Adrian Reid explains that trading systems degrade over time (short-term/high-frequency systems decay faster) and emphasizes the need to monitor performance. He describes monitoring the equity curve and tracking statistics over time — win rate, sizes of wins and losses, and average profit per trade — to detect when metrics move outside historically normal ranges. Reid runs weekly backtests, performs deep dives every couple of months, and uses a system-monitoring spreadsheet that aggregates backtest results and flags warning signs so traders know when to investigate or turn systems off.
Questions this video answers
Weekly.
Every couple of months.
How do I know if my system is degraded and how do I know if it's degraded to the point that I should turn it off? This is really important because unfortunately, and you might not want to hear this, but unfortunately, no system lasts forever.
Edges do decay. Short term systems, very high frequency systems, tend to decay more quickly than longer term systems. But all systems can decay over time. So you have to monitor their performance. And the way I monitor performance is I run the back just on a regular basis and I look at how the equity curve is performing. That's one thing. I also look at the underlying statistics of the system over time. I monitor the win rate, I monitor the size of the wins, I monitor the size of the losses. I monitor the average profit per trade and see how they're trending. If all of the stats are starting to trend down in the wrong direction and they start to move outside of the range of what's historically normal, then you know you've got a problem. So I have tools and templates to monitor this. You basically take the backtest you put into the tool and you can see, all right, is it degraded or is it okay, is the current performance normal or not? And so you want to know basically answer the question, is this performance measure within the range of what's historically normal for this system? If it is, great, okay, the systems are going through a rough patch, but it's normal because performance comes and goes, right? But if the performance statistic is starting to trend outside of the historically normal range is getting worse and worse, then you know you've got a problem and chances are the edges decayed and it's time to really do a deep dive and look into that system or maybe turn it off. And I think a lot of people don't monitor their strategies close enough and so leave them keep on trading beyond when they should go, oh, wow, that's a deep drawdown. Maybe I should turn the system off. Often it's too late, something you should still turn it off at that point. But often they've left it too later than they should have. So monitor your systems closely. I back test all my strategies every week and then I do a deep dive every couple of months to check all of the performance stats and make sure they're all within range. And what I've developed is a tool, the system monitoring spreadsheet, where I have this great process where I run all of my systems one after another. Automatically all of the results get sucked into an analysis spreadsheet. That spreadsheet analyzes the equity curve and the performance stats and tells me which ones are flashing warning signs based on how they're behaving recently compared to historical behavior. This is something all my students get so that they can quickly assess should I be worried about this system or not? And you need a process like that because by the time you've got a couple of strategies, you then need to start monitoring them to make sure that they're still performing. This becomes really important to make sure you know when to switch a system off.
