Weekly update on automated systems
My sixth red week. A lot of reasoning has to be done, but it will be simple. The question is: are the bad results being produced by the systems I added a couple of months ago or are the bad results coming from all systems in general?
Because in the first case, I am all right, but in the second case I am screwed and my systems are
not looking good.
[...checking... studying...]
Ok, studied and checked quite a bit and here's the deal. It wasn't just the recently added systems that caused the massive losses in the past few weeks. It was the Intraday systems that have lost because maybe the markets didn't go up and such systems mostly only go LONG, and of course if the markets mostly fall, it's not going to be easy for them to catch the LONG opportunity.
If I were to differentiate between each individual system, I might get rid of those that don't work, but I might be overoptimising. Instead, by looking at categories of systems, by common concepts, I should manage to avoid overoptimization (I mean: one system could have gotten lucky, but all of them put together is more unlikely).
So, as a category, the weekly bias systems seem to have failed so far. But they only go LONG and since in the past few weeks we've fallen, then that might be the only reason. These systems may start performing very well in the future, but am I prepared to take such a drawdown, whereby I go to 50k and then come back down to zero pretty much? Mind you, getting to zero doesn't mean my equity is wiped out, because the chart does not represent my capital but only the profits of the systems.
Let's go over the categories of these groups of systems just in case anyone cares:
1st hour bounce: a lot of screwing around in the first hour (trades 15.15 to 16 CET), but there's a tendency to bounce within the daily high and low. Well, I call it "first hour" because it's the first hour of "open outcry" trading, but of course the futures trade 24 hours a day.
ON bounce: after the close (22 CET), tendency to go in the opposite direction of where we've gone during the day.
opening gap: no explanation needed.
overstretched: self-explanatory title.
volatility breakout: no explanation needed.
weekday bias: on some days of the week markets go up.
with id trend: during the day (from 10 to 22 CET), whatever the trend, it tends to continue.
There's groups of systems that produce consistent results and others that have bad periods. I need to identify what doesn't work with the latter group. The weekly profit by groups table identifies the WeekDay Bias and the WITH ID trend systems as the unstable ones.
I wish I knew what went on in the markets in the past 6 weeks that might have interfered with my systems. But I am too busy posting guitar videos to be able to do any testing on this. The truth is that I am tired of testing, backtesting, automating systems.
When I sum up all the losses by those two groups in those 6 weeks, I get a total of 40 thousand dollars. If I had a way to avoid those 40k of losses, my systems would be ok. As I said, since those weeks are all packed together... I mean this is not bad at all: even considering the 42 systems all together, the red weeks all come in a straight series of losses. I could simply stop trading whenever I get a red week. How about it?
But this will become an issue only when I'll be trading all the systems. Right now I am only trading the best system of all and so it is not an issue.
One day I should start trading all of the systems at once, and that day, I should use this method of not trading after a red week.
One may wonder at this point, after how I mentioned posting guitar videos instead of working hard on the systems, one may wonder why I am not busy working and am here relaxed, wasting time. I am not wasting time. The quantity is all there, and it would be negative to focus on quantity any further on quantity, because I'd be missing out on quality and the much bigger benefits that I can get from it. I have built 42 systems, and there's enough good stuff in there to keep going for years. Now I need to focus on making a few good moves in the right direction and I am all set. Both for automated and discretionary, where making one good trade (sniper approach) per day is better than 10 pretty good trades. The real objective now, after all this hard work, is to slow down (in discretionary, simply to trade less) and focus on making a few good and simple moves, like these ones I've been talking about, such as not trading for the week following a red week.
Actually that's why I started this journal. Because I realized that the quantity was all there, and that I needed to slow down, sit back, and look at things from a different perspective, get a different approach, and add whatever quality was missing. Now I'll go and implement the weekly filter, whereby the systems don't trade at all after an overall bad week. Can't hurt me much. At the most I'll miss a good week. But since I haven't been getting one good and one bad week but a bunch of red weeks all stuck together, right now this seems the best solution.
Well well... check this out... was I saying six bad weeks? Look at this weekly chart of the EUR, which pretty much represents all my systems, since 7 systems are on the EUR alone, and most other futures I trade move the same way the EUR does:
Ok... now I've added the red and green weeks of my system to the chart and the picture is pretty clear and self-explanatory.
Now I see that maybe instead of waiting for the red week to happen I could prevent my systems from trading when the EUR is not going up. Well, a lot of thinking needs to be done outside of this post. Probably I will wake up one morning with the idea in my mind, or while in the bath tub. What I need to do is remove the worries I am having about my job, because right now what I am dreaming of is the suspicious transaction reports, and that mother ****er, the boss.
Link from where I took the weekly chart.
Ok, this is it, but I don't know yet if that's it. I set up a chart with the 10 period moving average ona weekly timeframe, and that seems to get rid of part of the red weeks:
http://futuresource.quote.com/chart...a=W&z=800x550&d=LOW&b=CANDLE&st=MA(10,10,10);
But you know what? The best method still seems to be not to worry about what the EUR does, and just to stop trading the systems whenever the previous week was a red. I will implement this at once.