In July 2026, a fund running roughly 4x leverage was liquidated inside three weeks. Situational Awareness had returned 439% through the first half of the year and grown to a peak near $45bn before its concentrated positions in AI infrastructure names reversed. Losses reached 67% for the month. Margin pressure forced the sale of the public equity book to Citadel.
Over the same period, Millennium lost 2.1% while carrying gross leverage several times higher. ExodusPoint lost 0.9% and stayed up 3.5% on the year. Verition lost 1.1% and stayed up 4.5%. Point72 lost 3.3%. Citadel’s Wellington fund gained 5.9%, in part by buying the positions the forced seller had to move.
Four times leverage ended one fund. Eleven times leverage produced a rounding error at several others. That outcome is impossible to explain if gross leverage is treated as a measure of risk, which is exactly how most people outside the industry treat it.
The multiple carries very little information on its own. Everything depends on which balance sheet you divide by, which financing channel supports the position, and what happens to your margin requirement when the positions inside your book start moving together. This piece works through each of those in order, with the numbers attached.
Gross leverage is long market value plus the absolute value of short market value, divided by net asset value.
Net leverage is long market value minus short market value, divided by net asset value.
Take a fund with $1bn of investor capital holding $3.3bn of longs against $2.7bn of shorts. Gross exposure comes to $6bn, giving gross leverage of 6.0x. Net exposure comes to $600mn, giving net leverage of 0.6x, or 60%.
Gross leverage answers a financing question. It tells you how much of a prime broker’s balance sheet the fund occupies, how much collateral has been pledged, and how large the position footprint is relative to the capital standing behind it.
Net leverage answers a directional question. It tells you what a 1% move in the broad market does to the fund before any stock-specific effects arrive.
A market-neutral book at 8x gross and zero net has no index sensitivity and very high financing sensitivity. A concentrated long book at 2x gross and 2x net has very high index sensitivity and modest financing sensitivity. The two shapes fail in completely different circumstances, which is why the July outcomes diverged so sharply.
The fund that was liquidated held the second shape while being described publicly with the language of the first.
Anyone comparing leverage figures across funds needs to know which measurement produced them, because the definitions do not agree with each other.
Prime brokerage gross leverage counts the market value of equity positions financed on that broker’s book, divided by the capital allocated against them. A fund with three prime brokers has three of these figures, and none of them captures anything financed elsewhere.
Regulatory gross assets to net assets, taken from Form PF filings, counts total assets against net assets. Fixed income positions financed through repo enter close to full notional, so a fund with a large Treasury book will report a much larger multiple here than any single equity prime would recognise.
Margin to equity counts the collateral actually posted against the capital base. This is the figure that determines survival, because it measures how much adverse price movement separates a fund from a forced sale.
The gap between these is not small. Across the industry, aggregate gross assets of roughly $11.8trn correspond to average leverage of 2.6x. Within that same population, multi-strategy funds have reported gross assets to net assets of 11.8x, while their equity prime brokerage leverage sat closer to 4.4x. Industry leverage measured against NAV has been put at approximately 8x, against roughly 5x a decade earlier.
All of these numbers are correct. They describe different denominators.
Fundamental long/short equity operates at the low end of the range, and the levels reached over the past eighteen months are historically extended for that category.
Gross leverage across the largest prime brokerage books rose for a third consecutive year to a record at the end of 2025, with net leverage near three-year highs at the same point. The global gross figure stood at 285.2% in November 2025, having risen 12.4 percentage points across that year, and one large prime book put it at 297.9% by the end of the month, a five-year high.
By June 2026 the measure reached roughly 294%, and the cumulative increase across the first five months of the year was the fastest recorded since the series began in 2016.
For context on how modest the equity numbers are in absolute terms, US long/short strategies more recently showed total leverage of 208.1% with net leverage of 52.8%, the latter around the 45th percentile of its trailing one-year range.
A long/short equity manager at 2.9x gross has roughly $2 of financed exposure per dollar of capital. Anyone arriving at a multi-strategy platform from that background will find the balance sheet unrecognisable.
At the end of November 2025, quantitative funds averaged 645.3% gross leverage on prime brokerage books, with multi-strategy funds at 444.3%. On the regulatory measure, multi-strategy gross assets to net assets had already reached 11.8x by the third quarter of 2024.
Those figures represent the normal operating state of the businesses, not a stretched position ahead of a specific trade. A platform that dropped to 3x would be unable to deliver the return profile its investors have paid for.
Understanding why requires working through the volatility arithmetic, which is where most explanations of hedge fund leverage stop short.
Consider a platform running forty pods. Each pod operates a market-neutral book that generates roughly 2% annualised volatility on the capital allocated to it. Risk teams enforce low correlation between pods through factor neutralisation, position overlap limits, and sector caps, so assume average pairwise correlation of 0.05.
The volatility of the combined portfolio at one times gross works out as follows:
2% multiplied by the square root of [(1/40) + (39/40 × 0.05)]
2% multiplied by the square root of 0.0738
Approximately 0.54% annualised
A fund producing 0.54% volatility earns almost nothing above cash once a pass-through expense structure and a performance fee have been applied. Allocators underwriting these platforms want something in the region of 6% volatility with a Sharpe ratio above 2.
Getting from 0.54% to 6% requires multiplying gross exposure by roughly eleven times.
The 11.8x figure reported in regulatory filings is the arithmetic consequence of that target, arrived at by dividing the required volatility by the volatility that a genuinely diversified book produces on its own. Nothing about it reflects an appetite for risk in the ordinary sense.
The alternative construction, concentrating the same capital into a smaller number of higher-volatility pods, reaches the identical 6% target at far lower gross leverage. It also produces much deeper drawdowns, far more dependence on individual portfolio managers, and a return stream that allocators price at a lower multiple. The industry chose the leveraged version deliberately.
Every figure above depends on the correlation input of 0.05.
Raise average pairwise correlation from 0.05 to 0.30 and hold everything else constant. The unlevered portfolio volatility rises from 0.54% to roughly 1.13%. At the same eleven times gross, realised fund volatility moves from 6% to approximately 12.5%.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.