Eclipse Technologies

Today, I want to introduce you to a new way of thinking about selecting rule-based investments. Nearly every long term fundamental/ quantamental strategy begins as a portfolio containing zero stocks, and your algorithm’s job (or yours) is to buy stocks before they appreciate in value. Even momentum strategies focus on adding stocks to a portfolio as they continue to appreciate in value.

Recently, I’ve done some experimentation with the exact opposite of this idea and found interesting results- begin a portfolio by owning every stock in your investable universe. For example, every stock in the S&P500. Then, instead of adding stocks that are expected to appreciate in value, you remove the ones that underperform over a period of time, let’s say, the poorest performing (bottom 25%) stocks each quarter, and gradually increase your exposure to the best performing ones. You can also adjust this strategy to make sure that you’re never overexposed to any particular stock over the portfolio’s lifespan. (Given the lifespan of the portfolio and rebalancing frequency, it’s pretty easy to find the exact percentile to remove so that your final percent exposure to each stock remains below (or above) a certain threshold.)

Before doing a backtest on this strategy, let’s take a closer look at the philosophy behind it.

Contrary to the method of picking stocks before they become winners, we are already invested in those winners, so we capture all of their upside rather than just a part.

The opposite side of this argument is that we are also invested in the downside of these stocks until we remove them. The only way to determine if this is the beginning of a worthwhile strategy is to backtest. But what question are we specifically trying to answer by backtesting?

The question then, is whether or not capturing all of the upside and all of the downside of the entire investable universe (considering decreasing diversification) on average results in outperformance or underperformance of the benchmark S&P500. More coming soon on this.