Trading Strategies: Curve Fitting and Over Optimization

Chris and Tyler talk about curve fitting and over optimization of trading strategies. 

This episode is a lot shorter than normal, we touch on a single subject vs going deep on the markets. If you like it and want more, let us know what you’d like to hear. Also if you want more of the long form, tell us what YOU want!

Chris talks about what he looks for in a strategy to consider:

Predictability - the ability to predict a trade is forthcoming, not predicting the results. 

Repeatable - does this setup repeat over time.

Define-able - can you clearly define the parameters.

What does over fitting look like in trading setups? 

“I just want to be consistent.” 

Seeking high win rate systems lead us down the path of over fitting. 

Random walk price data and finding setups that you start believing are edges, but end up being curve fit on random price series. 

Testing in-sample and out of sample data (training and test data sets).

Train model on in-sample data, then test it on the out of sample data. If it works on both, you have a system that can initially be considered more robust and not curve fit. 

Resiliency is more of what we are looking for to get us in the ballpark on a system. It isn’t fragile, it has more potential to last over time. 

Oh but what about Black Swans?

Position sizing, taking exits, adding to positions, staying out of the market are where the magic happens when you add them to a resilient system. Where as applying these techniques to a fragile system increases the risk. 

Spoiler: Robust systems are uncomfortable to sit through.

Building a system that fits your temperament is a key to having success trading a robust system. Get comfortable with being wrong. Backtesting systems, getting experience with the system, learning how you respond while trading the system. 

Practice the basics of a system over and over. Get the reps in!

You can reach Tyler at @tylerhkling on Twitter and [email protected]

You can reach Chris at @chrisdmacro on Twitter and [email protected]