PDA

View Full Version : Do trading systems stop working?



rnufnwft38
07-07-2015,
Hi there,

I have a trading system with clear entry/exit/stop/limit rules. I don't take trades very often, with setups occurring on average around 2-3 times a week.

I trade one index only.

I've forward tested it for around 17 months now and have around a 85% success rate with a 1:1 risk/reward ratio.

Having researched this topic, I can't find a clear answer, it seems like systems need to be tweaked over time, but has anyone experienced a system that has completely stopped working having had consistent results for over a year or so?

thanks

Jim

Rudolphdal
07-07-2015,
Yes ofcourse , your win rate can go down significantly , maybe it wasn't working in the first place , it just happens that your system is going through a good batch . 17 months isnt much , i've seen trading systems collapse after 5 years of trading .

fqphxycc14
07-08-2015,
A nice system will be responsive or filtered according to context. e.g. fewer or no signals generated as wider market direction becomes unclear or is reversing. Or you manually filter signals so that e.g. you ignore buy signals as the wider market is starting to tank. There are always clues to be gathered from correlating charts.

irogonosulugy
07-09-2015,
Quote:
Originally Posted by miles123 View Post
Hi there,

I have a trading system with clear entry/exit/stop/limit rules. I don't take trades very often, with setups occurring on average around 2-3 times a week.

I trade one index only.

I've forward tested it for around 17 months now and have around a 85% success rate with a 1:1 risk/reward ratio.

Having researched this topic, I can't find a clear answer, it seems like systems need to be tweaked over time, but has anyone experienced a system that has completely stopped working having had consistent results for over a year or so?

thanks

Jim
It doesn't necessarily stop working, it depends on how much effort was done to refine the entry/exit signals which could lead to curve fitting. Then there is whether optimisation was used in any of the parameters and whether the periods reflect all market phases.
Systems shouldn't need to be teaked. if they do, then they are more reliant on market cycles which inevitably change and are therefore less reliable

if your system is simple, and can "adapt" to trend state then I'd say its more robust and could be left alone

eyfmeofh22
07-10-2015,
All system stop working when market conditions change so that they cannot longer deal with them.

"I''ve forward tested it for around 17 months now and have around a 85% success rate with a 1:1 risk/reward ratio. "

These number mean nothing if the system is curve-fitted because it will continue working well until the market changes behavior. How significant was your system in forward testing? This blog offer a good methodology for testing for significance of a golden cross system but the principles apply to all systems. You can try to estimate the system significance in a similar way.