Knowing if your system is performing as expected is one of the most crucial question for a systematic trader. This question is mostly raised in times when you “feel” that your returns aren’t in line with what you had expected. However, I strongly think that a reality check should be part of a periodic system check-up.  How or how often to evaluate your systems performance depends on the type of system and how frequently it trades, e.g. a system that trades infrequently (a few times a year only) might need a longer look back period than a system that trades a couple of times a day. Analyzing your systems past performance needs to be a simple and effortless process. So it has to be automated and realistic. If one needs to spend a lot of time preparing the data, he or she is less likely to do it on a recurring basis. That’s just my opinion. So for this post I want to focus on how I analyze my systems past performance.

I decided to track and compare my trades as they occur on a daily basis (read here for more). In case I make an error in the execution or something else it gets noticed and analyzed at a later point.  This part tells me how well I execute vs. my back-tested results. Of course the goal is to get as close to 100% match as possible.  Analyzing the outcome of backtest vs. actual trades isn’t exactly the objective of this post.

Equity curve or trades

In the beginning your need to take a decision if you want to compare your systems performance with the actual trades or the resulting equity curve. I decided to run all my statistics on the daily equity curve.  This way I can have one unified approach for all of my systems. Furthermore I get more data points than just taking the individual trades, especially for my longer-term systems trading on a weekly time frame.


What to measure?

I got a number of key performance indicators. These indicators are calculated on a rolling basis (daily):

  • Max draw down vs. current draw down
  • SPY correlation: 60 day correlation between equity curve and SPY
  • Daily Directional Correctness: number of up-days (ROC1>0)
  • Volatility: 60 day standard deviation of one day returns (ROC1)
  • Compound returns (60 days)
  • Sortino Ratio (252 days)

To gauge if a system is within it’s upper and lower borders I compare these rolling values with the median of the last 1000 and 252 days.


Implementation

Practically I create and store the equity curve after each backtest run. AmiBroker treats a stored equity curve like any other symbol. So based on this “new symbol” I created a template with a number of custom indicators.  As said in the beginning, I find it very important to have a effortless procedure in place, even though it might not be be the most advanced one.

Click on the picture to enlarge it. In the top of the of the picture you see the actual equity curve incl. draw downs and the max. draw down that occoured. Underneath the equity curve you got a number of indicators with a black, red and blue line. The back line represent most recent calculations as defined above. The red and blue line are the median of the last 252/1000 days.


What to do after a draw-down?

The most difficult trading time occur during a significant draw-down. You take severe losses one day after the other, feeling an ever-increasing amount of pain. You might begin to question the system. Asking yourself if the market has changed. That’s the point were emotionally weak traders begin to not follow their plan (system) anymore. Unfortunately that’s exactly the wrong thing (most of the times). A trading system is likely to have it’s best time right after a severe draw-down. The most profitable trades occur when your pain hits a new 52 week high and your equity a 52 week low! That’s why I have the believe that trading is counter-intuitive to the human brain and emotions.

I run a few simple systems to trade the equity curve for some of my systems. Bottom line: when any kind of RSI is bellow 50 you have a higher next day return than otherwise. The lower you put the RSI trash hold the higher the next day return will be. Bottom line: continue to execute your trades as you system suggests unless you have a valid reason that the market has fundamentally changed.

Share you experience how you monitor your systems health and how you deal with draw-downs. Please enter your comments bellow.

– Frank