Author's dialogue. Alexander Smirnov. - page 25

 
ANG3110:
VBAG:
I'm sorry to interject myself into your conversation. How about tying the value of the threshold to some functions. For example ATR or to the slope angle of T3. So to speak, let's make a lopsided trigger in direction of the trend. I am currently working out how to run EMA back and forth. As soon as I make a kite I will try this idea myself.
It's quite a sensible idea, maybe something will work out. The only difficult moment may be that in flat the threshold should be abrupt and in most cases it is sharp and breakthrough and the reaction may be delayed. Rather, it could be a deflection type of both Std and max. deflection set.
Or with loaded threshold, but in order not to pass the breakdown drag in a tight pending stop. In general, who will open ahead. Yeah... I guess so. But that's a completely different kitchen.
 
ANG3110:
VBAG:

Hello!



ANG3110 You mentioned T3. Could you please elaborate on its advantages( what is the gimmick) and preferably in comparison to the others.

Thanks.






It's very simple. We make an Expert Advisor that would buy and sell when the muving direction changes to the opposite. We can add a small threshold value, 1-3 points to the moving averages so that they do not wobble within them, otherwise there will be a lot of false positives. We also add all sorts of indicators (MA, EMA, LWMA, JMA, Linear Regression-LR, weighted regression, parabolic regression, DCT, regression sines, cosines, logarithms, adaptive family, VIDYA, AMA, by Std, by momentum or angle LR, Highest-Lowest, Parabolic, Lag indicators, Step indicators, cluster Neural network, etc.п., and T3. And run it through the optimizer. In this experiment, T3 gives the best results. Then comes LWMA, and then EMA. The rest are noticeably worse. This experiment is not very correct, like fitting, but it will immediately show, for instance, that Jurik is a cheat, an exotic. And it will show that determining the state of the market at any given time is far more valuable than all that filtering and ironing.



T3 is essentially EMA taken from itself six times in a row and summed up according to a certain law with balanced lag factors. That's why it has such a high smoothness.

ANG3110! I have thoroughly compared the T3 algorithm to other algorithms of averaging in Expert Advisors and I can assert the following: the only algorithm that is far more profitable in the optimizer is JMA, while all other algorithms have no advantage over each other! Of course I'm referring to the JMA algorithm in my article and not the code base. I'll just tactfully and politely refrain from commenting on JMA from the code base. I can only add that my EA writing speed is several orders of magnitude faster than that of all participants of this thread taken together! I write any EA in such a way that I can
select any smoothing algorithm I want from among those available to me by means of just one function, and simply change the number by which this algorithm is represented in this function! Another thing is that for any trading instrument
there may always be a preferable averaging algorithm for any test period!

 
GODZILLA:
The only algorithm that significantly exceeds all other algorithms in optimizer profitability is JMA, and all other algorithms do not have any advantages over each other! Of course I'm referring to the JMA algorithm in my article and not the code base. I'll just tactfully and politely refrain from commenting on JMA from the code base. I can only add that my EA writing speed is several orders of magnitude faster than that of all participants of this thread taken together! I write any EA in such a way that I can
select any smoothing algorithm I want from among those available to me by means of just one function, and simply change the number under which this algorithm is represented in this function! Another thing is that for any trading tool
there may always be a preferable averaging algorithm for any test period!


Well what I wrote is not made up. I checked it last week ago in the interval between 01.10.2007 and current time. Perhaps you tested in other conditions than the ones I described. If you are interested, I can post the results or test pundits. I have specifically tested GBPUSD15 on opening prices (for a quick experiment it is good enough), on Alpari quotes.

And in general I'm surprised by the statement about the speed of writing EAs. Do you know the potential of the participants in this thread?

 
shall we compete with or without boxing gloves?)
 

Here I have not been lazy and re-tested T3 and JMA.

JMA put in (from GODZILLA).

The best "fitted" variations.

Depo=10000. GBPUSD15 Alpari at opening prices. From 01.10.2007 to 07.02.2008

Passes are optimized by Risk as a percentage of a deposit.

1-Risk=0 (only at 1st lot); 2-Risk=10; 3-Risk=20; 4-Risk=30; 5-Risk=40; 6-Risk=50; MaxLots=15;

JMA

Passage Profit Total trades Profitability Expected payoff Drawdown $ Profit %
3 14955.20 116 1.12 128.92 30725.20 72.39
1 12932.00 116 1.42 111.48 6464.60 28.40
2 9379.50 116 1.27 80.86 7415.40 30.99
4 5461.60 116 1.03 47.08 48615.00 88.46
6 -375.40 36 1.00 -10.43 61522.10 86.47
5 -2821.50 116 0.99 -24.32 69350.20 96.48

Can't say it's too bad.

T3

Profit Total trades Profitability Expected payoff Drawdown $ Profit %
6 174019.60 166 1.29 1048.31 139782.00 62.20
5 162630.30 166 1.27 979.70 139782.00 65.52
4 136153.20 166 1.23 820.20 139782.00 74.81
1 13485.60 166 1.33 81.24 9318.80 35.57
2 7620.00 166 1.16 45.90 13788.70 55.59
3 7438.20 166 1.03 44.81 59415.60 87.17

But T3 holds Risk up to 50% and even higher and JMA only up to 30.

I have checked before on other historical data and currency pairs and the picture in JMA/T3 ratio, was about the same, even slightly better towards T3.

 

to ANG3110

Question - on what period was the T3/JMA tested ?

 
VBAG:
Сейчас как раз разбираюсь с прогоном EMA туда-сюда.
Чо Вы под этим подразумеваете?
 
I don't understand how you can talk about the trading parameters of the indicators outside the context of the EA that uses them. ANG3110, please clarify which strategy. And second: what is Risk, as a formula (everyone understands it differently)?
 
ANG3110:

Here I have not been lazy and re-tested T3 and JMA.



JMA put in (from GODZILLA).



The best "fitted" variations.



Depo=10000. GBPUSD15 Alpari at opening prices. From 01.10.2007 to 07.02.2008



Passes are optimized by Risk as a percentage of a deposit.



1-Risk=0 (only at 1st lot); 2-Risk=10; 3-Risk=20; 4-Risk=30; 5-Risk=40; 6-Risk=50; MaxLots=15;



JMA




PassageProfitTotal tradesProfitabilityExpected payoffDrawdown $Profit %
314955.201161.12128.9230725.2072.39
112932.001161.42111.486464.6028.40
29379.501161.2780.867415.4030.99
45461.601161.0347.0848615.0088.46
6-375.40361.00-10.4361522.1086.47
5-2821.501160.99-24.3269350.2096.48



Can't say it's too bad.



T3




ProfitTotal tradesProfitabilityExpected payoffDrawdown $Profit %
6174019.601661.291048.31139782.0062.20
5162630.301661.27979.70139782.0065.52
4136153.201661.23820.20139782.0074.81
113485.601661.3381.249318.8035.57
27620.001661.1645.9013788.7055.59
37438.201661.0344.8159415.6087.17



But T3 holds Risk up to 50% and even higher and JMA only up to 30.



I have checked before on other historical data and currency pairs and the picture in JMA/T3 ratio, was about the same, even slightly better towards T3.

I don't take time formats less than 1 hour seriously, while I use all available history for running Expert Advisors on all imaginable currency pairs. My EA's code contains a possibility to use one of the available algorithms of averaging. It is quite trivial to implement given a good understanding of MQL4. I'm just stating the fact that I have empirically seen countless times on a great number of quite different Expert Advisors: The JMA algorithm in the strategy optimizer comes out on top disproportionately more often than all other algorithms taken together! Obviously I have absolutely no need to bother with such nonsense like spending time on proving such things. It's a completely useless task! Actually, the most logical thing to do in situations like this is to always be able to choose from the various possibilities which one is the best in the given case! And at the same time keep in mind that it may very well happen that after a while the choice of options will be completely different. In general, the most ideal algorithm for averaging in Expert Advisors is not the one that shows the maximum profitability in the optimizer, but the one that is just more or less stably profitable beyond the right border of the optimization period. But even with this approach I have had two averaging options at the top, the second of which is JMA with Lengh parameter of 2,3,5,7,9. But, of course, I'm not going to argue or prove anything, I just think that I shouldn't get hung up on any single moving average. I can only add that the classical variant of the unpretentious MTS using changes of the trend moving average direction as signals for entering the market at the period of the chart less than four hours appears to be a failure beyond the right border of the optimization period, at any moving average and at any positive result in the optimizer! It is easier to guess by coffee grounds or stars, or ask "psychics" for advice than to use an EA of this kind to make a decent real profit within a testing period that is commensurate with the Championship period! But once I had a very fragile computer and tried my best to do at least an equivalent replacement of the resource-intensive JMA algorithm by the incommensurably less voracious T3. It was for this reason that I made the T3Series() function! But the number didn't work, and I had to play around on my computer primus to optimize my system four times as long, using JMASeries()! So good luck to all!

 

I wrote the currency pair GBPUSD15 - so the period is naturally M15. If we are talking about period T3, then just use optimizer, I have T3 of my own making and it is slightly different from the one that is available on the Internet. Not only the period is optimized, but also the b-coefficient. Same with JMA, I don't know what variant you might have.