132 - How to track poor performing trading strategies

A few weeks ago I got the fright of my life.

I found myself in a situation where I could potentially be crushed by a car, or perhaps even a bus or truck.

As I’ve mentioned in the past, I like to go riding on my mountain bike, and when I ride my bike, I like to go pretty fast and do things that perhaps the casual rider may think is a little crazy.

On this particular day, I was hurtling down my street like a missile, however, I didn’t realise I was about to find myself in a situation that could potentially put my life in danger.

At the end of my street is a pretty busy road, it’s one of the major roads for the suburb that I live in, and as I was approaching the intersection I squeezed on my back brake to start slowing down.

But as I squeezed the brake lever, barely anything happens.

I hardly slow down at all.

So I squeeze the brake lever more, and then as hard as I can, until the brake lever can’t go any further, but I’m still not slowing down, and now I’m much closer to speeding into oncoming traffic.

Now luckily, I’ve still got the front brake, however it can be dangerous to try braking from high speed with just the front brake, it could be quite easy to brake too hard and end up flying over the handlebars onto the road.

Anyway, I managed to stop just before I get to the traffic, and I take a quiet moment to pause and reflect on what’s just happened.

And then I remember…

Over the past few months, my back brake has started losing its effectiveness.

In previous rides I’d noticed that I had to squeeze the brake a lot harder to stop and I’d even considered taking the bike in for a service to get the brakes fixed, but it had slipped my mind and on this day I had completely forgotten about it.

I was aware of the brake performance degrading, but I hadn’t kept track of it until it was almost too late.

One of the challenges we have as traders is keeping track of strategy performance and trying to determine when a strategy has degraded enough that it’s time to stop trading ir, or take it in “for a service”.

There are a number of metrics and techniques we can use to monitor strategy performance, and we’ve covered quite a few of them on the podcast already. Today I want to share an approach that can be used to measure not only how much a strategy degrades from in-sample to out-of-sample, but also how the strategy degrades over time, so you can determine when it’s dangerous to keep trading it.

But this technique doesn’t just do that, it can also indicate if a strategy is potentially curve-fit or overoptimized.

Sounds pretty good right?

Let’s listen to Perry Kaufman now as he explains what this technique is and how we can use it.