Académique Documents
Professionnel Documents
Culture Documents
The image above shows the trend in Sharpe ratios of a representative long/short equity trading
strategy and similarly of a representative market-neutral ETF trading strategy. Based on the data,
we feel that market/neutral ETF trading should have better returns than market-neutral stocks
trading in future.
Allocators’ perspective
We believe that institutional investors may realize better returns when they invest in ETF based
quants as opposed to quants who are trading single stocks. In theory, there are a lot more
differences between single stocks and the conventional notion is that single stocks have a lot more
alpha to be made. However, that is not the case today. We will show here that a quant trading
ETFs is much more likely to realize alpha today. A quant-trading system that tries to derive alpha
from single stocks is more risky, more costly and more prone to overfitting.
Given the low cost, liquidity, availability of long historical data (either of the ETF or that of its index)
trading ETFs and not single stocks helps us remove any survivorship bias and allows for strategies
to be meaningfully tested over possible scenarios of future market events using walk-forward
optimization. Drivers of performance are easier to identify and to be controlled, improved and
fine-tuned for different market views.
ETFs allow us to source alpha from other asset classes
While old-school long/short trading on stocks has been limited to equities, the same on ETFs
allows us to cover virtually every asset class now. We can try to derive alpha in government bonds,
real estate, commodities, corporate bonds, and all of these both domestic and international.
ETFs require less operational effort than a portfolio of L/S stocks
If a data science team like Qplum focuses on trading ETFs, they will have to handle a lot less
complexity, like corporate actions, mergers and acquisitions, splits, reverse splits, dividends and
distributions. This allows the team to be more focused on finding alpha. By a very crude estimate,
a trading desk would have an extra 1-2 hours a day to work on alpha research, if they didn’t have
to spend it on operational management of a portfolio with 500+ stock. This, in turn, increases
returns and reduces costs.
As Brian Peterson at DV Trading remarked [2] at a panel along with Gaurav, it would take upwards
of $200 million today to setup a high frequency trading firm. The same is thus true for an equity
long/short trading firm since they are competing on the same alphas.
Does long/short trading work better on stocks than on ETFs?
We looked at returns of market neutral strategies on ETFs and stocks and this is what we found...
Till 2004 L/S strategies indeed had better returns on stocks
Net log-returns of long/short strategies on ETFs and stocks before 2004
The chart above compares the gross returns of long/short strategies on stocks and ETFs. Prior to
2004, it was indeed the case that there was a lot more alpha in stocks than ETFs. However things
changed since 2004. Perhaps due to Reg-NMS, which came into effect in 2005, the amount of
technological infrastructure needed to trade stocks against the basket became a lot more. As you
will see below, since 2004, long/short trading on ETFs has been performing much better than
stocks.
Since 2004 L/S strategies on ETFs have done better than stocks
Returns of long/short strategies on ETFs and stocks since 2004
Since 2004, long/short equity market neutral strategies on stocks have not done well. In fact since
late 2010, returns have not been positive. On the other hand, ETF based strategies have been
doing much better. The equity curve of ETF long/short strategies is much more stable. Look at the
financial crisis years. It is hardly noticeable. Even since 2010, while stocks based long/short funds
have failed to generate returns, ETF strategies have been growing steadily.
Dispersion between ETFs is high enough to achieve risk targets
A commonly quoted argument against ETFs is that they are very similar and there isn’t much
dispersion between them. It is claimed that exorbitant leverage is required to take risk required for
substantial returns. But this isn’t the case. In the chart below, we show the risk achieved by the
same strategy as above. Even when we constrain long + short leverage to under 6, the strategy
manages to achieve a risk of 6% in the low risk environment of today.
Annualized volatility (in %) of a Long/Short strategy on ETFs
There is a decent amount of dispersion being captured by ETF long/short trading.
A quant ETF strategy is less risky than a L/S strategy on stocks
Thanks to their idiosyncrasies, individual stocks are much riskier than the broad market. Therefore,
in order to manage their risk, quant shops investing in individual stocks generally end up artificially
limiting their allocation to some of the stocks. But diversification comes naturally with ETFs.
There are multiple stages in the adoption of any innovation. It has been studied time and again that
the early majority stage is when the reward to risk ratio peaks. ETFs are just entering the early
majority stage. The growth of ETFs is still picking up. There are about 2000 ETFs and many more
are being launched every year. Hence there is a lot of opportunity in ETFs right now.
ETFs are new and not many quants know how to trade them
Long/short equity trading has been done for many decades now. Books have been written about
how to trade stocks long/short. People know this trade very well. It is very difficult to source alpha
in it.
It is clear from the charts above that much of the alpha that was there earlier, pre-2004 that is, was
actually between the stocks and the basket themselves. Now that high frequency trading index
arbitrageurs are doing that trade, the alpha is not realizable for traditional long/short firms.
ETFs long/short on the other hand is new. ETFs themselves are very new. They were less than a
fifth of their current size ten years ago [5]. Read more about w hat US ETF market looks like
today[5].
There is more alpha to be derived in ETF long/short trading.
Conclusion
In many alpha seeking strategies, we are seeing higher returns when trading ETFs. All strategies
that look at longer term prediction often related to broader market factors and these trades can be
expressed more efficiently through ETFs. Much of the s hort-term alpha in single stocks has been
captured by High Frequency Trading firms, leaving little for equity market neutral StatArb portfolio
managers [2].
While hedge funds have been underperforming the average investor[6], and CIOs have been
shifting to ETFs[7], long/short funds should wake up.
References
1. Qplum’s flagship portfolio
2. The ten year evolution of quant: Trend Following, StatArb, HFT, Deep Learning
3. Three sources of Alpha and where high frequency trading fits in
4. Is yesterday’s quant strategy in tomorrow’s ETF?
5. What US ETF market looks like today
6. A detailed report on the investment patterns of US investors
7. Smaller endowments should focus on ETFs
8. ETF inflows are almost double of last year
9. Quant firm AQR seeks SEC approval to sell ETFs
10. Dynamic Flat CIO office needed today
11. One trillion more in ETFs than hedge funds now!
12. Qplum’s perfomance attribution
13. Reg-NMS invigorated the HFT eco system and make equity market neutral alpha tougher in
stocks
Disclosures
All investments carry risk. This material is for informational purposes and should not be considered specific investment
advice or recommendation to any person or organization. Past performance is not indicative of future performance.
Please visit our website for full disclaimer and terms of use.