Friends, what a weekend!
I must say it is with quite an emotion I'm writing today.
First: Dale! It might sound "easy" what I'm going to tell, because I'm not one of your customers, so I have no direct financial interest in the matter. I hope things will get fixed soon also in this direction. For what it matters, anyway, your contribution to this and other threads has been for me one of the most relevant learning experiences in the field of trading that I ever did. Part of this experience was also constituted by your mistakes, your "experimental nature" and, last but NOT least, your tenacity in not giving up, going always on, improving and fighting. Thank you a lot for that! And, whatever happened, I think this is of so much value, for all of us (also as an example to follow), that, in the middle term, will probably make all of us quite happy and wealthy traders. So I hope with all my heart that nobody will kick you off here, rather encourages you to go on, as always. I sincerely hope to be able to give somehow a contribution, with all the excellent "brains" of this thread, to make not just my life, but also the one of the brave followers of this "adventure" easier and better.
Back to the task. I must really thank every single person "picking up" my trial on the TBPS! A lot of great inputs.
Rekaats & J: great! Yes, I cross-checked a
part (LOL! It'll take A LOT of time for checking ALL) of your data with mine. They are not perfectly identical for the following reasons:
1. I'm backtesting on the FXDD platform. Although it should "close the bars" at the same time as Delta, probably there are some minutes difference, so sometimes the "close" does not perfectly correspond. Oddly enough some bars are also missing (01.01.2009, e.g.)
2. With the backtester, there, I get values in $$ and not in pips. Although I can set (in my script) the lot amount to be fixed (like = 1), this corresponds to pips just for ???/USD pairs. For GBPJPY I just manually calculated the data of the last month in pips, and, besides the missing days in FXDD, the results are identical to J's worksheet (which is identical to Rekaat's), but...
3. For coding reasons, making metatrader placing the order "at close" as per the old man, is quite painful, so I place orders "at open". Normally in the forex this means only a few (less than 10) pips difference, except over weekend. So my results are not 100% identical to yours also because of these few pips difference.
Anyway: for me all these backtesting's have the role of
giving me an indication about the profitability of a system over long times. I am glad of the "convergence" of our results, because this means we are speaking of the same thing.
Then: on GBPJPY, the system as it is (Rekaatz: I already said I had NO FILTER at all: just like the book worksheet), looks already pretty good... unless some TERRIBLE drawdown.
Thank you very much for the SAR and the ADXR suggestion. I think I
must now fix this terrible metatrader indicator. I found not less than 3 algorithms around, all quite far from Wilder's. I have already an idea on how to make it, but coding the indicators for metatrader is anyway quite painful (that's for you, jcc1761).
By the way, as I did not have time to go on with math. SO: the Wilder's smoothing. We already agreed here that the ADX smoothing can be accomplished by taking the exponential moving average with ((2*ADX periods)-1) periods. Let's assume a 14 days ADX. This can then be smoothened by using an EMA(27).
This is due to the fact that ADX(n) = 13/14*ADX(n-1)+1/14*DX(n);
Well: for the calculation of TR and DM's, the indicated procedure is quite different, because you take 13/14 of the previous "indicator" and add the
ENTIRE value of today: TR14(today) = 13/14 TR14(yesterday) + TR1 (today).
Question for the "matthy'es": is
also this second smoothing resolved by a EMA((2*periods)-1)?
Further subject: filters. Besides the ones suggested by Rekaats I want also to thank Dale for his tips. It boils down, again, to ADX (so, to my future work). Just a question: when you say that you must just take the trades in the "direction of the ADX", do you mean the direction indicated by the +DI/-DI? Because the "direction of the ADX" should always be upwards, isn't it? And again: here we are speaking about ADX. What about ADX
R (LOL, my old friend: here we are always fighting, you and me, with or without this damned R

). Rekaatz was suggesting looking ad the "R": do you mean trading the TBPS only when ADXR is lower than 25?
Finally, dear jcc, I don't remember to have seen your name before in the long story here! Welcome!
You know: I'm NOT an MT4 "guru", nor a professional programmer. Actually I have a major in applied
genetics!!! But I had the chance of working some years in a bioinformatics company, and I had to "absorbe" a bit of bits. Metatrader is for sure NOT the best platform: a lot of mistakes (also in the language: poorly object oriented!). But it is quite nice for "poor guys" that do not have the superb skills of somebody like J! Or the luck of so much time and perseverance of heroes like Dale able to backtesting visually (which is for sure better) for weeks.
How to learn it? I cannot help very much. If you have some experience in programming and especially a bit of knowledge of c (++, officially, but then is quite "poorly implemented) it will take very few time for being able to write your first scripts just by looking at the online "help" pages (the link is present in Metatrader help menu). For me this was the way: I'm far from being a "complete" MQL programmer, but, with the reference guide in front of me, I can manage to make almost all that I want. If you do not have ANY programming experience, well, then always on the same site you find a "BOOK". Please: follow it patiently and entirely, making all the example programs, otherwise you'll spend your hours for finding the period missing in your code (by the way: there is NO DEBUGGER

). Consider that it's difficult to become quite "fast" in the thing, if you have no previous programming & c experience, in less than two months. But patience then pays back! Feel free to ask more, if you want.
Sorry for the lenght! Good night (especially if you got asleep at the end of this novel)!!!!
Fabio