StockViz

Share this post

Beginner's Luck in Quant Investing

stockviz.substack.com

Beginner's Luck in Quant Investing

Success has many fathers, but failure is an orphan.

Shyam Sunder
Sep 18, 2021
Share this post

Beginner's Luck in Quant Investing

stockviz.substack.com

You only control the process, not the outcome. No matter how robust the research, it is inevitable that investment strategies will hit rough patches. There's a fair bit of beginner's luck involved as well: a good strategy that underperforms at launch is abandoned quickly, never to be revisited. Automating robust strategies and allowing them to paper-trade in perpetuity solves this.

One such example is our set of Factor Momentum strategies. The theory is pretty simple: just buy whatever factor (value, quality, momentum, etc.) that worked in the previous year and hold it for one month.

Soon after we first read the research, we setup Factor Momentum portfolios for both US and Indian stocks. So far, it has worked great for US stocks but not so great for Indian stocks. We wrote up an update here.

Does this mean that the strategy “doesn’t work?” Hardly.

Twitter avatar for @ReformedTrader
Darren 🥚 🐣 🕊️ @ReformedTrader
"I simulated ten different 5-year GBM processes with expected annual returns of 20% and 10% volatility. "You can reasonably expect a strategy with a known Sharpe 2 edge be underwater for at least 2 out of 5 years if your luck falls in the bottom decile." robotwealth.com/what-pl-swings…
Image
3:28 PM ∙ Apr 11, 2021
53Likes4Retweets

The “luck” factor plays a very big part in short-term investment results.

Speaking of luck, what if…

The Fama-French Value Factor premium was due to luck

We are big fans of factor investing, especially momentum. However, like everyone else, we started with value investing but quickly found out that we didn’t have the stomach for it. A few months ago, Mathias Hasler of Boston College published this:

The construction of the original HML portfolio (Fama and French, 1993) includes six seemingly innocuous decisions that could easily have been replaced with alternatives that are just as reasonable. I propose such alternatives and construct HML portfolios. In sample, the average estimate of the value premium is dramatically smaller than the original estimate of the value premium. The difference is 0.09% per month and statistically significant. Out of sample, however, this difference is statistically indistinguishable from zero. The results suggest that the original value premium estimate is upward biased because of a chance result in the original research decisions.

Back when most of these landmark research were published, data was hard to come by, difficult to store and often had to be processed by hand. Understandably, academics took shortcuts - using monthly returns, having Dec-31st as cut-offs, not discriminating between revised numbers vs. those first published, etc. While the underlying alpha might have still survived, the magnitude of it is often questionable in the present day and age where none of those data storage, management and compute problems exist.

And then, there is the famous “specification problem” in quantitative investing - the same underlying alpha can be harvested in multiple ways (different variable configurations) so each model will be (un)lucky in different time-frames.

The case against Value in three parts

The market, and by extension, is not as dumb as professional investors like you to believe. Mispricings are rare and fair-value is in the eye of the beholder (bag-holder?)

If the S&P 500 or any stock practically never is at or near fair value, the question is why bother calculating the fair value in the first place? If the Second Law applies and a market drifts towards the most likely state, it will never end up at fair value. If true, that would mean that fair value calculations are entirely useless.

Klement on Investing
Should we abandon fair value calculations? Part 1
Some time ago, a reader sent me a collection of essays by a strategist from a renowned asset manager. In one of these essays, the author tried to tackle a seemingly philosophical problem and observed that the price chart of stocks or a market index like the S&P 500 is not time invariant, i.e. one cannot reverse the chart in time and expect to see someth…
Read more
2 years ago · 4 likes · 3 comments · Joachim Klement

Ending wealth is multiplicative, not additive. If ergodicity economics is right, then fair value models are indeed useless and misleading. And well-established investment approaches like value investing would turn out to be a historical aberration that worked for 70 years or so but did so more or less by coincidence.

Klement on Investing
Should we abandon fair value calculations? Part 2
Yesterday, I talked about how thinking about markets in analogy to thermodynamics leads us to the conclusion that if we take the Second Law of Thermodynamics seriously, share prices should almost never be at or near fair value. Similarly, a market index like the S&P 500 should almost never trade at fair value because it is incredibly unlikely that the v…
Read more
2 years ago · 4 likes · 4 comments · Joachim Klement

People try to maximise ending wealth and not returns, they are simply refusing to enter gambles where they can lose a large amount of their wealth even if the expected return improves. And because many investors independent of their risk aversion refuse to participate in stock markets at the same time, the drop can be fast and furious.

Klement on Investing
Should we abandon fair value calculations? Part 3
Over the last two days, I have done my best to lose as many readers as possible by talking about the Second Law of Thermodynamics, its implications for finance, and the ergodicity problem. For the few readers whose eyes haven’t glazed over yet, I want to talk about an intriguing experiment today and what that means for financial markets…
Read more
2 years ago · 3 likes · Joachim Klement
Twitter avatar for @PolemicTMM
Polemic Paine @PolemicTMM
Dispite all the billions of words written, courses studied and papers authored, company stock prices go up for only two reasons. 1. The value of the company goes up. 2. People think the value of the company will go up. 99% of the time it's 2.
5:53 AM ∙ Sep 4, 2021
206Likes21Retweets

Quite a bit to chew on this weekend!

Meme of the week


The best way to get started investing is to Get Started!

Share this post

Beginner's Luck in Quant Investing

stockviz.substack.com
Previous
Next
Comments
TopNewCommunity

No posts

Ready for more?

© 2023 StockViz
Privacy ∙ Terms ∙ Collection notice
Start WritingGet the app
Substack is the home for great writing