Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Is it a bias to use all historical constituents of index in backtesting?

I calculate 30-year equally weighted return of stocks that has ever been constituents of S&P 500 and S&P 400. The return is much more higher than the index itself. Is it that the portfolio is biased?
I'm thinking that stocks may outperform the market before they are included but may underperform after they are excluded from the index.

2 responses

I believe the S&P indices are market cap weighted, so if you equal weight the components you might be getting a bit of small cap as well as value effect, which might explain the outperformance?

Usually stocks get a boost when included in an index, due to index funds and ETFs having to include them, and stocks that are excluded get sold off for the same reason.

Stocks are born and eventually they die. By using an equal weighted index of a large number of stocks you are essentially trend follwing the huge growth of stocks which rise from nothing. More mature stocks in the S&P 500, for instance, simply don't have the same growth potential. How many McDonald's outlets or bottles of Coke can the world put up with? Of course, you also pay the price for stocks which fall from grace, but that too is a part of the trend following nature of indices. In practice of course, even an equal weighted S&P 500 can take only very limited capital. But for the smaller investor that is a very different story.