Dear Line,
thank you so much for so an inspiring and deep thread. I'm so sorry to be late here. My time available for trading and studying is so reduced that I really didn't manage to digest a bit of your teachings before.
And very openly: I think I have digested just a small fraction of the whole. The big picture, anyway, is pretty clear and attractive for me, now.
What I am commenting now is just the still smaller fraction which was clear enough for me to have it coded and backtested. I have made a little script which implements the T-trigger at GMT-1 00:00 and 12:00 and then goes on swapping at close the way you have shown in the most recent series of examples. Profits are taken by initiating a quite tight trailing as soon as (but not before) price has reached 50 pips (or more) from entry point. The whole using the M15 time frame. Note that, for this testing purpose, no particular money management nor compounding is applied. Trades are initiated with 0.1 lots, period.
Well: I made an interesting observation.
I started backtesting data from Oct 7th to date, i.e. one month in the past. This is the picture:
http://www.forexfactory.com/attachme...1&d=1320510922
Well: this looks actually very very promising for a "fast-and-dirty" testing script, actually, and the system, when visually backtested seems waterproof in its simplicity.
Then, unfortunately, the hopes get a bit cooled down if you try to repeat the experiment let's say from Jan to Jun this year, just to chose a longer period a bit back in the past:
http://www.forexfactory.com/attachme...1&d=1320511167
Mmm... Many possible explanations:
1. Me dumb , I made some stupid mistake. Very possible. I share this (otherwise useless) script for you to test [take care: timing is hard coded, and adjusted to my FXDD testing broker GMT+2, so you must adjust it to your broker's Time Zone. Code is super-easy, commented and should be self-explanatory. If not just ask!]
2. Among the weakest features of metatrader, there is surely the management of historic database. So: it might well be that these dismal results are due to wrong historic data. It doesn't look like, for what you can control in a rapid double-check (history center, visual inspection, quality of modelled data in backtesting).
3. Otherwise: this is one result of a syndrome I am seriously affected by (or at least... was!), which is called visual overfitting. While developing systems, making a manual backtest of several thousand bars (like the M15 of Oct-Nov) is a huge, serious worthful job. Unfortunately it happens (at least to me) that, looking and re-looking back to (so many) bars we find sometimes repetitive patterns. As we have looked so many bars, then we get convinced that these patterns are "common" and "general" and "valid" for the market as a whole. This is where computers come into help, because they can test it not for thousands, but for 10.000 - 100.000 - 1 mio bars, and... eventually find out that these wonderful patterns are not so "general" and "carved in stone" on the market in its complex.
So, dear Line, I had the impression that, after having claimed in the initial post that you had found "the general rule" (let's not call it the Holy Grail), a fact that I still can believe in the general terms of "trading the line", "keeping the hand in the fire box", i.e. accepting a very high R/R in spite of a low "accuracy" (%win/%loss)...
... in spite of all these good ideas, I had the impression you were/are still "tweaking" quite a lot your system, and this T-trigger line could be no "general solution".
I would like to ask you, if you still have the patience and the goodwill of sharing a bit of your vast experience, how long are you already succesfully trading the line, the way you have shown here.
And, since "where to place the line" is what you wish we learn to do, if possible I will be so thankful if you can give some more insight into how to design the V-trigger, and how long you've already used it successfully.
I hope you'll not take my experiment as a criticism, but, as it is, as a sign of deep appreciation and interest in your exceptional skills.
Bye
BL
thank you so much for so an inspiring and deep thread. I'm so sorry to be late here. My time available for trading and studying is so reduced that I really didn't manage to digest a bit of your teachings before.
And very openly: I think I have digested just a small fraction of the whole. The big picture, anyway, is pretty clear and attractive for me, now.
What I am commenting now is just the still smaller fraction which was clear enough for me to have it coded and backtested. I have made a little script which implements the T-trigger at GMT-1 00:00 and 12:00 and then goes on swapping at close the way you have shown in the most recent series of examples. Profits are taken by initiating a quite tight trailing as soon as (but not before) price has reached 50 pips (or more) from entry point. The whole using the M15 time frame. Note that, for this testing purpose, no particular money management nor compounding is applied. Trades are initiated with 0.1 lots, period.
Well: I made an interesting observation.
I started backtesting data from Oct 7th to date, i.e. one month in the past. This is the picture:
http://www.forexfactory.com/attachme...1&d=1320510922
Well: this looks actually very very promising for a "fast-and-dirty" testing script, actually, and the system, when visually backtested seems waterproof in its simplicity.
Then, unfortunately, the hopes get a bit cooled down if you try to repeat the experiment let's say from Jan to Jun this year, just to chose a longer period a bit back in the past:
http://www.forexfactory.com/attachme...1&d=1320511167
Mmm... Many possible explanations:
1. Me dumb , I made some stupid mistake. Very possible. I share this (otherwise useless) script for you to test [take care: timing is hard coded, and adjusted to my FXDD testing broker GMT+2, so you must adjust it to your broker's Time Zone. Code is super-easy, commented and should be self-explanatory. If not just ask!]
2. Among the weakest features of metatrader, there is surely the management of historic database. So: it might well be that these dismal results are due to wrong historic data. It doesn't look like, for what you can control in a rapid double-check (history center, visual inspection, quality of modelled data in backtesting).
3. Otherwise: this is one result of a syndrome I am seriously affected by (or at least... was!), which is called visual overfitting. While developing systems, making a manual backtest of several thousand bars (like the M15 of Oct-Nov) is a huge, serious worthful job. Unfortunately it happens (at least to me) that, looking and re-looking back to (so many) bars we find sometimes repetitive patterns. As we have looked so many bars, then we get convinced that these patterns are "common" and "general" and "valid" for the market as a whole. This is where computers come into help, because they can test it not for thousands, but for 10.000 - 100.000 - 1 mio bars, and... eventually find out that these wonderful patterns are not so "general" and "carved in stone" on the market in its complex.
So, dear Line, I had the impression that, after having claimed in the initial post that you had found "the general rule" (let's not call it the Holy Grail), a fact that I still can believe in the general terms of "trading the line", "keeping the hand in the fire box", i.e. accepting a very high R/R in spite of a low "accuracy" (%win/%loss)...
... in spite of all these good ideas, I had the impression you were/are still "tweaking" quite a lot your system, and this T-trigger line could be no "general solution".
I would like to ask you, if you still have the patience and the goodwill of sharing a bit of your vast experience, how long are you already succesfully trading the line, the way you have shown here.
And, since "where to place the line" is what you wish we learn to do, if possible I will be so thankful if you can give some more insight into how to design the V-trigger, and how long you've already used it successfully.
I hope you'll not take my experiment as a criticism, but, as it is, as a sign of deep appreciation and interest in your exceptional skills.
Bye
BL
Attached File(s)
Sand2_0.mq4
11 KB
|
257 downloads
πάντα δοκιμάζετε τὸ καλὸν κατέχετε