UVIX, the 2x Long VIX Futures ETF, is where some element of nervousness show up though with implied volatility being maybe a little bit higher than where it should. We have been here before with this exact name. Back in early April we flagged UVIX as a short vol trade, took a fair amount of grief for selling volatility into a Middle East crisis, and two weeks later it had worked as implied vol collapsed and the position paid. When we ran our short vol list this week, UVIX lit up again at the 30-day horizon.

Let's have a look.

The Context

Look past the split and at what UVIX has actually done since. The fund drifted from around $62 at the start of July to a $54.75 low by mid-month as the tape stayed calm, jumped to $63.55 on July 17 as the Strait of Hormuz came back to the front page, and has since settled to $58.32. A round trip, driven by headlines, ending roughly where it started.

None of this is a UVIX story on its own. It is a VIX story, and the VIX is so far refusing to move. Hormuz is a live shooting conflict over the single most important oil chokepoint on earth, the heaviest week of tech earnings is days away, and the market has priced both into a 16 handle yesterday at the close and “only” 18 pre market this morning.

Realized volatility tells you how much this thing has actually been moving, and the answer is a fair bit, though less than the headlines might have you think. Thirty-day realized vol in UVIX sits around 84%. To put that in context, the two-year median for this product is roughly 78% and the 80th percentile is about 93%, so we are a touch above the middle of the range and well short of its extremes. The nine-day is a little warmer near 88%.

The question is whether it stays there, but so far our forecast leans towards roughly the same and maybe a little lower. The model points 30-day realized vol down toward the low 70s over the coming month, back below its median, with the shorter windows easing the same way.

In April the whole trade rode a de-escalating crisis, and the compression did the heavy lifting for us. This time the geopolitics are pulling the other way, so we read the forecast as a gentle drift lower rather than a collapse, and keep in mind that no forecast survive a pretty scary headline.

With that in mind, let's see how the options market has reacted and where the best opportunities lie.

Read more

China has always carried a bit of a risk premium of its own: as the sole contender to the US for hyperpower dominance, it has spent the last decade locked in tit-for-tat exchanges with Washington, whether on trade or, more recently, on geopolitical matters. And for that, investors often tend to overpay for insurance on Chinese equities.

This may sound a bit off-key with our usual data-first approach. Yet just seeing the market as a gigantic equation is rarely the right approach, and looking at it from a purely macro standpoint is the kind of skill we do not possess. For us, the beauty happens when you can combine human judgement with a strong data foundation. And this trade was no exception: for the reasons explained above, we were naturally inclined to follow our ML models when they came up with strong odds — especially because we also know there is some structural demand for hedging in Chinese equities.

Did it work in the end? Spoiler alert: It didn’t. And this Trade Anatomy is a testament to something we deeply believe: human judgement is often weaker than cold, well-calibrated statistics.

Let’s have a look.

The trade

Three weeks ago, we were writing about an opportunity to short the 30/34 strangle as well as the 32 straddle in FXI. At the time, we were pointing to solid odds (68%) of implied volatility exceeding the subsequent realized volatility over the next 30 days.

We were also expecting to see IV go down and potentially RV go up. We pointed out at the time that the negative VRP could correct itself, and this kind of scenario is sometimes the result of the market already anticipating things to calm down substantially, and refusing to overpay more than necessary for options. This time was no different.

Despite our prediction that RV would go up, it actually went down from almost 24 back to 21 shortly after the signal was issued. And while IV didn’t budge at first, it ended up going down too, to north of 21. However, something is striking about this chart: while RV likely dipped just below the IV we sold in the first days, the VRP stayed razor-thin throughout the entire trade. In effect, taking this trade was ultimately a bet that RV would drop substantially — enough to turn a profit.

And by the look of it, that wasn’t the case. Add to that the fact that the drop in realized volatility came alongside a steady trend up in the underlying, and you have the perfect recipe for a trade turning sour.

When we entered the trade, FXI was trading right under 32; it ended the period above 34, right on the upper leg of our strangle. The trend up was steady and detrimental to a strategy that deliberately chose not to delta hedge. We are happy not to delta hedge, and to bake into a volatility trade some element of terminal distribution (where the underlying sits at the end of the position) — and this time, it played against us.

As a consequence, there was never really a moment where this position turned meaningfully profitable: sure, after a few days in the trade and as implied volatility came in, you could have banked a 10% profit — but that was short of our 25% target, and with plenty of time left in the position, there was no urgency to close it.

You cannot win on every position, unfortunately, and having an edge means reducing how often things won’t go our way: this was our first negative trade in more than two months. Now could we have avoided the trade? That is a fair question: the expectation that RV would drop significantly enough to profit was ultimately a bit of a gamble, especially when our own model was saying this was not necessarily the favoured view. Another reminder that human judgement, while useful, is often weaker than cold, well-calibrated probabilities and forecasts… no matter your degree of experience.

Let’s now have a look at the Greek PnL attribution.

Share

The PnL Attribution through the Greeks

Here is the anatomy of a loss. On paper, our two core bets both worked: theta collected its rent (+$76) and our short vega made money (+$62) as implied volatility drifted lower over the two weeks. The problem was everything tied to direction. With a razor-thin — often outright negative — edge to begin with, there was no cushion, and once FXI started trending higher, delta, gamma and a vicious vanna print more than swallowed the profit. The book closed near −$140.

Read the stack from the top and the trade looks healthy: theta accrued steadily, and because implied volatility did fall, our short vega added +$62 on top. Together that is roughly +$140 — exactly the profit the thesis promised. Now read the bottom. Delta cost −$128, by far the largest line, as FXI ground higher and a position we deliberately left un-hedged drifted onto the wrong side of its strikes. Gamma piled on another −$61 of convexity damage, and vanna, of all things, took −$86. Two Greeks earned exactly what we expected; three Greeks tied to the trend took all of it, and then some.

Flip the path — a drift lower instead of higher — and that same theta and vega walk away with a clean win. That is the uncomfortable lesson of a thin-edge trade: you are no longer being paid for volatility, you are being paid to be right about direction.

That July 15th bar deserves a pause. Vanna measures how your vega shifts as the underlying moves — the cross between price and volatility, and a Greek that usually rounds to noise. Not here. We were short options whose vega was concentrated in the 32–34 zone, so as FXI punched through the 34 strike, our exposure to volatility lurched and the mark moved hard against us — even as implied volatility was falling in our favour. It is rare for vanna to headline a two-week trade. It did here because the underlying went exactly where it hurt most.

Now, the only question that matters: should we roll that position and try again? Should we assume the path will be better behaved this time?
Let’s have a look.

Read more

Last week, the US equities market went down, with the SP500 losing roughly 1.5% while the tech-heavy Nasdaq lost almost 3%. The losses were driven mostly by the reversal of fortune in semiconductors, which alone lost 6.5% — their worst week since Liberation Day, 15 months ago.

And as if that alone was not enough, oil went up sharply again as tensions in the Middle East resumed: the US announced the end of the ceasefire with Iran two weeks ago, after the Islamic Republic shot at ships in the Strait of Hormuz, and has been dropping bombs on the country with no sign of stopping for most of last week.

That resurgence of fear has rattled the market, despite a pretty good inflation print this week: inflation came in at only 3.5%, down from 4.2% in May and below the 3.8% economists expected.

As one duly noticed, this is a very different landscape from the lazy summer we had been anticipating — and praying for — during most of Q2. In that context, it is not too surprising to see the VIX settling at 18.53 this week: while that number is pretty normal in and of itself, it signals quite clearly that the market is not willing to keep the next 30 days of risk unhedged.

So what are the data showing?

First of all, our models just detected a push back into Stress territory. It could easily revert over the next few days if peace and quiet were to come back to the marketplace. But that is not the general theme these days, especially as the semiconductor unwind has been a worldwide story and not just a US one. It’s one thing to react to American news, but these days, whatever happens in Korea, China, and Japan definitely has an impact on the mood the next day in the US.

And while realized volatility is not skyrocketing, it went up quite a bit on Friday as the market weighed the implications of the release of the Kimi K3 model — another DeepSeek moment for most of US tech. The Chinese lab delivered a model roughly on par with the best US LLMs at a third of the cost, casting a shadow over the current valuations of some public companies, but also private ones: yes, Anthropic and OpenAI, we are looking at you, especially as your IPOs are supposed to come sooner rather than later.

At 14% over the last 30 days, realized volatility is give or take where we expect it to be over the next 30 days. Is this high or is this low?

It’s roughly around the 50th percentile of observations over the last two years, so one can totally argue that there isn’t much to be worried about… just yet. And a VIX at 18 — or even 19, as we briefly saw on Friday — could be good for business: that is far from the 8-point margin in VRP we got accustomed to in the year after Liberation Day, but it’s absolutely manageable. To give some perspective, we took the occasion on Friday to sell with our group the 7100/7800 strangle expiring August 31st — a position that seems well timed to profit if calm comes back over the next week or so, or at least if there is no further acceleration.

And continuation is really going to be the theme over the next few months: despite the first three weeks of July traditionally being pretty quiet, realized volatility in the tech sector is still boiling hot. As per the chart above, QQQ has come back down from the 30% threshold observed at the end of June, but to put things into perspective, that is still roughly on par with what we saw in August 2024, when the market got caught off guard by the unwinding of yen carry positions… in Asia, already.

Implied volatility in SMH has been climbing steadily over the last eight months, from the high 20s in December 2025 to a solid 60% at the close on Friday. We haven’t recorded any sign of extreme stress yet, but this kind of trend in SMH is not exactly reassuring: the market can collectively overreact over a few days, or do something completely senseless for long periods of time. But it rarely buys insurance for no good reason.

An astute reader would question that sentence and ask: well, how do you explain the 8 points of VRP in SPY during the twelve months following Liberation Day? Isn’t that a sign that the market can overpay for insurance for a long time?

Absolutely, and the answer is in the question: after a shock, it is well known that the market can exhibit a strong recency bias and stay hedged just in case. Think of fund managers respecting their risk limits more rigorously. But that tends to come after a shock.

Has there been a shock yet in the semis…? Not really.

Yet the price of the two-week 10-delta put stands at a staggering 75%, which only reflects the unusually high realized volatility in semiconductors, without again, any major shock to be recorded and by that we mean a -10% in one day for instance.

We know that IV Rank measures are particularly popular in retail communities: it is the number measuring how high implied volatility is against its own history. These days, by any metric, IV Rank will look astronomical, and it will be tempting to sell options. Be very aware that you are providing liquidity to a marketplace that collectively seems to agree that insurance in the semis is, at the very least, fairly priced — and could potentially keep getting more expensive.

If you’ve been a reader of Sharpe Two for a while, you know that a much better metric to assess whether or not you want to sell options is the VRP — the difference between implied and realized volatility. Plotting it against every expiry listed in SMH, the term structure doesn’t look great either.

In the short term, you would actually be better off” being long options. And in the 30 DTE+ area, the VRP is so slim that selling here is more a recipe for problems than a moneymaker. And what about SMH’s cousin suffering from ADHD — SOXL — particularly popular in retail communities addicted to leverage?

Realized volatility has been pretty stable now… at 100%, with IV sitting at 180%. On paper, it looks like a great trade: what can go wrong? You sell 180, wait a few weeks, and collect the difference between implied and realized that never lives up to expectations. But be mindful of what 100% realized volatility means: SOXL opening down more than 10% one morning has a probably north of 10%.

This is exactly why the market is willing to pay so much money for insurance. Selling junk insurance contracts is often a very profitable business… until it’s not. As per our model, while it was pretty confident you could harvest that premium a month ago, the probability has since dwindled — down to as little as 59% at some point this week, despite the gigantic premium.

Once again, there is no badge of honour in making money the reckless way, and as far as we are concerned, we are not touching SOXL these days.

Better safe than sorry.

In other news

The good thing about the current US administration is that it never ceases to amaze us in its ability to make deals in places where we market operators would never have thought a deal was even possible. It just keeps showing what little schoolboys we are on a trading floor, and how much bigger we need to dream when it comes to making money. This week, we learned with contemplative marvel that Trump Media was pitching the idea of direct access to a private feed of the President’s Truth Social posts — the account where he regularly taunts global leaders, political or economic, makes grandiose declarations, and usually triggers a market reaction for the ages. So for the casual sum of 100k USD per month (!!!!) you could be informed milliseconds before everyone else of what Trump may be saying.

You have to respect the audacity of the current President of the United States, who has reportedly already made a few billion from crypto markets, a few more from various business endeavors, and allegedly a few more still in the oil market. When Michael Burry opened his Substack, we estimated he instantly made 3 to 5 million a year. If Trump were to proceed with this, he could instantly make a billion dollars out of what would arguably be the biggest conflict of interest ever recorded. But hey, why not after all? If there is a willing buyer?

The question is: what does that signal? If you ask us… actually the opposite of what willing buyers may think: when you can’t make money from the wire anymore… just sell it by telling everybody what a great wire it was.

Just our two cents.

Share

Thank you for staying with us until the end. As usual, here are two great reads from last week:

• The Terminalist is back at it again: a very thorough analysis explaining why, when MCPs and AI have been all the rage, Bloomberg has been pretty silent about it and unwilling to participate. Another masterclass in business strategy, and totally worth the 30-minute read.

• And here is a fascinating interview of Eric Ho, the CEO of Goodfire, a company specializing in LLM explainability. Ever wondered how agents do it? And why it feels like magic? You are not alone, and while we do not have all the answers, some people are actively working on these problems.

That is it for us — we wish you a wonderful week ahead and, as usual, happy trading.

Ksander

Leave a comment

While the S&P grinds out fresh record highs and the VIX sits parked around 16, far far away from the geopolitical activities in Q1, one corner of the market did not get the memo. Chinese equities just had their worst week in more than a year, with the Hang Seng shedding better than 5% and Hong Kong closing at a one-year low. The trigger was familiar: an unwind of leveraged AI and semiconductor bets that started in Korea and Taiwan and rolled straight into the mainland tech names, with a dose of renewed Middle East tension on top.

FXI, the iShares China Large-Cap ETF, wore all of it. The fund topped out at $38.26 in mid-May and has ground lower almost without a pause since, closing near $31.95 to end June, a 16% drawdown that turned violent in the final week of the month. When a market falls like that, the options market does what it always does: it gets expensive.

And this is what caught our attention.

Let's have a look.

The Context

FXI spent the spring drifting lower and the last two weeks falling. From the $38.26 high on May 13, the fund slid to a $31.59 low on June 26 before steadying near $31.95. Most of the damage came in a single week, when the AI trade that had carried the mainland tech names all year went into reverse and dragged the index to its lowest level in a year.

The driver underneath is not really a China story, or not only one. The same names that led global markets higher, the chips and the AI infrastructure plays, are the ones that led this pullback lower, and China's large caps carry a heavy dose of exactly that. When Seoul and Taipei started unwinding their most crowded bets, Hong Kong followed within days, and FXI is the most direct ETF expression of that move on a US screen.

Realized volatility tells you how rough the ride has been. Thirty-day realized vol in FXI now sits at 23.9%. To put that in context, the two-year median is 19.5% and the 80th percentile is 23.4%. We are sitting just above the ceiling of the normal range. At nine days, realized vol is running even hotter at 26.4%, also above its own 80th percentile. This fund has been moving more than it usually does, and the recent sessions are where most of that movement lives.

The question is whether it stays that way, and our near-term forecast leans lower. The model points realized vol down over the coming week, from roughly 23.7% back toward 20.7% over the next week, though it eases only modestly further out. The single worst week is now behind the index, and unless the AI unwind finds a second gear, the sharpest stretch has likely passed.

So we have a fund that just fell hard, that is realizing volatility at the top of its range, and whose options are, by every account, expensive. That combination usually means one of two things: either the premium is worth selling, or it is a trap where the options are dear for a very good reason and the underlying is about to prove them right. Telling those two apart is the entire job.

With that in mind, let's see how the options market has reacted and where the best opportunities lie.

Read more

And that is okay. You do not always need a story to flag when volatility is overpriced. A reason is a double-edged sword: it can scare you out of a trade when the context looks worse than it is and leave you on the sidelines, and it can just as easily keep you in a position far longer than you should, as the narrative pulls you away from what the data actually says.

In the case of SOXL, and tech in general, the data has been stubborn. Realized and implied volatility have both been exceptionally high, hinting at a possible shift in regime for how the market behaves over the next few months, and that alone is enough to keep people on the sidelines. The same data, however, still showed a very healthy variance risk premium, which once again let us land this trade profitably. What it also showed, in plain terms, is that realized volatility, the measure of how much the underlying actually moves, was very high, so you should have gone in expecting wild swings in the underlying while you collected premium for a risk that never fully materialized.

That is the nuance most people miss when trading the variance risk premium. Harvesting the gap between perceived risk and realized risk does not mean there was no risk to begin with, and tunnel vision on the potential profit can have catastrophic consequences. The SOXL strangle went for about $30. The slightest adverse move could have put a small account under severe stress. Whether you should have taken the trade is not something we can answer for you, but now that it is closed, with no dramatic ending, it is a good time to put on the risk-management hat and take a hard look at your own balance between size and risk.

That, again, we cannot do for you.

The Trade

Two weeks ago, we were highlighting an opportunity to short the 190/410 strangle in SOXL, expiring July 17. At the time, we pointed to the pronounced gap between implied and realized volatility as the main driver behind our north star: a 75% probability that the implied volatility sold that day would exceed the realized volatility that followed.

There was only one small catch: realized volatility sat at a ginormous 103.4%, and we were still forecasting it to climb from there. Yes, the VRP was healthy at entry, close to 70 points, and yes, we forecast implied volatility to come down, which should have helped the PnL. But 100% realized volatility should make you think at least twice. The risk of a jump one way or another is extremely high, and so is the risk of a strong trend in either direction, leaving little room for any outcome other than a loss on a short strangle.

The market is not stupid, and that is exactly why it charged so much to insure that risk. It does not mean the risk goes away…

And that is why our first prediction went awry: implied volatility rose instead, as the market braced for a potential wild move.

It went from 166% up to 185%, against our call, and was almost certainly a drag on the PnL early in the trade. Realized volatility, meanwhile, did not materially rise; it hovered around 100%. That is what ultimately opened the door to harvesting the large gap between the risk that actually showed up and the risk that had been sold to the market when we struck the strangle.

That, however, still depends on how jumpy the underlying is. As regular readers know, we do not delta hedge our positions, and at 100% realized volatility a hedge does far less than you would hope anyway: you can wake up to a 10% overnight gap that no intraday hedge would ever have caught, so the premium and the sizing end up doing the real protecting.

In the end, the underlying behaved at first, then fell far enough to occasionally flirt with our put side. Without an edge, that moment could have looked scary very quickly. But once again, the premium baked into the position, especially on the put side, made it manageable from start to finish. We lost money for the first few days after entry, then time did its work, because the risk never caught up to the insurance premium we had sold early in the trade.

If you followed the position, you entered at the open on Thursday somewhere between $30 and $32, and could have closed it the following Monday near $20, comfortably past our 25% profit target. Was it luck? When an asset realizing 100% volatility does nothing dramatic, you can count yourself a little lucky: at that level, a violent move is a feature, not a bug. But we also charged 70 points over the realized risk, and the decision to put the trade on was not made recklessly.

Let’s have a look at the Greek PnL attribution.

Share

The Greek PnL Attribution

Looking at the attribution, this clean win was in fact a genuine tug-of-war. A short strangle on a 3x semiconductor fund is never a quiet position, and the Greeks make that plain. The equity curve dug a hole of roughly $550 in the first sessions before climbing back to about $960 by the exit. Here is how each Greek got us there.

Two Greeks paid the bill here. Implied volatility rose from 166% to 185% instead of falling, so our short vega bled close to a thousand dollars over the life of the trade. Gamma cost about the same, the standing price of being short a strangle on a fund that can swing 20% in a single session; the two are really one story, since gamma is just the cost of the very moves that delta later cashed in. Against them, theta did exactly what it is paid to do, collecting rent every day the strangle survived, roughly $1,500 by the end. And delta, worth about $1,400, was the piece that tipped the whole thing into the green.

The daily chart shows where that delta came from, and it is worth being honest about it, because it was the path we needed, not a structural edge. SOXL had rallied toward the low 300s in the first week, which left our strangle leaning short delta. So when the fund tumbled from around 300 toward 250 on June 23, the move ran in our favor instead of against us, and the bounce back over the final sessions did little further harm once theta had done its work. Had SOXL kept falling straight through our put, that same delta would have turned from friend to executioner. That is the whole game with a short strangle: we sold volatility that was too rich for what the fund went on to realize, and over enough repetitions that edge shows up, even if any single path can still breach a strike. Which is exactly why we kept the size small.

The only question that matters now is whether we should put it back on. The market is very likely still charging a hefty price to insure the variance risk in SOXL. Let’s have a look.

Read more

Indeed, the move was not tied to equities alone: gold lost another 3% as well, now far from the all-time high it made early in the year, and the reopening of the Strait of Hormuz has USO falling further still. The price of oil is already well below where it sat before the war, as demand is obviously weakening.

You did not need much to start hearing talk of a potential deflationary regime waiting around the corner, despite some inflation numbers coming in slightly higher than anticipated this week.

Needless to say, there are conflicting reports about the true health of the global economy and very few answers. What has been striking, though, is that the volatility episode that started in mid-June picked up again this week, after softening the week before, pushing 30-day realized volatility well above 15%, to 16.92%.

Is this high? Yes, it is. To offer some perspective, at the peak of the geopolitical tension in H1 2026, realized volatility flashed 19.5% on the S&P 500. The Iranian vol premium is getting slimmer and slimmer, and what is interesting is that no obvious catalyst has been driving the recent moves. What is even more impressive is that realized volatility in the Nasdaq is now at a level we last saw in August 2024, when a gap down in equities on a poor NFP combined with a sharp move in the yen to trigger a selloff in overleveraged positions.

Needless to say, this change of pace in realized volatility was not on our bingo card a few weeks back, and at this rate it could start to challenge the lazy-summer hypothesis over the next few weeks. At the moment, options at the very front of the curve are still not paying enough for the risk in the marketplace to be worth selling, unless we were to see a steep drop in the daily swings.

That hypothesis still has some legs, and waiting until the turn of the month, when the portfolio dressing is finished, matters before drawing any firm conclusion. However, the cool-off will have to happen sooner rather than later. Many portfolio managers follow volatility-based allocation, and with levels climbing dangerously high, they may have to trim their holdings, potentially accentuating some of the swings.

One thing is for certain: the variance swap curve is as flat as we have seen it through Q2, where a strong and healthy contango held for most of the period, and it stayed glued to the 20% level for most of last week. Nothing meaningfully above it, and any attempt by implied volatility to move lower was constantly met by more swings in the equity market dragging it back up. That is pretty much in line with the current level of volatility of volatility, which has also climbed back up over the last few weeks.

It sits in the reasonable part of its range, the kind of level where a 3-point intraday move in the VIX should not feel strange, but well short of the genuinely scary readings we saw at the peak of the Hormuz crisis.

But let’s go back to QQQ for a second: what is particularly striking is that while realized volatility is now past its August 2024 level, vol of vol is behaving pretty well. Obviously it could still spike higher over the next few weeks, but at this stage that is not what our models are saying.

The likelihood of it cooling off over the next few weeks is high, at roughly 70% for now. It is obviously just a probability, and with the right piece of news an adverse move could still occur. But we are not there yet.

So, looking at the tech sector, which is the object of so much wild speculation, we end up with a paradox: realized volatility is clearly at an unusual level, while volatility of volatility, though elevated, is pretty much normal. What to make of it? As far as we are concerned, vol of vol has the last word here: plenty of things can make stocks move, that is the life of the market. What is much more concerning is when the price of insurance itself is not stable, because that reflects real, deep indecision among market participants.

Still, we are not denying the stress, or the clear change of regime, where the floor in VXN is probably closer to 22 than the 17 we saw last year. But the market has not shown it is ready to pay much beyond 32. Until when? Well, if realized volatility keeps climbing, a spike into the high 30s or low 40s would be far from farfetched. Again, that is not what our models have in store for the next few weeks, but let’s look past the summer: if that June moment proved not to be an anomaly and high realized volatility is still a feature of the market in early September , the price of hedges may not stay as “low” as it is right now.

We do not think this is a reason to panic. But it is definitely worth respecting the data, and maybe buying some hedges that expire at the end of 2026. Better safe than sorry.

In other news

Here is another piece of news that may not soothe investors’ nerves: we learnt yesterday that OpenAI may be postponing its IPO to somewhere in 2027. Of course, there has been no official communication on the matter, but the speculation cites volatile market conditions as the main reason.

This is obviously only the tip of the iceberg. The finance game is often won by dealing well with incomplete information, and we did not need much to draw our own inference about the state of the frontier lab’s balance sheet. The market may be moving more than usual, but having to show a balance sheet as bleak as some of the wildest figures suggest may simply be the nail in the coffin.

It is easy to get consumed by the day-to-day of the market and end up thinking you hold the smartest view in the room. But we will dare a more cynical take: does the balance sheet really matter? Can the United States afford to lose ground to China in the AI race? We do not think so, and the current and previous administrations have repeatedly proven they will not let that lead go easily. So the only question that matters is: what are the conditions that could hand an irrevocable lead to the Chinese? We have a few ideas: a stock market down 40%, for instance, and a severe, prolonged bloodbath in their tech sector. We are not saying it cannot happen, but a very powerful government has every incentive to do whatever it takes to make sure it does not.

Share

Thank you for staying with us until the end. As usual, here are two great reads from last week:

• Bitcoin has hit a 20-month low and is failing to be anything other than a risky asset, fairly correlated to the rest of the market. Yet here is an interesting paper that uses order flow in currencies worldwide to anticipate moves in the digital currency. The study stops in 2022, unfortunately, right when a new regime started for crypto. It would be interesting to see whether the predictability is still there.

• We also learnt this week that one of the most influential figures in modern finance and markets has died at age 100. Alan Greenspan, often called the Maestro, oversaw some of the most important events of the last 30 years, whether directly at the helm of the Fed or later as a kind of policy oracle. Here is a long piece from the FT in his memory.

That is it for us this week. We wish you a great week ahead, a little less realized volatility, and, as usual, happy trading.

Ksander

Leave a comment

IBIT, BlackRock's spot Bitcoin ETF, has worn all of it. The fund topped out at $46.47 on May 11 and has ground lower almost without a pause since, closing yesterday at $33.87. That is a 27% drawdown in six weeks, and the flows tell the same story: spot Bitcoin ETFs have just posted six straight weeks of outflows, nearly $6 billion pulled out, with only a small green print on Monday hinting that the selling might be tiring.

We have been here before with IBIT, on both sides. In April we flagged it long, when the moves were big and the options were cheap. In early May we flipped to short, and we will be honest that selling premium into what turned out to be the opening act of this exact selloff was no free lunch. The signal is back on the short side now, and our short vol list lit up on IBIT at both the 9-day and the 30-day horizons.

Let's have a look.

The Context

IBIT's chart is a one-way street. From the $46.47 high on May 11, the fund slid to $33.87 by yesterday's close, a 27% drawdown that played out in steady daily increments rather than one violent crash. And because bitcoin trades around the clock while the ETF does not, the slide did not even stop overnight: IBIT closed at $33.87 and was indicated near $32 before this morning's open, a reminder that the gaps in this product happen while you sleep.

Every leg of that move has the same fingerprint. When the dollar firms and the market decides rates are staying higher for longer, the most speculative corner of the risk spectrum is the first to be sold, and bitcoin still sits at the far end of that spectrum no matter how much institutional plumbing now surrounds it.

Realized volatility has followed the price down the stairs and then jumped. Thirty-day realized vol in IBIT now sits at 47.5%, just under the 80th percentile of its two-year range (48.09%) and well above the 40.03% median. The selloff has done its damage to the vol numbers. The front is calmer: 9-day realized vol is 41.88%, sitting right on its own median, because the most recent sessions have been a grind rather than a gap.

Our forecast expects the elevated reading to fade. The 30-day realized-vol model points from roughly 47% back toward 41%, with the 7-day and 21-day forecasts leaning the same way. The model is not calling for calm in bitcoin, which would be a brave forecast in any month. It is calling for the crash-level daily moves to roll out of the window and realized vol to settle a few points lower.

Just to be clear; we are not calling for a calm period in IBIT. However, with July around the corner, it is likely that that forecast realized itself and that the swings in IBIT slow down.
With that in mind, let's see how the options market has reacted and where the best opportunities lie.

Read more

And this is reflected in the hedging needs: implied volatility shot up over a few days in early June before calming down post-FOMC, when we learnt little beyond confirmation of the likely rate hikes. Then, earlier this week, with no obvious reason, implied volatility went up again and challenged some of the most leveraged positions in the tech sector.

It is in situations like this that we are happy to sit on a delta-neutral position. Or at least in theory, because as the market moves, your risk profile changes, and that is exactly why we run this exercise: if you enter a 20-delta strangle and the next day the market rallies or tanks 3%, you now have little volatility exposure left and are mostly hoping the trade comes back inside your strikes. And while a certain level of variance is acceptable, if it happens too often, your thesis, your metrics, or whatever system you use to get in may be the culprit. Never the market.

In our case, here is a trade that once again paid off mostly as we expected, with a welcome bit of help from variance at the very end, which this time put our position back on track. That is the thing with variance: it is mostly random, and it mostly cancels itself out over time, so that your true expectancy can show up. It is still a little early to do a one-year round-up since we started this series, but in our group we are up 40% with a Sharpe comfortably above 2. The secret? Using machine-learning models to detect when the implied volatility you sell today is likely to exceed the realized volatility that follows.

Let’s have a look.

The trade

Two weeks ago, we were writing about shorting the 690/760 strangle in SPY, expiring July 17. At the time, our models put a 72% chance on the implied volatility we sold that day exceeding the realized volatility over the next 30 days, a great setup to harvest the gap between anticipated risk and actual realized risk, otherwise known as the variance risk premium.

The variance risk premium in SPY was about 9 points at the time, which left comfortable wiggle room even though we forecast that realized volatility was almost guaranteed to keep climbing from there. We were also fairly confident implied volatility would come down, which should have helped carry the position forward.

In the end, things happened pretty much as we anticipated.

Implied volatility fell from its 21.5% peak, all while movement in the underlying kept dragging realized volatility higher, north of 15%. And while things calmed down over the holiday-shortened week, yesterday’s session brought a few concerns back to the equity market, and with them, implied volatility in SPY ticked back up. Even so, we should still expect it to carry the position, especially early in the trade.

How about directional risk? As a rule, we do not delta hedge our positions; we accept some path and variance risk as the market unfolds.

In this case, the market actually started by rebounding pretty strongly as implied volatility came down, which hurt us a little, until the long weekend, combined with yesterday’s move lower, helped us close the position right on our profit target.

The PnL curve is very telling: the trade played out cleanly over the weekend, when the market stopped moving up, and you could have taken your 25% somewhere between Monday and Tuesday.

Let us now look in detail at how each Greek influenced the PnL, and confirm that our thesis played out again (implied too expensive versus realized) and that the lucky move down cannot, on its own, justify why this trade is in the green.

Share

The Greek PnL Attribution

On the surface this is another clean win, but underneath it was a genuine fight, and a good reminder of why we run these positions delta neutral. Start with the cumulative attribution. Vega did exactly what we wanted: implied volatility fell back from its highs, and a short volatility position cashes that in. The catch is spot-vol correlation. In equities, implied volatility tends to drop when the market rallies, and that is precisely what happened here. The same move that paid our vega pushed the underlying up toward our 760 call, and delta and gamma took the other side of it. They hurt, dragging the position to its low around June 15. This is the part most people never model: the Greek making you money and the Greek costing you are often two faces of the same move. Then the market did us a favor by doing very little. Over the long weekend, with the underlying parked, theta finally had room to compound, and it became the largest single contributor to the trade. The late leg lower on June 23 turned delta back in our favor, handed us a little more, and carried the position past our 25% profit target.

The daily chart tells the same story at higher resolution. Two outsized delta days do most of the damage and most of the repair: the drop on June 15 and the snap back on June 17 very nearly cancel each other out, which is exactly what you expect from a position you refuse to delta hedge. Underneath them, theta does its work relentlessly, every single session, because realized volatility never managed to catch implied. That distinction matters more than the happy ending. The move down on the final day was a welcome bonus, not the reason the trade worked: we sold implied volatility that was simply too expensive for what the S&P went on to realize, and over enough sessions that edge shows up regardless of the path the market takes to get there.

Now that we have a good view of why this trade made money, the only question that matters is whether we should put it back on.

Let’s have a look.

Read more

While we are still a few points away from an all-time high, the new Chairman, Warsh, inherited a market accustomed to setting fresh records rather than stagnating, and it will be interesting to see what kind of market he leaves behind four years from now. Especially because, as he chaired his first FOMC last week, he already marked a stark contrast with his predecessor, Jay Powell. Warsh has never hidden that he doesn’t believe in forward guidance, and that, to him, the best economist is the market, so (central) bankers should be taking their information from it rather than feeding information to it.

And that was exactly what he did: a pretty slim statement to announce that rates would be on hold, and a pretty short press conference to comment on the decision. Time will tell whether he adapts his style to the market over time, or the other way around. What is for certain is that, after a rocky session on the back of a “hawkish” (really?) press conference, the market recovered to finish on the highs of the week. Was it a free pass? Do they truly not care? It’s the beginning of what promises to be yet another lazy summer, and we wouldn’t draw conclusions too quickly: the first big test for Warsh will be in August, at Jackson Hole.

In the meantime… VIX 16 … tells you everything you need to know! Yet a few things will have to be verified over the next week or so before we officially start the traditional cruise toward July 4th. First of all, we left realized volatility at a weekly high last week, and while things have quieted down, it was still pretty shaky.

The market initially reacted pretty poorly to the remarks from Warsh before reversing most of the losses the day after. This was also the end of the quarter and the half-year, and it is not unlikely to see “weird” moves and acute volatility in that period. So was it really Warsh? Was Warsh an excuse for portfolio readjustment? The truth is often pretty blurry, and we would be careful with any direct causal explanation at this time of year.

What is for certain is that we still forecast realized volatility to come down a bit from what we observed over the last two weeks:

From 20% to 15%, and it will be interesting to see on Monday whether some volatility is left, or whether we open the door to coming back down to 12 much faster than this chart currently anticipates. With the Iran situation cooling off for good and no major new numbers on the horizon, the catalysts may be a little hard to find to keep the swings we’ve seen going. Yet until proven otherwise, the first few weeks are not the best for options sellers:

We were already writing last week to be careful with short-dated options, and the swings over the last week proved us right. If you were patient, you found some situations, especially in the wake of the Warsh conference, where selling vol was a better bet from a VRP perspective than it was at the close on Thursday: less than 2 points at around 30 days starts to be pretty slim. Obviously, if subsequent realized volatility turned out to be 8, that would be a hell of a trade. At the moment, our signals show that the implied volatility you sold today has about a 60% chance of exceeding the subsequent realized volatility. While that is not completely flip-of-a-coin territory, it should push you to consider something else. If you really want to trade equities, a Nasdaq 20-delta strangle presents much better odds at the moment.

However, for all of them except the Nasdaq, we currently anticipate the skew to flatten a bit, which means puts repricing favorably relative to calls. This aligns with a regime where we anticipate realized volatility cooling down overall. Yet another sign that if you are extremely bearish, you may want to reconsider your thesis. We are not at a point of real worry in the marketplace: yes, we know the financial media will insist on the shaky situation with Iran, or potentially inflation, but that is not where the market’s focus is at the moment.

When looking at the forward volatility term structure, the next bump in the curve is scheduled for the next FOMC week at the end of July. Therefore, let us hope for a truly boring market in early July so we can enjoy a few days off, and let’s see what the second half of the year has in store for us.

In other news

How hawkish was Kevin Warsh exactly? Some people will talk about this much more eloquently than us. However, what is interesting is the slight contradiction currently playing out. The dollar is arguably getting stronger over the last few weeks on the back of constant talk of a Fed hike, while bonds are… getting slightly more expensive, with rates giving back some of the gains they made a month ago when Powell was leaving the Fed.

In normal circumstances, we would say that one of them is lying and the situation may correct itself. But can we really say that we are in normal circumstances? Let’s take a concrete example: oil is finally back at its pre-crisis level, well below $80 a barrel, down from $120 a barrel no more than a couple of months ago. Yet there has been no sell-off in emerging market currencies, in particular the ones heavily dependent on oil like India and the Philippines, and no major sell-off in riskier assets overall: none of the stock markets worldwide went through hell on the back of a truly disruptive situation. In fact, on some interesting occasions, they went up like crazy (Korea). So why is that? We definitely do not have the answer, and it is interesting to watch economists scratch their heads to explain what is a blatant contradiction of every economic textbook.

Another proof, if needed, of the weak relationship between the economy and the market’s behavior. The lesson? Always the same: be careful with what is for certain. The odds that it doesn’t happen are always underestimated.

Share

Thank you for staying with us until the end. As usual, here are two great reads from last week:

• A year and a half after the release of the MCP protocol, you are starting to see a much deeper integration between merchants and consumers through … agentic pipelines. And while the idea of having agents purchase things for us looks like a slam dunk, this article actually questions that. If you are building an agentic business at the moment, this is worth a read.

• We haven’t talked about prediction markets in a little while, but that doesn’t mean they are going away. In fact, they may only be hardening themselves to become stronger. A big problem for customers over the last few months has been ghost fills, a phenomenon that could be characterized as the reverse of the early-2000s experience at the beginning of HFT. You thought you had a trade on… only to discover that it was never registered on chain. Here is an interesting paper explaining and quantifying the case in detail.

That is it for us this week. We wish you a great week ahead and, as usual, happy trading.

Ksander

Leave a comment

Nothing in that mix is good for long-duration growth, and nothing in long-duration growth is more leveraged to it than the semiconductors. After a year where the AI trade carried the entire market, a Fed that is suddenly worried about inflation rather than cuts is exactly the kind of headline that turns a one-percent index move into something far larger in the chips.

Which brings us to SOXL, the Direxion Daily Semiconductor Bull 3X. It is the most aggressive way to be long the chip sector that exists in an ETF wrapper, three times the daily move, and over the past two weeks it has behaved exactly like you would expect: a fund swinging fifteen to twenty percent in a session, both directions, sometimes in the same afternoon. The semis rally is still the story of 2026, but the ride has gotten violent, and the options market has noticed.

Let's have a look.

The Context

SOXL is built as a daily bet. It promises three times the daily return of the semiconductor index, which means it amplifies everything: the rally, the drawdown, and the chop in between. And 2026 has handed it all three. The fund rode the AI rally to a high near $280 at the start of June, then the last two weeks turned into a rollercoaster: a sharp drawdown toward $180, a snapback toward $270, and a close near $234 yesterday once the Fed was done. The chart is mostly spikes and reversals, with very little trend in between.

The driver underneath is the one moving the whole market. Semiconductors have been the engine of the 2026 rally, and a 3x fund sitting on top of them is the purest read on how much conviction is left in the AI trade. When the Fed signaled higher-for-longer on Wednesday, it leaned on the most rate-sensitive corner of the market, and SOXL, by construction, felt it three times over.

Realized volatility tells you how violent the ride has been. Thirty-day realized vol in SOXL sits at 103%. To put that in context, the two-year median is 78% and the 80th percentile is 88%. We are well above the ceiling of the normal range. At 9 days, realized vol is 119%, also above its own 80th percentile. This fund has been moving more than it almost ever does.

The question is whether it stays that way, and our forecast says no. The model points realized vol lower across every horizon, from roughly 104% toward 85% over the next 30 days, with the 7-day and 21-day forecasts leaning the same direction. The Fed meeting was the big known unknown of the month, and it is now behind us. Unless a fresh shock lands, the most violent stretch has likely passed.

With that in mind, let's see how the options market has reacted and where the best opportunities lie.

Read more

In the end, this trade in the homebuilder sector was also a good one, and while the PnL doesn’t show our traditional 25% take profit, it is green enough for you to take it off the table and not be greedy. We will see at the end of the article if it is worth putting it back on.

There is often a debate about our 25% take-profit rule. While we recognize it may at times be suboptimal (why would you get out of a trade when everything has been going smoothly?), it is first and foremost an operational guardrail for retail traders. We have it in place as a milestone to force a thesis reassessment — not as some amount of PnL we think is “embedded” in a short strangle position, which is the default view you will find on social media.

Indeed, volatility trading is not theta trading, as you can often read online. And if the market gives you 25%, there is enough here to take a step back and at least revisit the original market view: do you still want to be short volatility at these levels? Has the market moved, and have you picked up some unwanted directional risk? If the answer to both is yes, you are in a good position to “roll your trade”, or push it back and recenter, which is nothing more than… closing the trade and opening a new one.

Why 25%? We’ll admit it is an arbitrary number. However, it seems to be a good balance between how quickly the market will monetize your edge (had you had one in the first place, because once again, theta trading is no strategy) and the amount of commissions you would pay if you were constantly rebalancing (think about tight delta bands for instance) or, on the other hand, waiting for the home run and trying to squeeze every dollar out of the current position.

A position is nothing more than a tool. It helps you express a view and monetize your edge. But as the market moves and readjusts, so should you.

The trade

Two weeks ago, we were writing about an opportunity to be short vol in XHB through the 95/115 strangle expiring July 17th. At the time, the market was still fairly quiet, and nothing pointed to the intense second week of June that was coming.

At the time, we were predicting that both implied and realized volatility had a good chance of coming down, and that the odds of implied volatility exceeding the subsequent realized volatility over the next 30 days were 66%. This is the threshold we often consider in the app before putting a trade on. In the end, implied volatility went slightly up at first before shooting down this week as the market calmed down, while realized volatility closed essentially flat.

This is good for our position: if subsequent realized never caught up, time decay was very likely to bite into the options we sold, as they turned out too expensive relative to the embedded risk premium and the reality of forward risk. We insist on this: theta without IV > RV is an illusion, and we’ve had a few good examples this H2 for you to refer to.

The other thing that could be detrimental to our position is directional risk. We do not delta hedge, and we accept that — while our thesis is built around volatility-premium harvesting — a strong trend in the underlying can be catastrophic for our position. Once again, if you have a strong edge, more often than not that won’t be the case, and we are essentially swapping an easier operational constraint (delta hedging can be expensive and cumbersome) for more variance in the distribution of our PnL.

And in that case, the directional risk was real; we entered the position at around 104 on Wednesday, only to watch the underlying drift up for most of the time we held the trade. We should expect this to hit performance — though not to the extent where you would lose money, since the strikes were never challenged. Once again, the reason the strikes weren’t challenged is that implied volatility was overpriced. Had we had very little edge, the 20-delta might have become a 100/110, and we would have written a very different article right now.

In the end, while we didn’t reach the 25% profit target, we got really close, and we will happily take another win.

Let’s now have a look at the PnL attribution through the lens of the Greeks.

Share

The PnL Attribution

Looking at the attribution, this is about as textbook as VRP harvesting gets: theta did the earning, and gamma did the complaining. We sold options that were modestly too rich, realized volatility never caught up to what we collected, and time decay quietly did its job. The only real drama came from a couple of overnight gaps mid-week — exactly the risk you accept when you sell options and refuse to delta hedge.

Theta is the headline, accruing steadily to roughly +$110 — the single largest contributor and the entire reason the trade worked. We sold implied near 28.7 against realized that closed essentially flat, so every quiet day paid us. This is what we mean when we say theta without IV greater than RV is an illusion: here the premium was genuinely too rich, so the decay was real.

Gamma was the persistent headwind at about −$120. We were short convexity, and a couple of overnight gaps did most of the damage. Delta — the directional risk we refuse to hedge — actually netted a small +$40 despite the drift from $104 toward $110, simply because the strikes were never threatened. Charm added another +$25, leaving the book around +$50 net, a hair under our 25% target.

The daily chart is where the discomfort lived. Vega was a friend on June 4th, when implied collapsed right after we put the trade on, then turned against us mid-week as the market's bout of volatility pushed implied back up — a reminder that IV can rise against you even when your strikes are perfectly safe. June 9th was the worst session, almost entirely gamma: an overnight move gapped the book out of whack, with nothing to do about it by the open. But realized never sustained the spike, implied rolled back over, and the position recovered into the close.

With this in mind, the only question that matters is whether we should stick with the position. Things seem to be going smoothly — but is there merit in recentering and re-entering the position?
Let’s have a look.

Read more

The VIX spent most of the week above the 20 handle, with a few tense moments around 23, but it ultimately closed at the lowest level observed all week, at 17.76. What happened? Well, the usual. President Trump announced on Wednesday that he would bomb Iran to pretty much annihilate them. He had already made the rather stunning statement earlier in the week that the US had been taking 100 million barrels out of Iran without them noticing, and that the bombing, set to start on Thursday, would have as its main objective the capture of Kharg Island and its oil facilities.

24 hours later: reversal, no bombing, a deal announced (apparently the 37th time), and the stars were now aligned for the biggest IPO ever. 75 billion dollars raised from the public, a valuation of 2.1 trillion dollars, instantly making Musk the first (official) trillionaire, in theory at least. Not to hate on Elon but going back to the definition of a market and liquid/illiquid assets: one can only wonder what would happen to the valuation, and the price of these shares, if Elon was trying to swap them for dollars… the devil is in the detail of the order book and a mark, doesn’t equate liquidity… Some elements of answers in a few months when he, and the 4400 Space X millionaires, will be able to sell their shares.

So what’s next? If you ask us, still a little bit of the same for the week ahead: the first FOMC meeting under Chairman Warsh will be widely watched, and with inflation on the rise, or at the very least not cooling off, the market is now anticipating rate hikes rather than cuts.

We don’t know much about that, but what we do know is that the volatility from last Friday, which was fairly contained given the move down we observed, definitely carried over into this week, and that zone of turbulence may not be totally over. First of all, how volatile exactly was last week?

Pretty volatile. In fact, one of the most volatile stretches of the last two and a half years, and if you put aside the obvious shocks, the last seven days as a whole were pretty rocky. You never had a day where it was absolute carnage, the usual characteristic of real stress and tension in the marketplace, but the swings were turbulent enough to be felt by everyone.

When measured at 30 days, realized vol moved from 10 to almost 15 in about a week. That is a strong move, and rare enough to be noted. It happens once in a while and it usually takes a few weeks to diffuse. This is pretty much in line with our current forecasts for realized volatility.

We are currently right below 15, and if things were to keep up at the rate they’ve been at over the last week, it wouldn’t be farfetched to see 20 over 30 days, which would put pressure on options sellers and volatility harvesters. So, be aware. Yes, this is quite a stark change in tone from where we stood last week, but one has to respect what the data says, especially when the catalysts are there to offer up a few cold sweats.

That said, is this a reason to panic? Obviously not. Our regime model flagged some stress characteristics in the marketplace on Tuesday, but overall we are still in a period where harvesting VRP will pay off more often than not.

Sure, with VIX at 18, you may want to give it a few days to see if options buyers push prices back up, to give yourself more margin of safety. But is it likely that a true volatility event starts over the next few days, past the FOMC meeting? We do not think so, but we never know how the market will react especially if it really doesn’t like the “new guy”. In the meantime and data wise, this simply isn’t the term structure of a market that is overly stressed by what could be announced this Wednesday.

Sure, you may want to be careful, as usual, with the short-dated options, but otherwise the implied volatility on 20-delta puts, arguably the definition of insurance, is showing a pretty stable contango. Now, volatility of volatility has picked up over the last week, and by virtue of clustering, one has to assume that a 3 or 4 point intraday move in the VIX is much more likely now than it was up until just 10 days ago.

If anything, this is yet another warning for traders in super short DTEs: you must have had a couple of rough days, and with vol of vol rising, it can get worse before it gets better. So how do you extract VRP in these conditions?

There is no prize for doing it the hard way: if you don’t want to wait around for a moment when the marketplace is a little shaky, you can most likely target an expiration like July 31st, sell a 20-delta strangle, wait and enjoy the sparkles all over your social media timeline.

In other news

What exactly is happening at Anthropic? How did the company manage, in the course of three months, to go from competing for the title of one of the most loved companies in the world to one of the most hated? Nerfing Opus 4.6 and gaslighting the community about it was probably not the best idea; the very hefty price tag on 4.7 and 4.8 for the actual output was probably not great either. And then came Fable 5, supposedly so dangerous that it was outright configured to refuse answering questions on many scientific subjects, provoking the ire of the public and a plethora of excellent memes all over the internet. And as if this wasn’t enough, we learnt this weekend that Fable and Mythos would be purely and simply discontinued by order of the US government, citing national security concerns.

Let’s catch our breath for one second. Aren’t they supposed to IPO soon? We’ve known for a few weeks that they have filed confidentially, and while there is no official date yet, the success of SpaceX yesterday should definitely push Anthropic and OpenAI to present themselves to the public sooner rather than later.

Let’s hope for Anthropic that, by then, they will have engineered another brand reversal and be seen positively by the public again.

Share

Thank you for staying with us until the end. As usual, here are two great reads from last week:

• This week, some readers characterized our approach to vol trading as “recent high VRP + GARCH + macro vibes.” That made us smile. But we will never trade the simplification trading concepts over the complexity of the quant world. That said, here is a really good, strong yet approachable technical paper from the prolific Zohren that should spark ideas for anyone trying to predict realized volatility, particularly on short horizons. A quick note on this: it shouldn’t be applied as is, but it should give you a sense of how strong your competition is (Zohren worked at Virtu for a long time). And since it is in a research paper (from 2022, mind you), the technique is not novel. You should then adapt it to whichever use case you have.

• And because some may still be missing the point about how transformative agentic AI will be for technical roles, here is a dense yet fairly straightforward framework from Google drafting how data science and research may be done over the next few years. Models are only going to get stronger. If you are a data scientist now, getting comfortable with these pipelines will get you further than knowing every parameter in XGBoost.

That is it for us. We wish you a wonderful week ahead (and a first presser from Chairman Warsh), a little less realized volatility, and … happy trading!

Ksander

Leave a comment

Almost none of that 4.2% is the kind of broad domestic inflation a central bank can do much about. Core prices rose just 0.2% on the month. The engine is energy, and energy is the Iran war: crude is still elevated, the Strait of Hormuz is still a live headline, and gasoline did most of the work in the print. The trouble is the timing. A hot number landed six days before the Fed meets, at the worst possible moment for a new Chair trying to set the tone.

Kevin Warsh runs his first meeting as Fed Chair on June 16 and 17. The market expects him to hold rates, but it also expects him to retire the easing bias and lean hawkish, and after yesterday nobody is sure what the dot plot will say or how the first press conference will go. So the index walks into a rate decision with fresh inflation data working against it and a war premium that refuses to leave.

That is the backdrop for what our model is flagging. The selloff has been real, realized volatility has jumped, and the options market has repriced for a rough month ahead. But when we weigh how much trouble it has priced against what we think is actually coming, our short vol signal lit up on the S&P 500 this week, at both the 9-day and the 30-day horizons.

Let's have a look.

The Context

SPY spent the spring climbing out of a hole. The fund bottomed at $631.97 on March 30, in the worst of the oil-driven inflation scare, then rallied almost without interruption into June and printed an all-time high of $759.57 on June 2, a 20% move off the low. Then it turned. Over the six sessions since, the index has given back 4.5%, closing yesterday at $725.43. A sharp run to a record, followed by a sharp slap back down.

Every leg of that move carries a macro fingerprint. The rally was the market deciding the Iran war would not tip the economy into recession and that the Fed would eventually cut. The reversal is the market un-deciding both of those at once, as energy kept inflation sticky and yesterday's CPI print put a number on it. Strip the ticker off it and this is a rates-and-oil story wearing an equity index.

Realized volatility has followed a similar trend. Thirty-day realized vol in SPY sits at 12.85%, above the two-year median of 11.51% but still short of the 80th percentile at 15.34%. Elevated, not extreme. The front is louder: 9-day realized vol is 17.12%, above its own 80th percentile of 15.12%, because the past two weeks have carried the kind of one and two percent daily swings the spring rally never did.

And the forecast does not call for it to fade. Our 30-day realized-vol model points higher, from just under 13% toward roughly 18%, with the 7-day and 21-day forecasts pointing the same direction. The model is not telling us the S&P calms down into the Fed. It is telling us realized volatility has room to climb from here.

The forecast points up and to be clear are not betting the index goes quiet, especially after the last few days. However, we think that even that forecast may be a little overblown: 4th of July is just three weeks ahead, and once you remove SpaceX and FOMC… there really isn’t much left on the calendar.

With that in mind, let's see how the options market has reacted and where the best opportunities lie.

Read more

The madness around semiconductors did indeed reach South Korea, and the performance of companies like Samsung or SK Hynix explained in large part the day-to-day of the index. Lately, the volatility in the space is higher than usual, and that also shows in Korea. How did that play out for us? It was rocky, to say the least; it is now the second time we put this trade on and the high level of realized volatility means you need to be comfortable holding this position through pretty wild swings — amplified these days by all the different views, theories, and speculation around AI.

Although this trade turned out to be another winner, it should trigger another reflection about your own risk limits; as an option seller, you act as an insurance provider. And it is totally fine to focus on a segment of the market where the customers are inherently less risky. With realized volatility these days above 55% (not exactly your pristine AAA, but more like a borderline junk profile), the risk of a jump or an erratic move is high in a product like EWY. If you prefer a smoother risk profile and less risky situation (lower realized volatility to begin with and anything below 40%), there is nothing wrong with that at all.

Trading doesn’t have to be complicated, and it doesn’t have to be scary either. It will be stressful once in a while, that is ultimately the reason why you are being paid, but you are in control to which extent you are willing to lend your balance sheet. Do not let the market decide for you.

The trade

In our Signal Du Jour from two weeks ago, we brought to your attention that the 173/240 strangle in EWY expiring June 26 may be overpriced when one considers the risk embedded in the options market relative to what the index was actually showing.

At the time, our models were showing some pretty strong odds that implied volatility would exceed the subsequent realized volatility measured over the next 30 days. We also thought that implied volatility had good odds of coming down, which would have carried our PnL for a bit. However, we were mindful that realized volatility was likely to go up, with a probability of 85%. That said, this is an indication of direction, not magnitude: an 85% chance of going up means just that — not that the move will be explosive on the upside.

In the end, things happened pretty much like we expected them to: IV initially declined from roughly 77% to 72%, while realized volatility went up from 55% to 58%. Looking at it like this, nothing our position could not have handled, right…?

That is obviously putting aside the spike in implied volatility at the back end of last week, when a rout in tech stocks — and in particular semis — hit markets worldwide. Korea was not spared, and the resulting selloff was always going to hurt our non-delta-hedged position.

The underlying was roughly at $200 when we entered the position; it then ran up close to $220 before cratering down to $175. Needless to say, with gaps like this, your volatility-premium-harvesting thesis will suffer a little bit, as the market rushes for hedges and inflates the price of your straddle.

Yet, despite these pretty impressive moves, our position showed some decent profit midway through last week, before the relentless move down that prefaced what would happen in the US stock market last Friday.

Here is another reason why we often say to take profit sooner rather than later: in many volatility trades, you can let the thesis play out, hold, and make a bet on the terminal distribution. But in markets where realized volatility is already high and the gap risk is non-negligible, that is asking for trouble.

Let’s have a look at the Greek PnL attribution.

Share

The Greek PnL Attribution

Looking at the attribution, this was a tug-of-war between Theta and Gamma, with Vega leaning against us almost the entire way and Delta — of all the Greeks — casting the deciding vote. We collected a little over $580 of rent across the two weeks and handed almost all of it straight back to the convexity we were short. That is the trade in one line: we sold insurance on one of the most volatile products on the board, then spent the holding period paying claims.

Start with the rent, because it is the only Greek that behaved. Theta accrued steadily for the full two weeks to roughly +$580, the single largest positive contributor. It does not care about the headlines on Samsung or the tape on Friday — it simply collects, one decaying day at a time. In a normal week, that $580 is the entire PnL. This was not a normal week.

Gamma is where the money went. We were short a strangle, which means we were short convexity, and EWY spent two weeks doing precisely what convexity punishes: large, gapping moves in both directions. Cumulative gamma finished near −$680, more than enough to swallow the theta we collected. This is the structural cost of selling options on a name realizing north of 55% — you are paid to be short gamma, and then you actually have to be short gamma.

Vega was the persistent headwind. Implied volatility never gave us the clean compression we wanted: it dipped early, from roughly 77% to 72%, then snapped higher as the semiconductor rout peaked. Cumulative vega settled around −$220. The cruel part is the timing — vega hurt most precisely when gamma was already hurting, because a selloff inflates implied and realized vol at the same time. There is no diversification between your Greeks on a day like Friday.

And then there is Delta, the Greek we explicitly chose not to hedge — usually the line that ends a post-mortem in tears. Here it was the opposite. Because we let the position run unhedged through the entire round trip — $200 up to $220, back down to $175, then a partial recovery — delta finished around +$330 to the good. That is luck wearing the costume of a Greek: the path simply handed the strangle back its center of gravity. Net it out — theta and delta pulling one way, gamma and vega the other, with the second-order terms tipping the balance — and the book closed a little over $160 in the green.

This is why the daily chart matters more than the cumulative one. Midweek, with the underlying back in the middle of our range and delta quietly canceling the gamma bleed, we were sitting on our best mark of the trade — comfortably above $300. That was the moment to take the 25% and walk. We did not, and on Friday the 5th the bill came due: a single session erased close to $500 as gamma and vega detonated together. It took Monday's bounce to drag the position back into modest green. The lesson is the one we keep repeating — in a product already realizing 55%+, with gap risk this real, holding out for the last dollar stops being a volatility trade and becomes a directional bet you never meant to make.

After these wild swings, the only question left to answer is whether or not we should roll the position.

Read more

And it wasn’t just the stocks: silver lost more than 8% while gold was a little better at -3.5%. The oil complex was also down. In the end, except for rates, which stayed pretty much the same, this had all the characteristics of a bloodbath: liquidity being pulled out of the system while everyone rushed into hedges and safety. As a result, the VIX closed at 21.59 (despite the glitch in TradingView showing it right above 20).

The reason? Pretty straightforward, or very convoluted, depending on how lenient you want to be with the market these days: with 172k jobs instead of the expected 80k, the economy is running hot. Too hot. And for sure, the Fed will have to hike, especially with how the price of oil has behaved over the last four months.

Right. We’ve heard that one before, but after years of looking at the market, we’ve never seen a good justification for shorting an economy that is doing just fine. And we can’t remember a bear market starting because the economy was going creating job. Sure, we’ve seen these blips, but those are, well... rarely anything more than temporary. And if we were allowing ourselves a little macro observation, it is actually interesting to note every single asset class being obliterated on Friday except… the one that is supposed to react to inflation and potential rate cuts, ie TLT and the long dated run. Sure they strenghented but nothing in comparison to the movement in equities.

Yet the move is respectable, and it is important to do our due diligence and look carefully. You’d be a fool to ignore that spike: the VIX going from 15 to 21 is enough to pay attention to, and to at the very least assess whether you want to be the insurer of last resort.

Spoiler alert: absolutely.

Let’s start with the chart above: despite the respectable spike (once again, 6 VIX points are not to be taken lightly), our regime model is detecting a positive VRP regime — much more a regime favorable to option sellers trying to harvest the risk premium between implied and realized volatility than a regime of stress. Obviously, let’s see how the beginning of next week goes: if things stay agitated, we may start to see a few orange dots; however, that isn’t the case at the moment. What could be the catalyst? The never-ending saga between Iran and the US: we learnt last week that they were still playing with fire in the Strait of Hormuz, that Iran had unilaterally withdrawn from the negotiations over the unreasonable demands from Uncle Sam, while at the same time, that the discussions were showing great progress.

But apart from that, things are pretty light. Sure, we will have some actual inflation data this week; however, the move seems a little blown up for what is likely to happen: you currently have a staggering 8 to 12 points of premium along the term structure between what we just realized and what the options market implies for the next 30 days.

So let’s take a step back, while we still have some fresh memory of agitation and uncertainty in the market: remember the peak of the Iranian crisis in April, where Trump was raging live on social media, threatening “these bastards” with a return to the “stone age”? We were flashing realized volatility at 19 back then. Do you really think that over the next 30 days — especially with the 4th of July falling in 27 of them, and the SpaceX IPO — we are likely to see more price movement than at the peak of a Middle East war?

Everything is possible, obviously, but at the moment our forecasts are lazily pointing towards a slight push higher to 13 and not really more than that. And why would it be otherwise? On Friday — which, again, was a pretty steep decline in stocks and in dollar-denominated assets in general — the daily realized volatility was … right below 20%. Is that a lot? Not really — all in all, a fairly average day under the sun from a realized volatility perspective.

Over the last three years, it would sit right in the middle of the daily range. In comparison, Friday was the equivalent of a calm day during the six weeks of intense uncertainty in March and April. Once again, that doesn’t mean continuation is not possible. It can absolutely keep tanking over the next few days, and one will have to watch how the market behaves Monday and Tuesday. But without any meaningful increase in the daily movements, the options prices will start to hurt pretty quickly, as the premium evaporates and the risk that fund managers have hedged against doesn’t materialize.

If you ask us, this is typically the type of scenario where everybody wants to be a volatility trader, but when the demand for options is there — loud and ferocious — many discover that trading mean reversion on some supposedly more professional and quantitative asset class is no simpler than vile speculation on crypto, CFDs and other prediction markets. Don’t go all-in, obviously, but short the straddles and come back right before the 4th of July to see if the world really blew up.

In Other News

We have never hidden our disdain for Saylor and the MicroStrategists, and we won’t hide either that it is with a certain delight, and a mouth full of popcorn, that we are watching the price of MSTR tank week after week as the promise of a perpetual rise in bitcoin prices — meant to support a stock price designed to funnel hard cash raised from the public into crypto assets (and also into the pockets of its chairman, but that is beside the point) — fails to be fulfilled. Saylor and MSTR had to sell some bitcoin this week to honor their dividend strategy, and obviously the market noticed. Long story short, the company is now sitting on a pretty hefty unrealized loss of a little more than 11 billion dollars. Of course, all it needs is a nice little run in bitcoin for all of that to be forgotten. The problem is that bitcoin has some steep competition these days when it comes to liquidity: investors need cash for all the recent financial innovations — prediction markets and their very appetizing perpetual contracts and, obviously, the big and beautiful IPOs, IPOs like you’ve never seen before, so big and beautiful that everything is made for retail investors to participate. Finance is no charity, and when the entire industry is doing everything it can for arguably the least informed participants to dabble in it, one has to wonder what its true motivation is. You can ask Saylor — a man who, as a reminder, was charged with accounting fraud by the SEC 25 years ago, right before the dot-com crash, and settled without admitting wrongdoing — we are sure he would be pretty pleased to find some more degenerate liquidity so he could keep exiting his scheme.

Share

Thank you for staying with us until the end. As usual, here are two great reads from last week:

• We were always dubious about crypto, but we strongly believe that prediction markets can become a much, much bigger instrument than what options and futures markets on stocks have ever been. Case in point: this article from CNBC showing how a bar hedged itself during a Knicks NBA playoff night.

• We have spent our days playing very few video games, but Age of Empires was one of them. And when we saw a research paper trying to demonstrate that if LLMs had “human-like” capacities, you could argue that Age of Empires has the same (?!!), we were immediately interested. This is a great paper about how poorly defined hypotheses and methodology can lead humans to the conclusion… they want to hear.

This is it for us. We wish you a wonderful CPI week ahead and, as usual, happy trading.

Ksander

Homebuilders never got the memo. While equities relaxed, the long end of the curve did not, and that is the one place where housing has no defense. The Iran war has kept energy prices elevated, inflation has stayed sticky, and the thirty-year yield has spent the spring at levels it had not seen since before the financial crisis. Mortgage rates followed it straight back toward 7%. For a sector whose entire business model is a customer's monthly payment, the rate is everything.

XHB, the SPDR S&P Homebuilders ETF, has worn it. The fund topped out at $121.36 in February when the market still believed in rate cuts, then ground lower for three months as the rate-cut story died, bottoming at $94.86 on May 19. Since then it has bounced almost 10% as yields stabilized and builder confidence ticked up off the floor. It closed yesterday at $103.89, roughly flat on the year and still well below where it started the spring. A round trip that went nowhere, with a lot of motion in between.

That motion is the point. Our short vol signal lit up on XHB this week at both the 9-day and the 30-day horizons, in a name where realized volatility is sitting near the top of its two-year range and the options market is still pricing for the next leg down.

Let's have a look.

The Context

XHB is not a clean bet on six homebuilders. It is the whole housing-construction complex in one wrapper, equal-weighted across builders like Lennar, D.R. Horton, PulteGroup and Toll Brothers, building-products names like Builders FirstSource and Masco, and the home-retail giants Home Depot and Lowe's. No single stock dominates, so the fund moves on the thing they all share: the cost of money. When the thirty-year yield rises, the entire basket rerates at once.

The chart shows exactly that. From the February high at $121.36, XHB lost ground in near-straight lines into late March, recovered toward $109 in April as the bond market caught its breath, then rolled over again to the May 19 low at $94.86. The bounce since has been real but unfinished. We are 10% off the low and still 14% below the February high.

What drove it is no mystery. Every move in XHB this year has been a rates move with a homebuilder's name on it. The April rate shock, when the thirty-year auction went badly and the long end broke higher, is the same week realized volatility in XHB spiked above 32%. The May rally is the same week yields stopped climbing. The fund trades like a long-duration bond that happens to build houses.

So let’s look at realized volatility. Thirty-day realized vol sits at 26.97%, right at the 80th percentile of its two-year range. To put that in context, the two-year median is 23.59% and the 20th percentile is 21.94%. The fund has been living in the top quintile of its own volatility distribution for weeks.

But the trend underneath is down. Realized vol peaked near 32% in mid-April during the worst of the rate shock, cooled to around 25% by mid-May, and has only ticked back up to 27% because the May selloff and bounce kept the tape busy. The violent part is behind us. What is left is an elevated reading that is slowly stepping down.

Our forecast says more of the same. The 30-day realized-vol forecast points from roughly 27% toward 26.5%, the 21-day from 27.3% toward 25.8%, and the 7-day from 23.84% toward 23.48%. This is mild compression, not a collapse. The model is not telling us XHB goes quiet. It is telling us the worst of the realized-vol regime has likely passed.

So the picture is coherent. A rate-sensitive basket that took a real beating, has begun to heal, and is now carrying realized volatility that is high but fading. The question is whether the options market has noticed.

With that in mind, let's see how the options market has reacted and where the best opportunities lie.

Read more

Since then, things calmed down: equities are back to all-time highs, the reality-TV show between Israel, Iran, and the US is still running, and while bonds have relaxed a bit, they remain attentive to developments around inflation and growth, of course.

So how did that trade go? Not a slam dunk, but not a loser either. One of those where you can be right on volatility and still be flat because you accept some directional risk. This is another tale of the variance game: last week in KWEB, some strong overnight movement punished a position that looked set to play out perfectly. This week is a little different, and it reopens the old debate of whether you should delta hedge or not. We will answer it in very practical terms: does the price of the strangle justify the delta hedge? How about the size? Of course, if you trade millions on our signals (and we thank you for the trust), these questions matter. But we often encourage people to keep it small precisely for that: paying more in commissions and taking on the operational risk is not worth the effort over a structure costing 75 cents per contract. It really is not.

Secondly, the goal of trading is not to be right but to extract enough money through that variance game that your expectancy stays positive over time. And at this game, a tiny win… is still a win.

Let’s have a look.

The trade

Two weeks ago we were highlighting the opportunity to short volatility in TLT through the 81.5/86 strangle in the June 18 expiry. At the time, we were justifying the position by looking at the probability — roughly 68% — that the implied volatility baked into the options market would overstate the subsequent realized volatility.

We were also pointing to implied volatility having some good odds to go down over the period of the trade while realized could go up. However, the 2.9 points of variance risk premium were in theory enough to get us into the trade. In the end, realized volatility did go up, from below 9 to just above 9, while implied volatility came down quite a bit, settling below 10 over the last few days and offering, once again… a negative VRP.

With such a crush in implied volatility, you would expect vega to have helped our position quite a bit. However, as mentioned in the introduction, the bigger problem came from direction: when we entered the trade, the underlying was trading around 83.5; it closed yesterday above 85.5, after challenging the 86 strike for a few days.

That was indeed quite the move up and without delta hedging once again, the kind of trend that is hard to recover from. Could you have done something about it that goes beyond delta hedging and taking into consideration the cards that we were dealt with at the time? You could. In the Signal Du Jour, we were pointing at how the probability to see the skew flatten over the next few days was high (93% at the time of the article) and indeed the puts looked expensive. It is always easier in hindsight, but tilting the position to make it slightly delta-positive, or simply overweighting the put side, would have turned a pretty much flat trade into a winning one.

One last comment: although this trade was not as profitable as what we were hoping at first, with enough edge baked in, the premium is often your first line of defense against an adverse move. Hence the importance of being able to predict whether implied volatility will exceed the subsequent realized volatility.

Let’s now look at the analysis Greek by Greek.

Share

The Greek PnL Attribution

Looking at the attribution, this trade tells the story of a volatility call that was largely right and a directional bill that ate most of the reward. Vega and theta both did their job; delta and gamma quietly handed the proceeds back. What is left at the end is a few dollars of profit — a win on the scoreboard, even if it never quite felt like one.

Theta was the steady earner, accruing to roughly +$33 over the two weeks — the rent you collect for being short options while nothing dramatic happens. Vega added another +$10 as implied volatility crushed from the low teens back below 10, with charm and vanna chipping in a few dollars more. Against that +$48 of gains, though, delta gave back nearly −$29 and gamma another −$12.

You can read the toll straight off the equity curve. The position climbed to about +$16 by May 27th, then surrendered two-thirds of it in a single session once the underlying started pressing the 86 strike, before grinding back to settle a hair under +$7. Enough premium to stay green, not enough to keep what direction kept taking away.

The daily view shows exactly where it slipped. May 28th is the −$10 session that did the damage, almost entirely delta and gamma as TLT pushed up toward the 86 call. June 1st was the one day direction paid us, delta alone adding about +$24 — possibly the start-of-month reallocation into bonds — though vega clawed most of it back as implied volatility popped on the session. Theta, for its part, keeps printing its small positive bar almost every day: the one Greek you can count on when the rest of the book is fighting the tape.

The only question left to answer now is should we roll or not? Although, given what the attribution just showed us, the answer should be fairly obvious.

Let’s have a look.

Read more

Why is that? Because for the first time since the back-and-forth, almost public negotiation between Iran and the US, Iran confirmed on Wednesday that a deal was in the making, a material step in the right direction after months of open trolling of the US administration, claiming at times that President Trump was talking to himself.

The deal of course includes a reopening of the Strait of Hormuz, and, more surprisingly, a $300 billion fund invested worldwide for the reconstruction of Iran. We won’t comment on who won and who lost in that deal; that is far beyond our pay grade. We will just point out that, despite the market reaction, many are still convinced we are not out of the woods.

And why should we blame them? It has been going on like this for months, so why not an extra few months? First of all, because this is the first time since the beginning of the war that oil closed meaningfully down two weeks in a row: -5% last week, and now -8%. Once again, the market is a formidable machine at aggregating information. It is slightly imperfect, yes, but not meaningfully enough to say that, on average, an event priced at 10% was actually a 25%. And if risk perception is going down — whoever the next social media Cassandra may be — the market firmly disagrees right now.

And there have been many little signs like this, have you noticed? Let’s start with the obvious one - VIX on the 15 handle. This is genuinely the lowest level observed since early January, where a surprise attack on Venezuela was in fact just the early sign of what the rest of Q1 would look like. Is VIX too low? Once again, that may be your opinion, but as it stands, it is still materially above realized volatility.

Five points of VRP when the jump risk in equities gets meaningfully lower (realized volatility flirting with 10) is a genuinely good and much more stable condition for harvesting than the quite stressful weeks we had at the end of March — which feels like a lifetime ago now.
Yet we’ve been yapping about VRP for weeks now, so an astute reader could just say: okay Ksander, is that all you got? No, we have more. Once again, we don’t know if you’ve noticed, but VVIX is materially below its long-term average observed over the last 12 months. In fact, VVIX is so low that it is digging in the lowest percentiles observed since Liberation Day.

VVIX, or the implied volatility contained in VIX options, tells you the market’s expectations for VIX movement over the next 30 days. Last week it closed at 90, which was already the lowest level observed over the last 12 months and also happens to be its long-term average. Now it is at 85, one of the lowest readings recorded over the last year.

For many, this will be yet another sign that the market is completely dumb, out of whack, manipulated — whichever qualifier you may want to use. But it really is not. First of all, because a reading of 85 in VVIX, in the grand scheme of things, is really not that abnormal.

So essentially, in a matter of just a few weeks, the marketplace decided to stop buying these arguably most expensive insurance contracts. That points in one direction: risk perception is shifting. And that may genuinely be for good reasons.

With the Iranian war “out of the way,” the immediate catalysts for markets going down are off the menu. You would need something genuinely new and unexpected to rattle the markets. In other words, nothing in the current chatter of the day is representative anymore: it is all priced in.

And with summer now just a few weeks away, one has to squint really hard to find a reason why we would exit this quiet regime any time soon. The first FOMC under Warsh in mid-June? Likely a market mover, but we will have to wait and see if it turns out to be a true disruptor. How about inflation, the AI bubble, etc.? VIX 15 tells you everything you need to know: they may be an issue eventually, but not in the next 30 days. Lastly, we don’t know if you’ve noticed, but the market is clearly gearing up for some astronomical IPOs to land soonish.

Human beings are a paradox: when things are quiet we want movement, and when there is movement we want quiet. We forget the bad of one regime and fantasize about something new, something different. Be careful what you wish for. We don’t mind calm and boring — in fact, we crave it... as long as the market keeps baking in a healthy Trump Risk Premium. And while we won’t lose any sleep over a VVIX at 85, we will monitor closely to see whether these 4 to 7 points of VRP — observed consistently for more than a year in US equities — will erode, and with them, some of the fantastic sources of edge we’ve had at our disposal.

In other news

The biggest story of the week, by a wide margin, was the leak of the draft US–Iran agreement and the rather eye-watering price tag attached to it. According to The New York Times via multiple wires, the proposal floats a $300 billion “international investment fund” to rebuild Iran, paired with a 60-day ceasefire, a commitment to keep the Strait of Hormuz unrestricted, and even a Lebanon clause aimed at calming Hezbollah. The architects, reportedly Steve Witkoff and Jared Kushner, even floated joint ventures between US energy majors and Tehran’s oil sector.

For context, $300 billion is roughly the GDP of Finland and a good chunk larger than the entire Marshall Plan in today’s dollars. Whether any of this lands intact is another question entirely — Trump has not signed, the two sides still disagree on who controls the strait, and Iran wants its $24 billion in frozen assets released up front. But the symbolic shift is real. For markets, the message is simple: the geopolitical apocalypse the VIX was pricing in March has, for now, been swapped for a deal — and an oil chart down 10% on the week.

Share

Thank you for staying with us until the end. As usual, here are two good reads from last week:

• Here is an extraordinary read about crypto market, edge and making big (big) money while you can. And even if that source of edge is now gone and impossible to extract, it should give you some ideas about where to look for places where you have asymmetry of information and not a lot of competitor. Spoiler alert: not in 0 dte.

• If you still do not know what a perp is, you still have a few weeks to catch up. Here is an article from the FT talking about their raise in popularity and why the convergence between tradeFi and defi may be closer than one was originally thinking.
  
  

That is it for us this week. We wish you an excellent NFP week, and as usual, happy trading.

Ksander

For today, we want to look at one of the corners of the market where calm is not the word that comes to mind.

Today we turn our attention to EWY, the iShares MSCI South Korea ETF, that has been having the year of its life. The KOSPI broke 7,000 earlier this month for the first time ever, the ETF is up roughly 88% year to date, and President Lee Jae-myung's campaign promise of KOSPI 5,000 now reads like a quaint understatement. Samsung Electronics, which is about 29% of the fund, reported a 756% jump in quarterly operating profit in April on the back of HBM memory chips going into Nvidia's Vera Rubin platform. SK Hynix, another 20% of the fund, is in the same boat.

Then, on May 12, a government official floated the idea of an AI "citizen dividend", a profit-grab on memory-chip gains for the benefit of taxpayers. EWY dropped 10% intraday, closed down more than 7%, and yet the ETF has already rallied back to new highs. Needless to say that the options market is all over the place.

Let's have a look.

The Context

EWY started April at $126.22. It closed Tuesday at $200.65. That is a 57% rally in less than two months, on top of an already historic year-to-date move, in a fund where two stocks make up nearly half the weight. The May 12 plunge is barely visible on the chart now, a $14 round trip that took five sessions to recover and push to new highs by May 26.

Samsung is shipping HBM4 to Nvidia at gross margins above 80%, contract DRAM prices are up 50% in a single quarter, and the company has effectively booked its memory capacity for the rest of the year. The surprising part is not the rally. It is that the move higher has been delivered in straight lines punctuated by single-day shocks and that pattern keeps realized volatility elevated.

Thirty-day realized volatility in EWY sits at 55.46%. To put that in context, the two-year median is 28.72% and the 80th percentile is 46.44%. The fund is in the top decile of its own historical vol distribution and has been for several weeks. By a comfortable margin.

Our forecast does not say the storm is over. The 30-day projection points to roughly 47.6%, a modest compression from current levels but still well above the historical median. The 21-day reads 50.8%, the 14-day 49.3%. The model is calling for elevated realized vol to persist, just not at the 55%+ prints we have been seeing.

Let’s be precise here before moving forward. The trade is not a bet that EWY stops moving and in fact, if you are not comfortable with the daily moves, but most importantly the sudden jumps inherent with a high volatility regime, you may want to pass that one.

Otherwise, the thesis is the same: it is a bet that EWY moves less than what the options market is currently charging for. Two very different propositions, and the distinction matters when you are sizing a position in a fund where one comment from Seoul can shave 10% off the close.

With that in mind, let's see how the options market has reacted and where the best opportunities lie.

Read more

And while technology was central to many discussions, the reaction from a stock perspective has been particularly muted, not to say completely uninterested. And ultimately, while this trade lost money (our first bad call in Q2), it will be crucial to understand if it was because our reading on volatility was wrong, or because the markets continued on their slippery trend and their lack of love for anything Chinese.

We obviously know the answer to that question and will not spoil it just yet. However, we take that as a reminder that one has to get comfortable with the variance game. Once a trade gets “tested”, the number one reaction from retail traders is to “defend” the trade and avoid the loss at any cost. We barely ever do so. Some trades will make money, some trades will lose money. It is important to manage the size of your position so that no position is big enough that an occasional loser will wipe out the gains of many good ones. Obviously we will not say anything against anyone trying to delta hedge a position that has accumulated too much direction. But think about the opportunity cost: there are new opportunities every single day! Why would one try to repair or defend something that will most likely fall in the “bad trade” category, when one could just close it and focus on the next trade?
Let’s have a look.

The trade

Two weeks ago we were highlighting the possibility to short the 29/33 strangle in KWEB expiring on May 29th. At the time, Chinese equities saw their implied volatility rising as uncertainty around President Trump's state visit to China was building up. And it was building up so much that our signals were flashing very strongly.

At the time, we saw an 82% probability that implied volatility sold in the region of 9 dte would exceed the subsequent realized volatility. We were also pretty much convinced that implied volatility would go down, while realized volatility would potentially go up.

These predictions gauge direction but not magnitude, and therefore a 21-point variance risk premium between implied and realized volatility felt safe enough to get into that trade without thinking too much.

But occasionally what looks like an absolute slam dunk… will miserably fail. And the worst part? Not even because the plan and the forecast were wrong.

Indeed, as soon as the tone between Trump and Xi was understood to be much more cordial than what many had anticipated, implied volatility collapsed and went from north of 45% before the trade to below 30% a few days later. Surely, such a volatility crush would have helped our position.

But the problem is in the magnitude of realized volatility: it went up, and not by a small amount, almost doubling in just a few days before stabilizing around 35%. Now, 35% is still lower than what we sold, and normally this should have left us in a good position to still make money on the trade. But a big acceleration in realized volatility like this often comes with a strong trend, and more often than not, to the downside. And this time was no different.

First of all, in typical Chinese equity style, you can see that a good chunk of the movement is realized overnight, which ironically makes delta hedging such short-dated options both detrimental and unrealistic in practice. The real killer here is gamma: those overnight jumps showed up in roughly half of the nights when we had this trade on. And with such a trend, not only did our strikes get blown away, but the underlying never really gave any indication that it could recover from there.

We insist: these happen and one needs to get comfortable with them. When you have an edge, they do not happen regularly enough to be more than mildly annoying. But there is very little you can do: when you underwrite insurance, once in a while you will have to pay the claim, and there is no amount of “defending” that will get you out of your obligations.

Let’s now have a look at the PnL attribution for each of the Greeks to confirm our analysis.

Share

The Greek PnL Attribution

Looking at the attribution, this trade tells a clean and somewhat unforgiving story. It was less a failure of our volatility forecasting than the consequence of a one-way move we could not outrun. Vega paid us, theta accrued, but delta — really gamma in disguise once you are short overnight gappers like KWEB — took it all back and then some.

Vega did its job early. As Trump and Xi traded warm signals, front-end implied volatility crushed from 45% down to the low 30s, generating roughly +$20 of cumulative vega level PnL within two sessions. Theta also performed as designed, accruing quietly to +$35 by exit — the rent collection that shows up regardless of what the headlines do.

The problem is that delta dwarfed both. At its trough on May 22nd, cumulative delta PnL hit nearly −$180, recovering only partway to −$100 at exit. There is no amount of theta or vega that offsets that in a two-week window, short of a perfectly-sized 2–3% rebound which never came.

The daily view makes the path even more apparent. The two killer sessions are May 21st (−$47) and May 22nd (−$65), where the delta bars dwarf everything else on the chart. Both moves happened overnight, with investors dumping shares into the broader Hang Seng tech rout that hit Hong Kong in mid-May. Notice also how theta stops paying meaningfully once the strikes are breached: the options are now either deep in or deep out of the money, with very little time value left to decay on a day-to-day basis.

With all that in mind, let's see if we should roll the position and try to get back in.

Read more