Great EA in backtest! - page 120
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
how about someone who'se set up to do these 99% modeling tests do one for 2006 Jan 1 to present and post it for us?
The question still hangs
Wow what a touchy long winded response. I didnt ask for a phonebook answer on settings nor should you assume that just because a member asks something , they havent gone and searched heavily for the answer.
All I asked was why, on default settings this system wasnt generating trades (and yes, I damn well checked to see if they were the same settings you had posted on your results sheet before asking).
FxNorth
how about someone who'se set up to do these 99% modeling tests do one for 2006 Jan 1 to present and post it for us?
hey i've archived your same broker.. now the 1.93 starting to looks like yours profits
you have done a great work over the filter!
now i'm trying to improve the logging system to get more variables.. and put it as CSV to bind on worksheet
some variables i'm putting on was
CCI
DecisionValues
Divergence
orderType
hasWon
and some others...
i'm wondering to get the time that the trade spend opened to get win/loss
later.. put all that number sorted in some tables/charts and then improve some more the filter or do some changes on the trading logic
about the 99% from january until today, we would need that kind of data to make that test.. i've just found that project (some ppl that work together to log real market data classified by broker)
and about M1 feed from 1999 to 2006 sounds good?
New MetaTrader 4 build 199 (*Added downloading and import of data in the History Center)
it download the all data charts so we now could always get 90% model quality... that's all for now
hey i've archived your same broker.. now the 1.93 starting to looks like yours profits
you have done a great work over the filter!
now i'm trying to improve the logging system to get more variables.. and put it as CSV to bind on worksheet
some variables i'm putting on was
CCI
DecisionValues
Divergence
orderType
hasWon
and some others...
i'm wondering to get the time that the trade spend opened to get win/loss
later.. put all that number sorted in some tables/charts and then improve some more the filter or do some changes on the trading logic
about the 99% from january until today, we would need that kind of data to make that test.. i've just found that project (some ppl that work together to log real market data classified by broker)
and about M1 feed from 1999 to 2006 sounds good?
New MetaTrader 4 build 199 (*Added downloading and import of data in the History Center)
it download the all data charts so we now could always get 90% model quality... that's all for nowFantastic...
90% of modeling quality since 1999 to 2006 would be a really good field for us to test not only CT 1.93a, but any EA...
Still working on how to get 99% (that is not simple)...
Already installing MT4 v1.99 (lets remember that we can use this "neutral" platform and log in at any Broker's Demo or Real account server, by IP, Log in and password), but to avoid making a mess on the historical data, at the time I'll only use it for backtesting...
1.93ppf
PPF stands for "putting pips first"
yes I have one more change added to this and here's the concept I'm testing now.
yesterday I won two trades and lost one trade. The loss canceled out the two wins. I didn't like that. The stoploss at 17 made the loss -17. That part is fixed but what about the wins..ah now those are determined by CT's probability logics. It takes profit when it thinks the probabilities are turning. So sometimes it takes profit with less than 7 pip gain in the position. yesterday my two wins were +7 and +8 that made =15 total which was two less than the loss. I reasoned that if the stop loss is 17 than the minimum win I should accept is 9 because that is what is required to exceed 17 in two trades. In this version two winner WILL overcome one loser because I mandated that no matter what CT's probability logic says, if the position hasn't improved at least 9 pips then DON'T exit.
So this update has that command hard coded into the market exit function.
It appears from the test that yes this has some negative impact on the win loss ratios they are down to the mid 70's % instead of the mid and upper 80's % however if they are still over 66% then it's still winning more than 2 to 1 and that's all it needs to do now.
Again it remains to be seen how this works in live account. The backtester says is should still work. I'll let this run live with maxlots=.01 and count pips on the results. If we win the pip game the money game will follow.
seems like I tried something like this already, I don't remember. but it seems like this just might work...that relative drawdown is only 8.09% even when allowed to trade maxlots=10 and risk 1 which takes it to max lots really fast in my backtester. 8.09 isnt' bad.
Wow what a touchy long winded response. I didnt ask for a phonebook answer on settings nor should you assume that just because a member asks something , they havent gone and searched heavily for the answer.
All I asked was why, on default settings this system wasnt generating trades (and yes, I damn well checked to see if they were the same settings you had posted on your results sheet before asking).
FxNorthSorry you're offended but you know no one on this site owes you anything. Least of all me. I don't know why you're not generating trades. I didn't reply earlier, because I don't have a clue. I don't know what to tell you to do. Why loose your angst on me? My comments were not directed at you.
Good luck. I suggest that if you get it figured out post your solution so others who might have the same issues could resolve them following your lead.
PPF stands for "putting pips first" yes I have one more change added to this and here's the concept I'm testing now.
follow my backtests with PPF, also i'm attaching a binded worksheet with almost all variables from results of my last post (it has more than 1200 logged trades)
saying about pips, the PPF version got less pips than your last version posted
follow my backtests with PPF, also i'm attaching a binded worksheet with almost all variables from results of my last post (it has more than 1200 logged trades) saying about pips, the PPF version got less pips than your last version posted
Hey that's some really nice looking work templar! I really like your spreadsheet. (column b and c are reversed from their respective labels) What findings and conclusions do you draw from it?
I notice that your backtest says that you're winning 72% of shorts and 73.53% of longs. Now that looks good to me if the wins are no less than +9 and the losses are no more than -17. That should work shouldn't it?
okay... lol
let's complicate it a little bit more:
build 200 has already been launched:
to check the improvements: http://www.metaquotes.net/news
Now that looks good to me if the wins are no less than +9 and the losses are no more than -17. That should work shouldn't it?
As follow my last post's backtest, I could see that strategy *add minimal profit to 9 pips or simple PPF* wasn't so good as we got less profit in the end.
I need some help to make some things to the statistical file using mql4
confused)
- how to add data at end of an line in the file?
( because when a trade is open i write a line with many data, but when the trade was closed i need to add some more data at same line, all i can do now was to write a new line )
- how could i get time the trade spend opened, either to won/loss?
- how to call a function when some trade was closed at server side *StopLimit*?
The idea was to send to a worksheet using a csv file with following data per trade
- Pips won/loss
- Spent time until close*
- All CyberiaLogic variables
- CCI value (at open and *if possible* on close)
- Divergence
- Sugestions?
*I think that such information was really important, maybe later we could do some decisions as "90% trades that spent more than 3 hours opened loss, or never reach more then X pips of profit" and then do some changes at code.
I think that we could really improve this expert, with such statistical analisis tool working.
Also follow the changes that I've made on 1.93 version as a minor release A**:
- added functions WriteFileHeader and logTrade optimize the size of code.
- added external parameter to name statistical file to be writen at folder tester/files/NAME
- down the parameter OneTradePerBar to end of input list.
- translated the remaining russian comments to english.
**Just to remember that changes just apply to logging system results are the same!