What is Signal Timing Luck? (Regime Filters)

Hey guys, I just want to have a quick chat about a hidden risk in trading, systematic trading in particular, that we should be aware of and sort of do our best to mitigate. There are always going to be issues with luck involved in trading. You know, maybe a weather event comes through and knocks out the power in your house and that happens to be the best week of the year. There's nothing you can do about that. That sort of thing happens. There's a million things that can happen that could influence your trading that we have no control over. That being said, of course, being information technologyminded traders, there's always redundancies and backups and ways we can mitigate those risks and dilute those risks. One of the ones I want to talk about today, one of these luck risks is signal luck or timing luck. Specifically in relation to regime filters. So the concept of a regime filter, hang on, I just noticed the cats have knocked over all the crap on my desk last night. Um, the problem with regime filters, as this article will explain, and I'll link to this below, is signal luck. So the idea of a regime filter, let me bring up a chart real quick. Let me see if I've got a regime filter script I can add on. The script is using a exponential moving average. It really doesn't matter. The concept here is the main idea, right? Doesn't matter what regime filter you're using. So I think this script is looking for at least two closes below the moving average before it is considered. Yeah. This script down the bottom here, this colored ribbon is detecting more than two closes below this moving average. Obviously, you can get a bit more creative with your regime filters. You could use breadth. You could use volatility. You could use something other than a moving average, Ballinger bands. There's a lot of different ways you can apply regime filters. Some are more effective than others depending on what you're trying to achieve. However, all of them are going to have the issue of signal luck. A great example of this is back here very recently in 2025, mid 2025, April. We had this crash selloff here. Now, technically this month bounced very quickly. So, this is probably a bad example. Let me see if I have a better one. This is probably better. This is much better. This is the COVID crash. So let's say you've got a any kind of system which uses a reg a regime filter and it makes its decisions at the start of every month on the first trading day of every month. So using a 200 period EMA on the S& P 500 as your regime filter would have got you out on this day. Your system would have gone to cash on the first trading day of March which if you look down the bottom here is actually the 2nd of March. The first was a Sunday. Markets were closed. We get two closes below the regime filter right before the start of a new month. System goes to cash. What happens? We ride out this uh crash of about 27 30%. I'm sure you can see where I'm going with this. Now, let's throw on a simple moving average. 200 period simple moving average. In this case, it's not going to make much of a difference-c because these two candles closed below both moving averages. But what if this candle had closed higher above this purple line and we didn't get two closes below that moving average before the 1st of March and then we had to sit through this whole selloff for a whole month because our system didn't get us out. That is signal luck or regime filter luck in this particular case. Obviously this applies to any indicator that you're using for any purpose in your scripts. There is luck involved in when that signal triggers, especially for longerterm systems. For a shorterterm system, obviously it's not going to matter much at all. But for anything that holds for weeks or more on end, this is something to just be aware of. And if your system is very vulnerable to it, if we come up with some regime filter settings which uh let's say we were using a 20we uh moving average instead of the 200 day that would have kept us in the market until this selloff. The key takeaway from this video is there's a couple of articles below that I want you to read. Um these guys are way smarter than I am and they've done a better job of explaining this stuff than I can. But the key takeaway is if your system if you run a back test and you tweak your regime filter and you find that even minor tweaks really nuke the profitability or more likely

significantly increase the draw down the max draw down. One thing you can do to help with that is to split up your system into multiple time frames or have two different regime filters on a 50/50 split of the system. So this article by Nick Raj explains some of his research he did with this with one of his systems. So this is one of Nick's systems in their the Chartis platform, the signal service. Um unless you're a member of the Chartis platform, you won't be able to read all of this or see this system. There'll be a link to this below as well if you're interested in seeing Nick's personal trading systems. There's a 10% off annual link. It's an affiliate link that he gave me. Gives you 10% off. And as he says here, originally the strategy traded just once per month. So it rebalanced once per month. However, ongoing research suggested that signal timing luck could play an important role in performance. And then he links to this article. He says here, the portfolio we now offer has been divided into two. 50% has been allocated to the original monthly signals and weekly signals. The added weekly rotation now diversifies returns and reduces signal timing luck. Yes, it increases workload slightly, but it reduces the issue of what I just said. If you have two different regime filters and one triggers earlier, one might trigger up here, one might trigger down here. That way, if we scroll down here, here's his example of his two systems. He has the monthly rebalance, which is the orange line, and the weekly rebalance. As he says here, the returns drop. The max draw down is more or less the same. However, the smoothness or volatility of the equity curve growth is superior on the weekly version. So, by splitting the capital up 50/50, you smooth out your equity curve. You go through less of a wild ride. This particular system he's showing here trades the NASDAQ. So, that's pretty volatile at times. So, this is a desirable outcome. So, the 50/50 split, slightly lower returns, but a much smoother equity curve than either of the uh systems. You'll see the swings of the system are just a bit more muted on the combined system. Anyway, just some food for thought. You don't have to do this. It depends on the system, but if you find that your regime filters are keeping you in volatile markets longer than you'd like, there's not much we can really do about that. Signal luck is just inherent in a long-term systems trading decisions. And this technique of splitting up your capital 50/50 over two different time frames or using two different regime filters for half of the funds or capital in the system does help diversify that signal luck and rebalancing timing. And that's just something to think about. As I said, links to these articles will be below. Definitely worth your time to read them.