CFDs are complex instruments and come with a high risk of losing money rapidly due to leverage. 64 % of retail investor accounts lose money when trading CFDs with this provider. You should consider whether you understand how CFDs work and whether you can afford to take the high risk of losing your money.

Trading frequency

“Less is more” I agree with this concept.

Less trades or less rules?
I am a mechanical trader, so to reduce the number of trades I have to create rules to exclude false signals.
With a lot of conditions I can reduce 100 trades per year to 10 trades per year.
Are those 10 trades accurate “high probability” trades o I am just creating an overfit that hits few outliers trades that won’t happen anymore in the future?

3 Likes

the point is it works the other direction too. more trades does not equal more confidence. so you are back at square one , none the wiser.

the thing that is being misses here is assumption the “sample population” is the same.

1 Like

Agreed but less trades isn’t equal to more quality.
It could be just luck or “overfiltering”, many rules and constraints and few events.

3 Likes

agreed. I never made the claim it was more quality

1 Like

I will go for fewer rules and sacrifice a few percentages of profit.
If a basic Trading strategy provides 100 trade signals then filtering no more then 20%-40% may help to avoid an overfitted strategy.

1 Like

From my experience, less rules/filters = more robust strategy and I try to keep it that way. Whenever I applied and tested some additional filter to already working strategy Murphys laws immediately started to work against me and got worse results in the end. I implemented over 30 filters to my strategies and in the end sticked with 2-3 what I used since beginning…

1 Like

I have given up on “full mechanical” trading after years of trying to find something.
As a discretionary trader, two rules are enough:

  1. Entry Rule: something must happen before I enter, which I have programmed Alerts for to notify me.
  2. Context Rule: the ‘discretionary’ part where I look at the chart and decide for each Entry Alert whether to act on it or not.

I have found it impossible to parameterize the Context Rule in an Expert Advisor to get net positive results without looking at the charts. Every situation is different and requires a form of assessment that involves too many input parameters, including stuff that’s not on the charts like CPI, PMI, NFP, GDP, Trump Tweets, etc…

4 Likes

More logic and less “trying somenthing that seem to be awesome”

3 Likes

This is text book fitting but to what degree? I think it depends on the strat, filtration technique and sample size and maybe some other things. You can get a hint by doing some walk-forwards incorporating the filtration steps. (Carefully! Don’t mess around trying to get the perfect walk-forward either or you’ll just end up with a over fitted walk-forward.)

It seems like over-fitting by applying a basic non MA based filter (example: filter if yesterday’s close was lower/higher than today’s) is less likely to result in a losing strategy than say over-fitting by tweaking basic parameters of the strat (or applying a MA filter), I’m not sure about this though, what do you think?

FWIW I haven’t found any good filters to apply after inception. The only success I’ve had with filters are when they are in place at inception. I used to think having a large sample of trades first and then applying a filter after would be good but I wasn’t able to find a technique that proved that true. I’m interested in other’s experience and how your studies turn out though…

2 Likes

I beg to differ. Suppose you want to apply a long-term trend filter, then you should use a single filter. This may be a look-back period or MA or any indicator (such as 70 days look-back or 90 days look-back or MA-100 or MA-200) but not a combination of multiple filters. If you will compare the results, then most of the trades will overlap and give you similar results. But using combinations of filters result in over-fitting of the trading strategy.

1 Like

I have to say that my experience has been exactly the same as yours bendex. Before trade inception, using a carefully designed filter to provide a trade open “go/no go” decision helps improve results. However I have found that both the cessation of that filter, or the use of any other filter has never helped to inform a better trade closure decision. Good that we are coming to the same conclusions on that.

1 Like

I agree with both of you. The roughness of the equity curve does not automatically decrease when we change exit strategies, combine different systems on the same markets, or combine different markets on the same system. However, changing entry strategies can change the equity curve roughness significantly. Entry is the most important factor of a Trading System. This conclusion goes a bit against the popular wisdom that diversification gives a smoother equity curve.

Diversification works only after you removed all the lucky and overfitted stuff that worked only on the past, and this can be a looooong process.:slight_smile:
Overfits trade the noise so they are well uncorrelated.

Most of the time Different trading systems on the same instruments capture the same movement, so I don’t think there will be much effect on drawdown of the equity curve.

1 Like

Exact, even if the signal is very different the actionable movements are the same.
This is true also investing, many darwins trading very differently and managed by different persons are green or red in the same days.

1 Like