The other day, the author of Nik’s Blog left a really interesting question in the Endless Metrics comment section.
Answering it in full detail is fairly complex, so let me paraphrase here it as, “What would your performance look like if you bought the best performers in the S&P 500 each year?”
In essence, if at the end of each year, we looked at the top ten best performing stocks in terms of price return for that year, and bought an equally weighted portfolio, held it for a year, then sold it and bought a new portfolio using the same methodology, how would we do?
My gut told me that we should probably perform on par with the market or maybe do a bit worse. After all, if it were so easy to just outperform, then why wouldn’t it be a more obvious strategy?
Well, the resulting outperformance shocked me:
This is a crazy good record of outperformance. It seemed so good that I had to check my methodology, which consisted of using the stocks in the S&P 500 as of today. But, even when I didn’t count performance of stocks that weren’t historically part of the index during the past years, this approach still worked!
The question then is, how is a portfolio made up of the top ten performers in 2021 doing thus far in 2022? As of writing, the S&P 500 is down -15.6%. However, last year’s top ten is only down -5.4%, which means more than +10% in outperformance.
This is just a small case study and I’m sure people can poke holes in how this methodology isn’t a perfect approach (primarily, it’s backwards looking and the past is not the future) but I am still surprised at the ease in which we got these results. Winners just seem to win!
Have a question you want to ask or a topic you’d like to see covered? Let me know!
Thank you, Luke! I think you could make your job a little easier in the future via QuantConnect, which lets you algorithmically test various strategies starting from the 1980s.
Joe brought up good concerns. Another concern with this methodology is that it measures (cherry picks) across a long (mostly) bull run period. This further skews the sample.