Flows & Liquidity

• Preliminary hedge fund performance data by Pivotal Path
• revealed that TMT Equity Sector and Multi Strat hedge
• funds have been at the center of hedge fund deleveraging
• and forced tech position liquidations in July. The former
• lost 10.2% in July while the latter lost 2.3%.
• Given their more diversified equity exposure, Equity
• Long/Short hedge funds were flat in July helped by a flat
• MSCI AC World index on the month. Similarly, Discre-
• tionary Macro hedge funds were slightly up in July given
• their more diversified exposure across equity indices and
• across asset classes.
• It is worth emphasizing that the 10% loss for TMT Equity
• Sector excludes Situational Awareness, which reportedly
• saw its assets collapsing from $45bn to $10bn. This
• makes the 10% loss look even more striking and suggests
• that, like Situational Awareness, several other TMT Equi-
• ty Sector hedge funds suffered from forced liquidations of
• semiconductor/memory stock exposures. Indeed, Figure 2M
• onthly performance of TM
• T Equity Sector hedge funds
• shows that the July loss would be the most extreme ever
• for this hedge fund category if it proves representative as
• further funds report performance.

• This is also true with Multi Strat funds, where their 2.3% loss in July would be the fourth largest in their history, as shown in Figure 3M onthly performance of M ulti Strat hedge funds. One needs to go back to the pandemic crisis of March 2020, or the Lehman crisis of 2008, or the collapse of the dot com bubble in 2000 to see larger losses for Multi Strat funds.

• The extremity of the July loss in the preliminary figures
• raises questions about the risk management frameworks
• of both TMT Equity Sector and Multi Strat hedge funds
• given that it appears they failed to prevent concentrated
• positions in semiconductor/memory stocks from building
• up excessively in recent months. Was concentration risk
• underestimated? Was volatility/correlation risk embedded
• in option positions mishandled? Did stop-loss / risk-bud-
• get discipline fail or was it overridden? Were financing /
• margin /liquidity dynamics not properly stress-tested?
• Going forward, the AUM decline of these two hedge fund
• categories in July means that their risk budgets are
• mechanically shrunk. If they also apply more stringent
• risk management frameworks and concentration limits
• and if prime brokers reduce balance sheet space allocated
• to these strategies, then the capacity of TMT Equity Sec-
• tor and Multi Strat hedge funds to hold tech exposures
• would be structurally more limited going forward.
• If this assessment proves correct and the capacity of
• hedge funds to hold tech exposures is structurally
• reduced, the tech trade would become over the longer-
• term even more dependent on retail investors and thus
• more susceptible to the swings emanating from leveraged
• ETFs, retail option buying and retail margin accounts.

Room for policy uncertainty to push term premia higher

• The FOMC meeting injected volatility into bond markets,
• with 10y UST yields initially rising by around 13bp from
• the previous day’s close into Friday, before retracing
• much of that move this week. However, the curve remains
• steeper. In principle, the three dissents in favour of a hike
• were a hawkish surprise. But the subsequent comments by
• Chair Warsh have fuelled concerns of greater uncertainty
• over policy rates and questions over Fed credibility.
• Indeed, as our colleagues in US rates strategy pointed out
• (UST Market Daily, Jul 29th), a suggestion that policy
• rates might not need to do the heavy lifting raises the risks
• of more active attempts at shrinking the balance sheet,
• and Chair Warsh paid only lip service to the Fed’s 2%
• PCE inflation mandate, raising the risk that he may look
• at a wider array of measures to determine the future path
• of monetary policy and also rely on the inflation task
• force’s findings to ratify his world view.
• This in turn has raised questions in our conversations over
• the potential impact on term premia from an increase in
• policy uncertainty. One way to look at policy uncertainty
• is to look at the dispersion of economists’ forecasts for
• central bank policy rates. This is shown in Figure 4Fed, ECB, BoE and BoJ policy rate forecast standard deviation for 4Q27,
• which shows the standard deviation of policy rate fore-
• casts from economists polled by Bloomberg for 4Q27 for
• the Federal Reserve as well as the ECB, BoJ and BoE for
• comparison. It suggests that the period after Warsh took
• over as Fed Chair has coincided with an increase in the
• standard deviation of Fed policy rate forecasts relative to
• other G4 central banks.

• What about term premia? Measuring term premia, or the

compensation investors require for holding long-term fixed income instruments rather than rolling over a series of short-term bonds, is clearly not straightforward, given that they are not directly observable and must instead be inferred by trying to separate out term premia from expectations of the path of future short-term rates. This can be done, for example, by using surveys to estimate expectations about future interest rates and to infer term premia. Another way is the literature around term structure models. Survey-based methods and term structure models have their benefits and drawbacks. Using survey-based estimates of future short rates is in principle simpler, though the downside is that they tend to get updated less frequently and the expectations of survey respondents may differ from those of market participants. Term structure models, by contrast, are available at higher frequency, but are also computationally more complex, subject to parameter uncertainty, and the resulting estimates of term premia can be very sensitive to model specification. These estimates can also change over time as model parameters are updated. For a fuller discussion on term premia, please see our previous F&L from Oct 2023 and In the eye of the beholder, Sep 2023. • With these caveats in mind, Figure 5Estimates of term premia in 10y USTs shows survey-based measures of term premia from the Blue Chip and NY Fed surveys, as well as three model-based estimates published by Federal Reserve researchers. These are the ACM model from the NY Fed, the Kim-Wright model from Federal Reserve Board staff, and the Christensen-Rudebusch model published by the San Francisco Fed (for details on the differences between the models, please see the FEDS Note: The Treasury Tantrum of 2023, Sep 3rd 2024). The different measures have shown relative stability since early 2025 after a significant normalisation from close to or below zero before the 2022 bond market sell-off. More recently, all three term structure models point to a rise in term premia since 10y UST yields troughed in late June, consistent with the view that elevated policy uncertainty could be contributing to the recent backup in yields.

The flows into hyperliquid ETFs stalled in July and August after surging in May and June

• After heavy outflows in May and June, crypto ETFs saw small inflows in July and August MTD (Figure 13M onthly flows in $bn into spot crypto ETFs ). Beyond bitcoin and ethereum which account for the bulk of crypto ETF AUM (with $77bn and $10bn of AUM respectively), there is around $2bn-$3bn of AUM in other crypto, predominantly Solana, XRP and Hyperliquid.

• Indeed, if one looks at monthly ETF flows as % of AUM (Figure 14M onthly flows as % of AUM into spot crypto ETFs), hyperliquid is the one that stands out with the largest inflow as % of AUM during May and June. This year’s rise in hyperliquid is also seen in total corporate crypto treasury holdings, within which hyperliquid is the fourth largest after bitcoin, ethereum and solana (Figure 15Total corporate crypto treasury holdings by underlying token).

• Whether hyperliquid eventually surpasses in market cap
• other tokens such as Solana and XRP remains to be seen.
• Hyperliquid’s value proposition is closely tied to platform
• activity, in particular transaction fees from perpetual
• futures trading. To broaden activity further, hyperliquid is
• expanding into prediction-style markets.
• That said, we see significant challenges to the market
• share of decentralized platforms such as hyperliquid.
• First, competition from onshore centralized platforms is
• increasing, as offshore decentralized venues are still
• exposed to concerns around unlicensed derivatives activi-
• ty, limited KYC/AML controls, manipulation/attacks/ora-
• cle failures and weaker consumer-protection safeguards.
• The launch of U.S.-regulated crypto perpetual futures
• products could accelerate a shift in liquidity away from
• offshore and decentralized venues to onshore venues. Sec-
• ond, competition from existing and new entrants in pre-
• diction markets remains intense.
• Perhaps these challenges are one reason the flows into
• hyperliquid ETFs stalled in July and August, as shown in
• Figure 14M
• onthly flows as %
• of AUM
• into spot crypto ETFs. Tracking these ETF flows along with its mar-
• ket share in trading/prediction markets would be key to
• hyperliquid’s outlook going forward.

Appendix

Europe India Africa/Middle East LatAm Taiwan South Korea U.S.

Japan

Short Interest Monitor

Chart A11a: Cross Asset Volatility Monitor 3m ATM Implied Volatility (1y history), as of 3rd Aug -2026

This table shows the richness/cheapness of current three-month implied volatility levels (red dot) against their one-year historical range (thin blue bar) and the ratio to current realised volatility. Assets with implied volatility outside their 25th/75th percentile range (thick blue bar) are highlighted. The implied-to-realised volatility ratio uses 3-month implied volatilities and 1-month (around 21 trading days) realised volatilities for each asset.

Spec position monitor

Mutual fund and hedge fund betas

CTAs – Trend following investors’ momentum indicators

Corporate Activity

Pension fund and insurance company flows

Credit Creation

Bitcoin monitor

Japanese flows and positions

Commodity flows and positions

Corporate FX hedging proxies

Non-Bank investors’ implied allocations