About Win Rates and Signals

Posted on 28th Jun 2020 by the LSS team

I wanted to share a quick few thoughts with you about using our signals in your investment process, and what to expect in terms of win rates. Are we at LongShortSignal actually able to predict the market, if so why is no one else doing it and exactly how game-changing is our output? Those are a few of the things I will go into in this very short piece.

For starters, it is my firm belief that markets are inherently unpredictable to an extent and there's no magic potion that will allow anyone to predict how assets will behave today, tomorrow or a week from now. But while I don't believe in that, I do believe in changing momentum, statistical patterns and taking a probabilistic view.


If the S&P500 has cruised up to 3400 for the better part of a year and suddenly starts seeing 2, 3 or 4% down days, something is likely to be up and momentum is possibly shifting. Now while a human watching the market during those days may be inclined to give in to bias, statistical models can look at this and say "right, I've looked at this market in combination with a number of other assets or factors going back to 1990, and with 80% probability I know what comes next, let's get out".

This is essentially what our calls are based on. We are having the models make the tough decisions for us. Now does that mean that they will always get it right? Certainly not. If you look through our signals you will see that there have been plenty of times where they did in fact not get it right, though generally positions were closed when losses were still manageable. But our value really lies in *more often than not* being able to pick up on changing momentum and having you exposed in the right direction, whether long or short. 

On the whole, our models have a >70% likelihood of getting it right. We expect this number to decline slightly in the future but it is still a rather strong result. Now as some models may perform better than others (for instance our crypto altcoin models at times have a hard time keeping up with the erratically behaving assets underlying them), the real cherry on the cake comes from combining a few of our models in your own portfolio, like we do in our own Model Portfolio