Eyeballing the equity curve of a simulation may be a quick way of paging through a large number of simulations. It shows a lot.
But here comes the rub. Any single equity curve or simulation represents the position of only that combination of the unfolding of the number of wins and losses in a row. I can have a nice algo with a 70% win rate and have the losers grouped in such a way that the draw down is extremely painful.
Looking at what is possible and what is probable and where to draw the line between the two is necessary in order to set expectations of an algo's future performance (Disclaimers apply). Setting a stop loss for an algo is like setting a stop loss for a single trade. Knowing your drop dead stop is important in both. The charts below show a number of equity curve charts of a number of winning trading systems for the same symbol. Which one is "best" is quite a subjective judgement. Having firm requirements in your trading plan helps me make a better choice.
The charts below have a white and a green section. The GREEN is the out of sample performance.
The interesting feature on these equity curves is the drawdown that occurs a little way into the OOS (Out Of Sample). The issue for algos is like anything in trading: discipline. Would you keep on trading an algo if it goes into a drawdown a short time after it goes live? If the answer is "NO" and you had done the proper work to create the algo - backtesting and walking forward - and you don't believe that you could sit through the drawdown then fully auto trading may not be for you. However, Hybrid trading is a different thing. More as this series of posts talking about algos for fully automated trading continues.
The markets were quiet in the London morning. Volume was lower too. The trades paid, but not a lot. I guess I've been spoilt.
No comments:
Post a Comment