How it actually works
A market maker's revenue is a function of volume. It stands between buyers and sellers, quotes a price to buy and a slightly higher price to sell, and pockets the gap. To illustrate with round invented numbers: buy at 99.98, sell at 100.02, four cents a go, twenty thousand times a day. The inventory it picks up along the way is held briefly and hedged. Long the ETF, short the basket that backs it; long the option, short the stock. Done properly, the firm is close to indifferent to whether the market goes up or down. What it needs is that the market keeps trading.
A directional bet abandons that indifference on purpose. The firm buys something and holds it because it thinks the price will rise, and there is no hedge on the other side quietly cancelling it out. Profit is the size of the position multiplied by the size of the . Again as illustration: a $1bn position that falls 3% loses $30m, and the same position funded with borrowed money at five times leverage loses $150m of the firm's own capital on the same 3% move. Nothing about that is exotic. It is what a hedge fund does, and hedge funds are structured, funded and disclosed accordingly, with lock-ups, drawdown expectations, and investors who signed up for exactly this.
Where the line blurs
The two activities are not cleanly separable in practice. Every market maker ends the day with inventory it could not offload, and every hedge is imperfect, so some directional exposure is unavoidable. The difference is intent and . Residual inventory is a by-product you try to minimise and unwind within hours. A directional book is a position you chose, sized, and intend to hold for weeks or months because you have a view.
That difference is invisible from outside. Both show up as "trading revenue". Both sit on the same balance sheet, funded from the same pool. A counterparty extending credit to a market maker is implicitly assuming it is looking at a spread-earning machine with fast-turning inventory. If it is actually looking at a levered long position in a single theme, the risk it thought it had priced is not the risk it has.
Why it matters to this crash
This is the whole substance of the story we covered on 18 August. The firm lost about $15bn in July, its first losing month in a decade, on directional bets that Bloomberg reported go well beyond what a market maker does; in its most recent debt offering it described its as having evolved to include longer-term positions, in line with a hedge fund. The FT separately attributes a $35bn July loss to 's leveraged stock bets and ranks it top of its all-time leaderboard of trading losses. We could not reconcile the two figures from published information and still cannot. The firm has still generated more than $40bn of net trading revenue this year.
The structural point survives whichever number is right. A bank cannot run that scale of directional risk, because capital rules and supervisors make the position expensive and visible before it gets large. A private trading firm faces neither constraint, funds itself in the , and is simultaneously load-bearing infrastructure: the price you get on an ETF or an option depends on it standing there. When the balance sheet that makes prices in a selloff is also the balance sheet losing money in that selloff, liquidity provision becomes correlated with the thing everyone is trying to sell. That is the thesis in one firm. For scale on the layer it sits in, Federal Reserve research cited in our coverage puts the at roughly $830bn as of September 2025.
What would make this dangerous
Watch for market makers widening quotes or stepping back from size in the weeks after a loss month. That is the moment the directional book starts charging the rest of us rent. Watch for more describing longer-term positions in bond offering documents, which is where this disclosure surfaced in the first place. Watch their funding costs: if bond investors start pricing these firms as hedge funds rather than as toll booths, the cheap balance sheet that makes the market-making viable gets more expensive. And watch for the same theme showing up across several of them at once, because the failure mode is not one firm being wrong, it is every firm that quotes prices being wrong in the same direction on the same day.