Understanding the Walk Forward Analysis Technique
Walk Forward Analysis is an extremely effective technique to optimize trading system parameters. However it can sometimes be hard to understand why and how it works. This page explains.
As opposed to the basic way of optimizing a trading system, i.e. using one single optimization phase, followed by one single walk forward phase, Walk Forward Analysis (sometimes called Walk Forward Optimization) uses multiple stages.
The benefits of this approach are huge, since it enables trading systems to be optimized in a way that keeps them effective in changing market conditions.
A simplistic way of understanding this is to consider two distinct classifications of trading systems: 1. Mean Reverting, and 2. Trend-based.
Sometimes markets will be in a trading range, where mean reverting systems will perform well, but trending systems will get whipsawed and lose money. But when a market is trending, the reverse will be true. So the ability to optimize a system based on the market conditions at certain points in time is clearly going to be a more effective approach than optimizing based on averages of different market conditions and dynamics.
Of course this is a very simplistic example. Often market dynamics change in ways we don't fully undertsand. Where a simple moving average cross-over system used to work 20 years ago, it no longer does. This is because the market dynamics have changed, and are very different today.
The great thing about the Walk Forward Analysis technique is that you don't need to know exactly how the market dynamics have changed. The WFA process takes care of determining the best parameter values for your system for each time epoch, including producing the parameters that are working best right now, in todays's market dynamics - the parameters you will want to use for live trading.
You might now ask a valid question:
"So can't I achieve the same by just optimizing using recent data? e.g. from the last 6 months?"
Well of course you could. However, by doing this your trade sample size (the number of trades generated in the back test) will be so small that the results are likely to have little (if any) statistical significance. This means you will almost certainly be over-fitting your system, and producing parameter values that will lose money in live trading.
Walk Forward Anlaysis is such an effective optimization tool because it does both of the following:
It allows parameters to be optimized, based on the market dynamics at different points in time, including the market dynamics right now (enabling parameter values to be determined for live trading)
It maintains statistical significance, since the overall back test is still performed across a long period of time
We highly recommend you watch this brief One minute video on Walk Forward Analysis to understand the process better.
This is why Walk Forward Anlaysis is considered by many professional traders to be the de-facto, gold standard optimization technique.
Subscribe to our newsletter
We are passionate about algorithmic trading, and about helping other algorithmic traders reach their full potential.
We help traders to develop robust trading systems that deliver results in live accounts.