top of page

Lessons from China Quant Quake

Updated: Apr 24

Recently there has been considerable press coverage on the severe underperformance of numerous China quant funds.

On 22nd February Bloomberg reported:

“China’s quantitative hedge funds are admitting to unprecedented failures by their stock-trading models during one of the wildest two-week stretches in the market’s history. One manager described it as the industry’s ‘biggest black swan event’. Another said its models ‘switched from doing it right to getting it wrong repeatedly’. While historical data on China quant returns is limited, all signs point to record underperformance for such funds — a shock that Man Group has compared with the ‘quant quake’ that wreaked havoc on US managers in 2007.”

Most press articles have referenced a report published by Man Group on 20th February titled "China’s Quant Quake: How the Dominos Fell”.

According to this report, the main reason for the poor performance - or what they refer to as the “third domino” following the extended market downturn and increase in market volatility - was government intervention and the subsequent relative underperformance of small caps:

“The third domino tumbled following the intervention of state funds mobilised to stabilise the market during critical periods. However, this intervention unintentionally triggered market dislocation. The state funds’ initial focus on purchasing large-cap stocks in January and the beginning of February, followed by buying mid-caps, fuelled a divergence in the performance of large and mid-cap stocks versus small-caps, increasing the strain on small caps, exacerbating their losses, and thereby compounding the unwinding of the leveraged market-neutral strategies.”

It's not the first time state funds have intervened to support equity markets in China, so recent support measures should not come as a surprise.  It’s also not the first time we’ve witnessed quant fund contagion and forced unwinds.

And undoubtedly it will occur again.  Fortunately, it’s possible to avoid being a victim – so long as you heed several important lessons.

#1: Don’t underestimate crowding risks

Here’s my 10-step crowding lifecycle (as documented in the following research note:

  1. An investment strategy, or investment style, generates strong alpha.

  2. Allocators – who tend to chase what’s hot – invest more money in this strategy.

  3. New entrants emerge to satiate investor demand.  A virtuous circle ensues.

  4. It gets more difficult for the fund managers to generate alpha as the mispricing opportunities get arbitraged away.

  5. Fund managers deploy more leverage to generate the returns investors are accustomed to.

  6. A forced unwind event occurs.  It can be difficult to determine the precise catalyst (e.g. during the Great Quant Unwind).

  7. Panic ensues as fund managers are forced to de-lever at the worst possible time.

  8. Forced unwinding exacerbates the share price moves, resulting in extreme losses.

  9. Other investors sit on the sidelines until the opportunity set becomes so attractive that they take the opposite side of the unwind trades.  In extreme cases, a bailout package may be required to restore market order.

  10. The crowded investment strategy is now out-of-favour and the relative paucity of money makes it easier for the “survivors” to once again exploit the mispricing opportunities they’re targeting.

From what I’ve read, this is a pretty good depiction of how the China Quant Quake has played out.  In this case, the forced unwind event (step 5) is the Chinese government buying exchange traded funds to support the market.

It’s incredible how history has a habit of repeating in financial markets.  John Templeton famously wrote in 1993 (in his book “16 Rules for Investment Success”):

“The investor who says, ‘This time is different’ when in fact it’s virtually a repeat of an earlier situation, has uttered among the four most costly words in the annals of investing.”

Investing in crowded strategies has been – and will always be – risky.  Outside of China quant, it feels like index rebalance arbitrage is moving into the realm of crowded quant strategies.  Potentially other quant strategies favoured by multi-strats (who all tend to look at risk using the same framework) are also getting increasingly crowded.

As we’ve frequently stated, It’s a huge advantage running a differentiated strategy which targets an uncrowded alpha source.

#2: Over-reliance on backtest results is dangerous

I’ve reviewed countless backtests over the last 25 years – and over time I’ve gained a greater appreciation of the pitfalls which potentially undermine their usefulness.  I know from my own experience how easy it is to generate good backtest results.  Tweak a style timing model so that it favours growth from 2015 to 2021 and value from 2001 to 2007, and the results look spectacularly strong.

Some backtesting pitfalls such as lookahead bias are relatively easy to address.

Others such as underestimating costs are harder to overcome, particularly for high frequency strategies where market impact costs can erode pre-cost alpha and for long-short strategies where it’s difficult to calculate shorting costs.

However, the biggest pitfall which can completely undermine the veracity of the backtest results is overfitting.  Out-of-sample testing mitigates this risk, but the biggest safeguard is to understand the mispricing opportunity being targeted and to calibrate the investment strategy based on this understanding.

Related to this, a common-sense overlay is also appropriate.  Even though Chinese small caps may have outperformed large caps for a prolonged period, is this likely to continue?  Are there exogeneous risk factors that could jeopardise the continued outperformance of small caps?  Is it wise to rely on one key alpha source?  It doesn’t seem like the architects of the strategies impacted by the recent drawdown considered these issues when developing their investment processes, relying too much on backtest results.

#3: Size is a big risk factor

For any long-short strategy, the biggest risk factor is market beta.  Next comes country beta, followed by sector beta. 

After controlling for these risk factors, many other factors which explain returns on a univariate basis no longer exhibit statistically significant explanatory power in a multivariate framework. 

The exception is size.  Even after controlling for market, country and sector exposures, there are periods of time when large caps outperform small caps and vice versa.

The problem with getting short exposure via market index futures – which seems to be a common trait for the China quant funds that recently lost money – is that it’s impossible to avoid taking a big size bet.  When you short market index futures, you’re effectively shorting the largest and most liquid stocks.

More broadly, shorting index futures is problematic for other reasons.  This brings me to the next lesson.

#4: Index futures are a poor hedge for long stock positions

From a risk perspective, the best way to offset a long stock portfolio is via a short stock portfolio.  This facilitates risk exposure mitigation during portfolio construction.

In addition to size, numerous other risk factors can’t be properly constrained using index futures.  The extent of this problem depends on the futures contract.  In Asia, the ASX200 futures contract is problematic due to significant Financials bias, and the Kospi and Taiwan MSCI futures contracts are problematic due to the dominance of single stocks (Samsung Electronics and TSMC).  It’s very hard to constrain risk exposures using these futures contracts.

Onshore Chinese quants can short a broader range of index futures contracts than offshore investors (who are typically restricted to the China A50 Index).  Nevertheless, it’s impossible to escape the fact that shorting index futures will result in unwanted risk biases that cannot be effectively hedged.

#5: Excessive leverage is an unacceptable tail risk

While leverage is typically a great way to transform decent alpha into strong headline performance, it can be extremely risky.

The following quote is from a research paper I recently wrote on taking appropriate risk bets (

“[Leverage risk] is more of a ‘tail risk’ in that when markets are well behaved it’s not problematic. However, during market dislocations and extreme liquidity events, leverage risk comes to the fore.  Witness the demise of LTCM in 1998 and the downfall of numerous quant funds during the Great Quant Unwind in August 2007.”

High leverage can result in forced unwinds and often these unwinds occur at the most inopportune time, such as during market dislocations when short-term adverse price moves result in margin calls.  Without excessive leverage, it’s possible to withstand the adverse price moves, or even take the opposite side of the trade to profit from liquidity driven price distortions.  This was my experience in August 2007 when the Asian fund I managed registered a positive monthly return during a month when highly leveraged peers were decimated.

#6: Breadth is all-important

If I could use one word to denote the most important characteristic of quant investing it would be “breadth”.  Quant models don’t have strong predictive power, so breadth is required to generate strong risk adjusted returns.  This relationship is neatly encapsulated by the Fundamental Law of Active Management: IR = IC * sqrt (Breadth).

Breadth is required in terms of the number of positions, number of trades, and – importantly – the number of alpha drivers.

It appears that many of the China quant funds impacted by the recent drawdown were overly reliant on one alpha source: the outperformance of small caps.

#7: Stop losses are crude – and often ineffective – risk control mechanisms

I’ve been investing long enough to remember the role “portfolio insurance” played in the 1987 market crash, particularly on Black Monday (October 19, 1987).  This involved short-selling index futures when equity markets declined.

The following quote is from Investopedia:

“These computer programs automatically began to liquidate stocks as certain loss targets were hit, pushing prices lower. To the dismay of the exchanges, program trading led to a domino effect as the falling markets triggered more stop-loss orders. The frantic selling activated yet further rounds of stop-loss orders, which dragged markets into a downward spiral.”

This illustrates the danger of automatically selling based solely on adverse price moves.

I’ve never been a fan of stop losses, particularly for diversified portfolios with proper risk constraints, as documented in the following research note:

It appears that stop losses exacerbated the extreme market moves across A-shares and provided limited downside protection for quant funds.

On February 22nd, Reuters reported:

“Baiont Quant, which employs artificial intelligence (AI) in investment, said its strategies suffered record drawdowns, despite the market ‘tsunami’ triggering red alerts and stop-loss orders.”

Similarly, stop loss orders didn’t help quant funds during the Great Quant Unwind in 2007. 

A risk control measure that doesn’t help during periods when it’s most needed is of questionable benefit. 

So says history, including the very recent history of the China Quant Quake.


162 views0 comments

Recent Posts

See All

Investor sentiment & AI as a risk factor

“In the short run, the market is a voting machine but in the long run, it is a weighing machine.” This is a quote from Warren Buffett in 1987 who paraphrased a passage from Graham and Dodd’s seminal b


bottom of page