Once again, it's about the eternal: trend/flat. - page 18
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Then to say that only the TF has contributed to the improvement is no longer possible.
Then we cannot say that only the TF has contributed to the improvement.
Did I say that the TF contributed to the improvement? - I said that time filters can be applied to TF below H1, but not to higher TFs, which is why we need ways to detect TFs for higher TFs.
Here are the results for a flat system on M5 from 2:00-8:00 for 2014 to October 2016. Optimization 2014-2015, profitability 2.44, recovery factor 9.71, number of trades 2040.
And these are results for the same system but without time limit during the day (trading is allowed all day), profitability 1.28, recovery factor 1.92, number of trades 2240.
The result, as they say, is in your face. I think the number of deals is quite representative to think the results are statistically reliable.
Here are the results for a flat system on M5 from 2:00-8:00 for 2014 to October 2016. Optimisation 2014-2015, profitability 2.44, recovery factor 9.71, number of trades 2040.
And these are the results for the same system but without time limit during the day (trading is allowed all day), profitability 1.28, recovery factor 1.92, number of trades 2240.
The result, as they say, is in your face. I think the number of deals is quite representative to think the results are statistically reliable.
I do not understand. Why didn't the number of trades grow in proportion to the significantly increased added trading time?
Heh... :) I was just sitting and waiting to see who would be the most observant.
The reason is that at night (1/3 of the day) there are a lot more signals for a flat system than for the remaining 2/3 of the day, a lot more. It's obvious. This is a direct demonstration of the simplest time-based filter, and t/f detection is the same filter, but now it allows to go up to higher TFs, to get out beyond the boundaries of the day.
ZS. Hi Azulenko. Keep thinking that the analysis is crap, I will get more money).
ZZZY. Would love to see Swinosaurs here.... And to Matemat, Metadriver, Avals, hello (this hello is real, unlike the one above a line). What a discussion we used to have in far 2007-2008, it makes me feel nostalgic... All that was said then is still relevant today.
Heh... :) I was just sitting and waiting to see who would be the most observant.
The reason is that at night (1/3 of the day) there are a lot more signals for a flat system than for the remaining 2/3 of the day, a lot more. It's obvious. This is a direct demonstration of the simplest time filter, and t/f detection is the same filter, but now allowing to go up to higher TFs as well, to get out beyond the boundaries of the day.
ZS. Hi Azulenko. Keep thinking that the analysis is crap, I will get more money).
ZZZY. Would love to see Swinosaurs here.... And to Matemat, Metadriver, Avals, hello (this hello is real, unlike the one above a line). What a discussion we used to have in far 2007-2008, it makes me feel nostalgic... All that was said then is still relevant today.
1.
2. It is not clear whether the guys gave up or disguised themselves. I remember. There were identities, for sure.
Here are the results for a flat system on M5 from 2:00-8:00 for 2014 to October 2016. Optimisation 2014-2015, profitability 2.44, recovery factor 9.71, number of trades 2040.
And these are the results for the same system but without time limit during the day (trading is allowed all day), profitability 1.28, recovery factor 1.92, number of trades 2240.
The result, as they say, is in your face. I think the number of deals is representative enough to think the results are statistically correct.
The time restriction does not tell us anything, if we do not know the difference between the code for day and night. The point of my example is precisely that the code there and there is the same and consists of a single oscillator wpr, only for day and night it optimizes separately Tf and PERIOD. Even without performance comparison (that's another topic) we can see that with the same code the optimization doesn't tend to make different TFs, but tends to separate oscillator periods. In my opinion, this can be explained not by flatness-trends, but by a GREATER factor that distinguishes day and night -volatility.
In other words, in my opinion, it is not so much important for the system to predict the price direction, as it is for it to be able to determine the swing. It is usually smaller at night, so the system tends to make fewer reversals.
If there were an example of code for "trending" situations, I could check this thesis more concretely. For example, we could distinguish four states - trend-low volatility, trend-high volatility, flat-low volatility, flat-high volatility.
And see which option works best.
1. The time restriction does not say anything, if we do not know the difference between the code for day and night. The point of my example is exactly, that the code there and there is the same and consists of one single oscillator wpr, only it optimizes for day and night separately Tf and PERIOD. Even without performance comparison (that's another topic) we can see that with the same code the optimization doesn't tend to make different TFs, but tends to separate oscillator periods. In my opinion, this can be explained not by flatness-trends, but by a GREATER factor that distinguishes day and night -volatility.
In other words, in my opinion, it is not so much important for the system to predict the price direction, as it is for it to be able to determine the swing. At night it is usually smaller, so the system tends to make fewer turns.
2. If there were an example of code for "trend" situations, I could check this thesis more concretely. For example, I could distinguish four states - trend-low volatility, trend-high volatility, flat-low volatility, flat-high volatility.
And see which variant works better.
1. You are talking about something of your own, clearly irrelevant to the topic of t/f.
2. Example code? You must be joking... You can take my words at face value or try to comprehend them yourself. But in this case I am not obliged to prove anything and moreover to show you the code. If it is interesting to be convinced of something, please do it yourself.
1. You are talking about something of your own, clearly irrelevant to the subject of t/f.
2. Example code? You must be joking... You can take my words at face value or try to comprehend them yourself. But in this case I am not obliged to prove anything and moreover to show you the code. If you want to know something, please do it yourself.
1. I am simply trying to explain a simple idea: trend and flat is another name for volatility.
2. Well, what about the code, what is the formalization?
And yes, by the way.
It is absolutely incorrect to compare results for different code byresults of OPTIMIZATION. There is a very simple regularity here - the more code, the more variables there are and the better the result will be since there is more space for adjustment. One should compare results of TEST runs on non-optimized sections. Best of all in wolf-forward mode.
1. I'm just trying to make a simple point - trend and flat is another name for volatility.
2. Well, what about the code, what is the formalisation?
3. And yes, by the way.
It's absolutely incorrect to compare results for different code byresults of OPTIMIZATION. There is a very simple regularity here - the more code, the more variables there are and the better the result will be since there is more space for adjustment. One should compare results of TEST runs on non-optimized sections. Best of all in wolf-forward mode.
1. For God's sake, as they say. If it helps you to improve the results of your system by understanding the market situation as "volatility" - go for it!
2. Formalisation - the ability to describe it numerically and formulaically. This I have given.
3. I showed the results of my same flat system for 2 years, 1 year of which is optimization. It's clearly seen that the system behaves worse on the unfamiliar plot than on the optimization plot, but keeps stable growth. Everything is correct. Take a look above on this page.