Low win-rate strategies unsuitable for Amibroker's Monte Carlo backtests?

I read up on Amibroker's monte carlo implementation. It seems like a wonderful feature but I'm not sure if it's applicable to strategies with low win-rate like trend following.

I chose "Simulate using trade list". Based on my understanding, the new results are created by reshuffling the trades in the original backtests randomly. Is this correct?

When I look at the Monte Carlo results, a surprisingly high percentage of the outcome ended in financial ruin. This is quite unbelievable based on backtest results and actually using the strategy in real-life, although I don't follow the strategy exactly as I exercised some discretion in the trades.

The strategy I use has low win-rate but fairly good payoff ratio, so the strategy has an edge. I wonder if it's because the low win-rate and the way bootstrapping is done in Monte Carlo that is causing financial ruin to be over-estimated in the Monte Carlo results.

Anyone shares my observation?

1 Like

The math behind Monte Carlo simulation is simple: simulate ALL possible outcomes. With low win-rate, majority of ALL possible outcomes may result in loses. That is pretty obvious consequence of how statistics works. In real life you can be lucky or not. Single life experience is not statistics.


@Tomasz, thanks for the reply. As a side remark or compliment, I am amazed at how fast monte carlo simulation is done in Amibroker. The time taken for backtest to complete with Monte Carlo disabled is about the same as when Monte Carlo is enabled. I thought I can speed up backtest by disabling Monte Carlo but the difference was not noticeable.

This topic was automatically closed 100 days after the last reply. New replies are no longer allowed.