Beating the Efficient Market Hypothesis August 31, 2018
If you have some sort of a background in finance, you may have heard of the Efficient Market Hypothesis (EMH). It’s an economic theory that states that all available information is reflected in prices. Any pricing inefficiencies are quickly arbitraged away by the others. No fundamental or technical analysis will consistently produce superior returns within the market. EMH proponents believe that we are better off holding an index fund; being the market, than trying to beat the market.
A running joke on EMH goes as such:
A economist (EMH proponent) and a friend were walking along the street.
The friend stops and says, “Look, there is a $100 note on the ground!”
The economist, without flinching, replies, “Impossible! If there was a $100 bill on the ground, somebody would have already picked it up!” And carries on walking.
I’m not sure about you, but I will definitely (at least) take a good look at whether loose $100 bills can be found near me.
Now let us talk about “momentum” – a market anomaly where stocks that have done well (or poorly) in the past, will likely continue to do well (or poorly) in the near term. Momentum has been described as a powerful and persistent market anomaly. If you are a die-hard supporter of the Graham and Dodd school of thought, momentum may be a counter-intuitive concept.
Interestingly, an EMH believer holding a S&P 500 index fund is adopting some sort of momentum. Losers are systematically kicked out of the index and replaced with new winners. The result is a constantly rebalanced group of 500 winners with an upward bias in price. Going further, as the share price of an individual index component… say… the AAPL, goes up, index tracking funds will have to purchase more AAPL shares to maintain their weightage within the index; pushing their share prices even higher.
No fluff, no bluff
Yes, there are already many academic papers on momentum (Jegadeesh & Titman 1993 comes to mind). But for further validation, I felt a need to conduct at least one experiment (I actually did many), to go through the full journey and see some results for myself. After all, seeing is believing.
Hypothesis:The best performing stocks in the past will continue to be the best performing stocks in the future.
We define “in the past” as the “lookback period”, “X” number of months into the past. Suppose it is August 2018 now, so a 6-month lookback period will cover February to August 2018. “In the future” is the “holding period”, “Y” number of months from now.
We quantify and measure “best performing in the past” as the top decile of total returns in the sampled pool, termed Q1. Q2 will be the next 10%… until Q10, the bottom 10%.
“Best performing in the future” simply measures total returns from now till “Y” months later.
So we measure and buy Q1 from “X” months, hold for “Y” months then rebalance… not rocket science!
I chose 3, 6 and 12 months as lookback periods and 1, 3 and 6 months as holding periods.
The sampled pool is NYSE, NASDAQ, and NYSE MKT. As a proxy for liquidity, only companies with market capitalization of at least USD200 million are included. I’ve also removed ETF, ETN, REITS and ADR from the pool.
The sampled period is 22 years from July 1996 to June 2018. To mitigate effects of seasonality, combinations with 6-month holding periods are done twice – 6-month A rebalanced in April/October and 6-month B rebalanced in January/July.
This gives us 12 possible combinations:
|Lookback Period (months:m)||Holding Period||Q1 (Winners)||Q10 (Losers)||Q Spread (Q1-Q10)|
Like an avocado: Too soon, not ripe. Too long, too ripe.
So we see… winners do outperform losers! Momentum does exist.
Spread measures the difference between the highest and lowest deciles (Q1 – Q10) and is positive across all combinations!
However the outperformance varies across different combinations. We find:
A. Momentum is an avocado. It has a narrow window for optimal consumption.
A 6-month lookback period seems to be the ideal, producing double digit Q1 and Q Spreads.
3 months and momentum is not ripe. 12 months and momentum gets too ripe, especially if paired with a longer holding period.
B. The ideal holding period is more diverse. Shorter holding periods (1 to 3 months) seem to outperform, as long as momentum is properly formed.
The top 5 performing combinations plotted against S&P 500 in an equity curve looks like this:
So we have… momentum busting the EMH – our first of many hypothetical $100 bills to be found. And I suspect the same model applied on Chinese or developing markets may produce even better results.
So what now?
In closing, I would like to emphasize: There is no perfect system that works all the time. The lesson to be drawn here is that the basic momentum model will serve as the backbone for our future strategies.
Thanks for taking the time to read this, and see you all next time!
(*Please note that information contained in this article is given for educational purposes only. All information and opinions provided in this piece do not constitute as investment advice.)
About the author
Tan Chek Ann
I sleep in the day and head the dealing desk at night. In my nightly work, I attempt to do analysis and scout around for trading ideas. The work I do tends to lean towards math and science as I’m not too much of an artistically-inclined person. Join me on my journey as I share my thoughts on various topics – though mostly still financial stuff. Reach me at firstname.lastname@example.org