Robust signals or why you should not even consider building something else
Portfolio managers and quantitative researchers are always cautious about the use of overfitted signals or incorrect assumptions for trade decisions. The path of a quant is usually littered with dozens of research results that did not make it through the rigorous process of ensuring that a given alpha signal remains robust over time and adds sufficient value to existing strategies. Let’s highlight a few common pitfalls and discuss how to achieve robustness.
To provide an easy example, let's define non-robustness as a simple two period moving average cross-over. If the faster moving average is crossing the slower moving average from below, we would have a buy signal, and vice versa. Unfortunately, such a simple signal is representative of many technical indicators used in quantitative finance assumptions.
In the example signal above, two variables are responsible for trading decisions. One can add volatility filters or standardization methods but, in the end, the respective look-back periods of the two variables are the basis for trading decisions.
Any good researcher would now define an in-sample and out-of-sample period to determine robust results. In order to establish a repeatable process, we need to define the two most important characteristics of any strategy. First, the variable space and second, execution timing. To make this applicable to any number of signal inputs and time intervals, we define a lag metric between 1 and 5. For instance, our trading signals will be lagged from 1 to 5 periods between signal and trade execution, and a multiplier to scale our input variables by a factor. We choose this factor N to be between 0.1 and 5 and increase by increments of 0.5. As an example, we would multiply a 200-day moving average by a factor of 5 with N=5 and test our signal with a 1000-day moving average instead of 200 for one robustness check.
As the results in the surface plot reveal, this is not a reliable source of alpha. Cumulative performance is incredibly volatile and ranges even into negative territory.
To reveal how a better strategy would look like, the following plot shows an excerpt from the same analysis of strategies that would make their way into our portfolios.
As can be observed, volatility of cumulative performance is extremely low and even the worst combinations of input and lag variables show positive results. This signal would give us the great feeling of having a robust source of alpha within our portfolio.
Another topic we would like to cover today is a theory on how the failure of FTX might have affected the markets before the exchange's official bankruptcy.
Since July, crypto markets have been very challenging. Finding strong signals has not been easy. Absolute momentum did not get caught in the downside move, but trends were not persistent enough to generate outsized returns. Cross-sectional momentum did deliver slightly better returns, but still, nothing truly moved the needle. Carry barely added a different direction. Spot vs. futures arbitrage strategies had the spreads too low to generate returns and were rapidly jumping between positive and negative spreads.
Our hypothesis is, that since the 3AC default and due to the massive trading volume FTX had up to November, a lot of the trading done via FTX could potentially just have been done on paper, so that no real price discovery could happen via FTX.
Since November, we see signals emerge in all mentioned strategies, spreads have reduced volatility and under the surface of the markets there seems to be more price discovery.
If our hypothesis is correct, we expect much more exciting times ahead.
Please note, our commentary on FTX is just a speculation, we make no allegations and have no insight, other than the things we read and hear in the public domain.
This blog post is shared with the intention to provide educational content, general market commentary or company specific announcements. It does not constitute an offer to sell or a solicitation of an offer to buy securities managed by The Well GP.
Any offer or solicitation may only be made pursuant to a confidential Private Placement Memorandum, which will only be provided to qualified offerees and should be carefully reviewed by any such offerees for a comprehensive set of terms and provisions, including important disclosures of conflicts and risk factors associated with an investment in the fund.
Past performance is not necessarily indicative of or a guarantee of future results. The Well GP makes no representation or warranty, express or implied, with respect to the accuracy, reasonableness or completeness of any of the information contained herein, including, but not limited to, information obtained from third parties. The information contained herein is not intended to provide, and should not be relied upon for accounting, legal, tax advice or investment recommendations.
