Whoever has to sell sets the price

The Liquidity Map · Stock market report

Whoever has to sell sets the price

On 29 July, six unrelated stocks bottomed in the same session. Nothing had broken at any of them. One holder was being forced to sell everything.

August 6, 2026 · 10 min read

Key takeaways

On 29 July, Micron, CoreWeave, Nebius, SanDisk, IREN and Bloom Energy all bottomed in the same session. One fund, on borrowed money, was being liquidated.
Within five sessions they recovered between 21% and 50%. That day's price said nothing about the companies, only that somebody had to sell.
Leverage does more than magnify the gain. It moves the exit decision to the bank you borrowed from, and the bank takes it on the worst possible day.
Figma confirmed what we had been watching since 23 July and rose 67% from the June low. SpaceX went the other way: it passed through the 112-120 range and closed at $108.
I am short oil from the 87-92 area since late July. If it keeps falling into September and October, weakness in equities follows it, not alongside it.

On 29 July, six stocks with nothing in common bottomed in the same session.

Nothing had broken at any of them.

One holder, on borrowed money, was being forced to sell every last share.

Definition

Margin call

The demand from the bank financing your positions to put more money into the account, because the collateral has fallen below its threshold. If you do not have it, the bank sells. Not at the price you would want and not on the day you would choose. The more you borrowed, the closer that threshold sits to today's price.

The seller

Six companies, the same low, the same day

On 29 July, Micron, CoreWeave, Nebius, SanDisk, IREN and Bloom Energy all bottomed in the same session.

They had fallen between 33% and 53% from the month's high.

In one month.

The Nasdaq 100 was down 10% over the same stretch.

When six companies with different customers and different numbers bottom on the same day, price has stopped talking about companies.

The explanation reached the financial press over the following days. A fund holding roughly $45 billion at the start of July, built almost entirely on AI infrastructure, received margin calls from the banks financing its positions.

On 30 July, before the market opened, its entire listed equity book went to Citadel in a single block.

In one morning.

By the end of the month, around $10 billion of that $45 billion remained.

Six trading days.

From the 29 July low to yesterday, across five sessions, Nebius rose 50%, Bloom Energy 49%, CoreWeave 49%, SanDisk 35% and Micron 21%.

Not everything that fell in July fell because of him. The sector had been correcting since the start of the month, and the chart shows it.

What was his: the last leg, the vertical one, and the exact day it stopped.

+50%How far Nebius rose from the 29 July low to yesterday. Nothing about the company changed that day. Only who owned the shares.

What happened on 30 July, in three lines.

Four of the stocks that were sold, plus the Nasdaq 100, measured from 1 July. The low lands on the same day in all four, while the index barely moves.

The usable part starts here.

The lesson

What a forced seller teaches you

First: borrowed money does more than magnify the gain, it takes the decision out of your hands.

The fund did not choose to sell on 29 July.

The company did not fail.

It sold because the banks it had borrowed from asked for money it did not have.

With leverage, you do not pick your exit. The lender picks it, and picks the worst possible day.

Second: being right does not protect you, and we have written about that before. The fund is still up on the year, thanks to its private holdings. Its AI thesis was not disproven in a week, it was simply liquidated.

Third, and this is the one that shows up on the chart. When several unrelated companies bottom on the same day while the broad index barely moves, the cause is not at the companies. It sits with somebody who has to sell.

What I avoid on days like that is buying into the fall. From the outside you cannot know how much the forced seller has left, and cheap is an opinion rather than a level.

Citadel did not catch a falling knife. It bought a negotiated block, with the seller at the table and the certainty that nobody else was coming with the same stock.

For the rest of us, the proof that the selling is over arrives on the chart, a day or two late.

400%The fund's exposure above its own capital, according to press reports. At that level, a quarter off the positions wipes out the capital.

The same proof, in its slow version, has just shown up on a position we have watched since June.

The check

What Figma did after 23 July

On 23 July I wrote that Figma had made its first higher low after twelve months of decline.

It was the first sign then, not a certainty.

Yesterday it closed at $28, 67% above the 25 June low.

The signal was not a prediction. It was the moment there was nothing left to guess, because price had already refused three times to go lower.

That report lays out the pattern at length, with Coinbase and with the first year after a listing. I am not repeating it here.

What matters now is that the thing being watched actually happened.

Anyone who entered after the confirmation caught two thirds of the move without guessing the low.

The invalidation is where it was: the 25 June low, at $16.84. Below it, the thesis is wrong and you leave.

That is the difference between watching and acting. While every month brought a lower low, there was nothing to do.

+67%How far Figma rose from its 25 June low to yesterday's close, after twelve months of decline.

Figma from listing to today. Twelve months of decline, then an area that holds.

The other position watched since then went exactly the other way.

SpaceX

The area being watched gave way

SpaceX closed at $108 yesterday.

Its first trading day had closed at $161, and four days later it was at $211.

49% below the high, in eight weeks.

From 211 to 108, with no bad news about the company. Just supply outpacing demand.

With price around 150, the structure pointed to a continuation into the 112-120 area. It got there in late July and did not stop.

On Monday it touched $105, the lowest level since listing.

That area was a marker, not a floor. Once through it, the discussion moves toward the psychological $100 mark.

None of this is unusual, nearly every listing does it. Why it always goes the same way I explained on 23 July, with Coinbase and with Figma, and nothing has changed since.

I still think SpaceX has a strong chance of becoming the largest company in the world within ten years. Eight weeks of chart have nothing to do with ten years, and that is exactly where people get confused.

$108Yesterday's close, against $161 on the first trading day and $211 at the June high. A 49% fall in eight weeks.

SpaceX since the 12 June listing. The 112-120 range was the level watched from 150 down. Price went through it.

Across the rest of the market, the picture is entirely different.

The frame

The broad index is at a high, not in crisis

On Tuesday the S&P 500 closed at an all time high.

Through the whole July fall it was 5% off its peak.

What broke, broke inside a single sector.

A $45 billion liquidation did not move the index by even 6%. That is how large the market is against anyone inside it.

The central bank held rates again, and the yield curve has normalised without the recession people were writing about two years ago. I wrote at length about why high rates are not the danger.

Summer seasonality has held so far, as expected.

What happens in recent listings and in the forcibly sold stocks are stories about them, not about the market.

-5%How far the S&P 500 fell from its high during the July drop. On Tuesday it closed at an all time high.

The signal I care about for the autumn is neither in the listings nor in the index.

Oil

What I am watching for September

I was long oil from the gap fill, from $69-67.

From 87-92 I started selling, with a clear invalidation above the area.

Yesterday it closed at 75.

Oil moves first. If it keeps falling into September and October, weakness in equities follows it, after it rather than alongside it.

I have held the position for a few months, the same as on metals, and in May it went exactly the same way.

I only trade when the reward against the risk is obvious. This was one of those.

What I watch next: yields falling, inflation with them, and a few deflation scares along the way.

When they arrive, that is when you load the boat. Until then you wait, exactly as with the listings.

It is also why I do not want to grow the equity side of the portfolio. The liquidity that comes next goes elsewhere, I think.

-18%How far crude has fallen from its 23 July high, which is the area I started selling from. Yesterday it closed at $75.

WTI crude from 15 June. July's low was 68.55 and its high 92.19, exactly the areas being discussed.

That leaves the part that matters for you: what you actually do with this.

How we are positioned

That is the diagnosis. The part that matters for you is what you do with it.

Equities are 20% of the portfolio and I do not want to grow that share. The rest is Bitcoin, Ethereum and 25% cash, because that is where I think the next liquidity goes.

That cash is not idle. It is there so we can buy when the scares I mentioned above show up.

I do use leverage, but not like that. On metals I use three times leveraged ETFs, on a small share, on short moves, with the invalidation written down before entry. The fund in today's report held four times its own capital in positions and had nowhere to be wrong.

On oil I am short from the 87-92 area. If that area is taken back, I am out, no discussion.

On the stocks knocked down in the liquidation, we are not chasing the bounce. Anyone who was not there on 30 July missed the easy part, and what remains is a market looking for its price.

We hold no position in SpaceX. We watch, nothing more. Eight weeks of chart is not enough for a decision, however much I like the company.

On listings where the structure has already changed, positions get built in steps, not all at once. We do not buy the absolute low and we do not wait for it.

Conclusions

A fund of roughly $45 billion was liquidated in six trading days, with its AI thesis intact.

Six stocks bottomed in the same session, then recovered between 21% and 50% in five days.

The price on a day like that does not describe the companies. It describes who was forced to sell.

Figma confirmed the structure watched since June and rose 67%. SpaceX went through the area watched and fell below it.

The broad index closed at an all time high on Tuesday, so the break stayed inside one sector.

The signal for the autumn comes from oil, where I am short from 87-92. If it keeps falling into September and October, the weakness in equities arrives after it.

What to watch this week

The forcibly sold stocks: does the 21-50% bounce hold, or does it fade back to where the selling began? Liquidating a fund does not change the companies' numbers, but it does not improve them either.
SpaceX: does it stabilise below 112, or does a fresh consolidation form and break lower? That decides whether the discussion reaches $100.
Figma: does the 25 June low survive its next serious test? While it holds, the structure stays changed.
Oil: does it keep falling into September and October? If it does, the weakness in equities is being set up. If it takes back the 87-92 area, my position from there is wrong.

Forced Sellers Set the Price

Keep reading

2027 is THE YEAR

The consensus wants an October bottom. A green August says otherwise

The math for a rate cut