The following chart shows the 30-day moving average of the number of daily trades since the fund was launched.
Chart 1: 30 day moving average of the number of daily trades Source; OQFM
The number of daily trades is high and has been trending higher.
First, we’ll address why we believe it’s desirable to execute a large number of trades. Next, we’ll analyse why the number of trades has been increasing.
Why is the number of trades high?
Low turnover factors have some undesirable characteristics
When we analyse quant factors, there are numerous performance metrics that we review, including factor decay statistics which measure the extent to which predictive power declines over time. Factors with low decay rates typically have relatively poor short-term predictive power but their predictive power only declines gradually over time. Factors with high decay rates typically have relatively strong short-term predictive power but their predictive power dissipates quickly.
The most notable factors with low decay profiles (ie. low turnover factors) are value, quality, and profitability factors. Among this factor cohort, value factors are the most important as, over time, they have exhibited the strongest predictive power. Quality and profitability factors don’t work well in isolation and are only alpha accretive when used in combination with other factors.
The big problem with value factors is they’re prone to periods of underperformance. For example, value factors significantly underperformed from 2016 to 2021.
This is why many investors view value as a risk factor.
It can also be argued that the time required to close the valuation gap (ie the difference between fundamental values and share prices) has increased and is likely to continue increasing.
The following quote from Bloomberg references an October 2022 interview with David Einhorn, one of the best-known contemporary value managers.
“There’s just very few of us left,” he said, adding that most market participants these days are not trained or experienced in value investing, or have shifted to passive or quantitative investing. Fewer players means (sic) there’s no one to notice what’s happening to these companies and “nobody knows what anything is worth,” Einhorn said. “So there’s an enormous number of companies that are dramatically mis-valued in ways that we havent seen before.”
This is a potential structural headwind which will adversely affect the performance of low turnover value strategies. Stock prices will still converge towards fundamental values, but it may take a very long time.
Liquidity driven pricing distortions in Asia are persistent and exploitable
Trading flows which aren’t driven by stock fundamentals occur for many reasons, including fund subscriptions and redemptions, changes in leverage, and panic selling.
These flows create mispricing anomalies but if there are too many sophisticated investors targeting these anomalies, they’re ephemeral and difficult to exploit. Fortunately, Asia is very different to the United States in this regard. The weight of sophisticated money does not result in the opportunity set being arbitraged away.
The opportunity set is particularly large during market dislocations. For example, we have a strong track record of exploiting liquidity distortions following the earthquake and tsunami crisis in Japan (March 2011) and the Great Quant Unwind (August 2007). Undoubtedly the panic selling in March 2020 would also have generated numerous opportunities, but unfortunately our new fund (with OQ) didn’t launch until June of that year.
Maximising breadth
We talk a lot about breadth – and for good reason. With large breadth, we have a relatively low hurdle for success. In effect, we’re playing a “probability” rather than a “perfection” game. We just need to get it more right than wrong to generate very attractive risk-adjusted returns.
This is in accordance with the Fundamental Low of Active Management which states that a manager’s risk-adjusted performance is function of skill (IC) times the number of independent bets (Breadth).
We have over 1,000 positions in the portfolio. All other things being equal, more positions equate to more trades.
The breadth argument also applies to the number of trades we execute. Not every trade will be a winner, but so long as the majority of trades appropriately adjust position sizes, we can generate strong returns.
Minimising market impact costs
As quants, we know the extent to which transaction costs can erode pre-cost alpha. We can’t control stamp duty costs and we have limited control over brokerage and slippage costs. However, we can minimise market impact costs by trading appropriately.
The biggest driver of market impact costs is order size relative to market liquidity. By executing a large number of trades, we can minimise order sizes, thereby minimising market impact costs.
Our portfolio management system (Conquest)
Conquest facilitates intra-day trading by providing live relative volume data, dual listing spread data, and intra-day technical indicators.
Conquest also integrates seamlessly with the order execution system we use (Tora). Inventory for short selling is sourced in real-time to ensure all trades are sent directly to the relevant market.
We’re not hostage to backtest results
We maintain a healthy level of skepticism when reviewing backtests. I don’t think I’ve ever seen a backtest where the simulated long portfolio and short portfolios don’t move in opposite directions, showing consistent alpha generation.
We’re willing to make refinements to our investment process even if we can’t fully backtest them. We place a lot of emphasis on whether the refinements make sense given the mispricing opportunities we’re seeking to exploit.
Our backtest platform has improved (more on this later) but it doesn’t include tick by tick data to test intra-day entry and exit trades. We take a pragmatic approach, and this doesn’t preclude us from timing trades based on intra-day share price moves.
Preparing a trade list at the start every morning and executing those trades regardless of what transpires over the course of the trading day is suboptimal.
Our experience
Our experience executing a large volume of trades over the course of the trading day spans more than 17 years (I launched our Asian hedge fund at Macquarie in 2005). Our skills have been honed over many years and over time our understanding of Asian stocks and market structures has grown. This facilitates faster decision making.
Why is the number of trades increasing?
New rebalance heuristics
Our rebalance process comprises two components:
Our alpha screens which we run at various frequencies based on each screen’s decay profile.
Rebalance heuristics which are run throughout the trading day.
We’ve recently added several rebalance heuristics to our rebalance process. These heuristics include intra-day returns and relative volume indicators, combined with our alpha factors. They generate a relatively large number of trading suggestions which has contributed to the increase in trading frequency.
New alpha screens
Our investment process is constantly evolving. We’re always looking for quant factors which are positively correlated with stock returns and factor combinations which improve the predictive power of our stock selection process.
Recently, we’ve added new alpha screens using target price data, based on both individual analyst forecasts and consensus forecasts.
Improvements to our backtest platform
Our backtest platform includes over 20 years of data for all key equity markets, not just Asian markets. This facilitates out-of-sample testing across different markets and market cycles.
Until recently, however, we didn’t have intra-day historic pricing data. We now have hourly data to facilitate testing of higher frequency trading strategies.
Conquest enhancements
Conquest is vastly superior to the system we used at Macquarie and the gap will continue to grow. I enjoy adding new functionality to the system and recent improvements include “live” factor scores based on real-time data.
Opportunities are often time sensitive and having a user-friendly and bespoke system tailored to intra-day trading is all-important.
コメント