Incorporating Aggregate Predictions into Market Reports

I’m continuing to build out my vision for the State of the Market report with a new feature: aggregate predictions. For the uninitiated, the SOTM report is a daily snapshot of what some of the strategies that I’ve shared on this blog are predicting for the U.S. stock market the following trading day.

Previously, the report just showed each strategy’s individual prediction and left it up to the reader to combine and interpret. But often, those predictions ran contrary to each other and many readers had a difficult time making sense of it all. So this new feature provides all of that original data plus a combined aggregate prediction.

Take a look – it should be pretty self-explanatory.

Below are two hypothetical portfolios that assume an investor took a long/short position in the S&P 500 equal (in % terms) to either 1x (red) or 2x (green) the aggregate prediction from the close YTD (frictionless).

[logarithmically-scaled]

This is of course a backtest and all standard warnings about future performance apply. I am not implying that the report should be used this way or any way for that matter - trade at your own risk.

YTD, the strategy correctly chose the next day’s direction 58% of the time, with winning “trades” 1.5x larger than losers.

How the aggregate prediction is calculated:

It wouldn’t make sense to simply average all of the strategies’ predictions together. Some have performed well over decades of data, while some are very temporary themes specific to this market (such as the XLE Leader/Laggard Strategy). Some are very unique in their approach to the market (such as the VIX Spread) while some share common elements (such as the two proprietary indicators).

So I’ve used my own subjective logic to create the aggregate formula. I’ve over-weighted unique or particularly powerful strategies, and under-weighted others. But note that I’ve intentionally NOT optimized the weighting. In fact, other combinations worked considerably better in backtests; however, I felt that the way I did it was the most likely to stand up in the future.

One last note:

Bear in mind that we are dealing in probabilities. We will often be utterly and completely wrong – that’s the nature of trying to predict something that is inherently very, very difficult to predict. But also remember, we don’t have to be perfect – we just have to be right a bit more than we’re wrong to win the game.