Current Status - page 14
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
1. Fxfisherman bot might be finished by October.
I'm not going to make it by October. Let's not talk more about the release date.
In other words, it will be released when it's finished. 
Exactly as Microsoft would have said
And then again their the biggest software company in the world
Take some time to rest scorpion.
Dear Sirs,
Please don't take this as a criticism, because it certainly isn't that.
... but...
I suspect that you're finding the performance for your bot somewhat less than you had expected. How do I know? Because I've been developing my own for over two years now, and every time I think I've got 'the solution', there seems to be more work that needs to be done. I don't think that this is a reflection on your (or my ability), but you're not placing performance data on this site, and I suspect that this would be an immediate requirement, if you ever want to legitimize this process. This hasn't been done... ergo, the performance is not of a level that would justify publication.
But I've taken a different tack with my development, and am now hitting the new problems of order placing. Alas... still a little while to go to implement my system methinks.
Unlike you, I've been ignoring (for now) the front-end process of connecting to a given dealer, and instead, been focusing on the back-end system of logic, and continually back-testing it. The first problem I encountered was the sheer data amount, and selecting the appropriate functions by which a given trade decision could be made. I'm sure you've found that your various methods of selecting a given buy or sell signal inevitably conflict. One says buy, one says hold, one says sell. Picking the correct one, out of all of the options available, is about as complex as using only one... and then deciding when to apply it. I addressed this problem by an artificial life method of natural selection, and allowed this process to determe which are the most appropriate logic decisions, and when to apply them.
The next problem became obvious as soon as I implemented these systems. The sheer processing power became a mamoth issue. For a minutes trade processing, it was taking upward of five minutes to process. Not very helpful. So I wrote a client/server program that lived on many pc's and authenticated to the server, providing processing power, when the computers were idle. I placed the client on 20 computers, and only then could I start getting real-time data crunching.
I would be interested in your thoughts. Unlike you, I'm not interested in the charity issue... so forgive me for wasting your time if this is a problem for you. But methinks there are a few things you're all going to need to investigate before this system actually produces what is promised.
Sincerely,
Mleck
Thanks for your great comment, Mlect. Your word is informative indeed. I think your development doesn't differ much from mine. There're times that we think it's a final solution, but it isn't, and we wonder why another issue pops up when one issue is solved.
Well, Systems R&D is always put ahead of any other things for me. I'm not really that concerned about front-end thing, it's just because I talk about it more in forums. Not mentioned much before, I lease a few high-end servers connected to the net 24/7 in Canada. One server to test the client and server live and to test various MT4 experts. One more server to bruteforce millions of parameter combinations for various systems. Another spare one to collect ticks data from InterbankFX, FXDD and Alpari--now I want to solvely rely on ticks coming from real brokers for use in backtesting. Mentioned in blog section, I'm pledging some cash to try Olsen's ticks data out. Just 1 year of trial data ain't looks cheap to me! so it's a real dilema here.
Besides all above, I oftenly promote trading system research in Trading Styles and Strategies forum. All members are encouraged to bring in profitable stuff, and some of them did--and I'm lovin' it. Unlike non-programmers, once there's something profitable in sight, I'll dechiper it in free time, looking for the reason/fact that push up the profit.
You somewhat see the point why I don't post the result. For one reason, it's in fact just a result of backtesting data from 3rd party data feed. The PL number is so great in backtesting, but it is not the same when I test them live with real, dealable data from real broker. "I suspect that you're finding the performance for your bot somewhat less than you had expected" is somewhat the case, but for one thing I'm not gonna quit testing, researching, or developing, unless i'm cornered AND my project resources run out.
The day I post the finalized result is going to be day that I declare Real Success, if not Tragic Failure. But now it's not the time to go forward posting such result.