In the last 2 days I've been working on the optimalization of the ea. I've started with the eurusd it's still in progress. This is what I made:
- generated range bar history files from size 3 to 20 pips
- ran backtest on all of them from 2012. jan 1. till 2012. apr 17.
- get the 2 best performing size (9 and 16) and ran a 2 year backtest on them (2010 jan 1 - 2012 apr 17)
- get the better performing bar size (16) and run optimalization on that to find the best Ichimoku setting. I've used a 6 months time range, because I think longer time backwards doesn't necessary since it's good to re-optimize all EA in 3-6 months, the other thing was if I use longer timeframes back, then the optimalization would take waaaay more time.
- now I'm running the optimalization to find the best BEJS settings with the optimized Ichimoku settings - it will finish tomoroww the MT4 says it needs about 30 hours to finish this
- after that I will run a last optimalization to find a take profit setting
I've used version u7 for backtesting because the u8 version said all the time that "Trading suspended" after the first trade taken in backtest, so there was always only 1 trade no matter how I set the EA. It means there is no "AutoBuySell" only oscillating long-short trades.
Now here are the settings so far (the interesting part
):
Range bar size: 16 pips
Tenkan Sen: 9
Kinjun Sen: 5
Senkou Span: 100
UseCleanCandleClose: false
Tomorrow I will post the optimized BEJS and TakeProfit settings, when the optimization finished. After that I will move on to the next major pairs - GBPUSD, USDCHF, USDJPY.
Here are the reports from all the testing and optimalization:
http://dl.dropbox.com/u/1293643/euru...imization1.zip
- generated range bar history files from size 3 to 20 pips
- ran backtest on all of them from 2012. jan 1. till 2012. apr 17.
- get the 2 best performing size (9 and 16) and ran a 2 year backtest on them (2010 jan 1 - 2012 apr 17)
- get the better performing bar size (16) and run optimalization on that to find the best Ichimoku setting. I've used a 6 months time range, because I think longer time backwards doesn't necessary since it's good to re-optimize all EA in 3-6 months, the other thing was if I use longer timeframes back, then the optimalization would take waaaay more time.
- now I'm running the optimalization to find the best BEJS settings with the optimized Ichimoku settings - it will finish tomoroww the MT4 says it needs about 30 hours to finish this
- after that I will run a last optimalization to find a take profit setting
I've used version u7 for backtesting because the u8 version said all the time that "Trading suspended" after the first trade taken in backtest, so there was always only 1 trade no matter how I set the EA. It means there is no "AutoBuySell" only oscillating long-short trades.
Now here are the settings so far (the interesting part
):Range bar size: 16 pips
Tenkan Sen: 9
Kinjun Sen: 5
Senkou Span: 100
UseCleanCandleClose: false
Tomorrow I will post the optimized BEJS and TakeProfit settings, when the optimization finished. After that I will move on to the next major pairs - GBPUSD, USDCHF, USDJPY.
Here are the reports from all the testing and optimalization:
http://dl.dropbox.com/u/1293643/euru...imization1.zip
In the last 2 days I've been working on the optimalization of the ea. I've started with the eurusd it's still in progress. This is what I made:
- generated range bar history files from size 3 to 20 pips
- ran backtest on all of them from 2012. jan 1. till 2012. apr 17.
- get the 2 best performing size (9 and 16) and ran a 2 year backtest on them (2010 jan 1 - 2012 apr 17)
- get the better performing bar size (16) and run optimalization on that to ..................
Honestly speaking, I've not been able to follow the thread throught out, so just finished reading first 40 pages post by post.... remaining tomorrow evening and after that I hope to run some testing through this week and next one to get more idea on range bar sizes etc...
I hope this thread doesn't die... it is wonderful EA and I wish I knew coding, I would love to contribute
Hope we make more green pips...
- #974
- Edited 6:29am Apr 19, 2012 5:54am | Edited 6:29am
- Joined Apr 2012 | Status: Trader | 8,327 Posts
dear seller8 and unimak, you've done a great work and you've showed a great skill.
After months of study on Ichimoku Kinko Hyo I can assure you that any results with this indicator comes from the settings.
If using a very fast ichi (like 2-2-12 who gives in abstract optimal results) the main goal is to achive a great perfection on every open-close position.
To obtain this the ea needs to open on any possible operation and close immediatly if the price inverts.
Using the last version of seller8 I've studied a strange setting to achive this results also if the ea doesn't open on every possible position. But if Seller8 could modify his last version to obtain to enter any position in any moment (exept obviously on heavy news) we made it. Really: WE MADE IT
The target of a very fast ichi is to minimize the numerous little losses and to get any possible pips from the greater positions. With the following settings we will gain a sure loss of max 3-4 pips on any position open<3 candles (RB3-eurusd-spread2) and a sure win on any position open>=3 candles. If only the ea could open properly on any signal, i.e. buy-buy-buy-sell-buy-sell and so on with a very fast and random path...
These are the settings I'm using:
ichi: 2-2-12-0
clean:true
s/l: 5/-2/1 or 5/-2/0
BEJS: (its only purpose is to assure a fast hidden minimal s/l on any open) -3/-2/1
no t/p
So, please Seller8, coplete you're great work
After months of study on Ichimoku Kinko Hyo I can assure you that any results with this indicator comes from the settings.
If using a very fast ichi (like 2-2-12 who gives in abstract optimal results) the main goal is to achive a great perfection on every open-close position.
To obtain this the ea needs to open on any possible operation and close immediatly if the price inverts.
Using the last version of seller8 I've studied a strange setting to achive this results also if the ea doesn't open on every possible position. But if Seller8 could modify his last version to obtain to enter any position in any moment (exept obviously on heavy news) we made it. Really: WE MADE IT
The target of a very fast ichi is to minimize the numerous little losses and to get any possible pips from the greater positions. With the following settings we will gain a sure loss of max 3-4 pips on any position open<3 candles (RB3-eurusd-spread2) and a sure win on any position open>=3 candles. If only the ea could open properly on any signal, i.e. buy-buy-buy-sell-buy-sell and so on with a very fast and random path...
These are the settings I'm using:
ichi: 2-2-12-0
clean:true
s/l: 5/-2/1 or 5/-2/0
BEJS: (its only purpose is to assure a fast hidden minimal s/l on any open) -3/-2/1
no t/p
So, please Seller8, coplete you're great work
Ubi maior minor cessat
- #975
- Edited 6:30am Apr 19, 2012 6:14am | Edited 6:30am
- Joined Apr 2012 | Status: Trader | 8,327 Posts
First of all I'm using RB3 without wicks: this regolarize the kumo and let us to fix a very precise setting for our ea.
Second:
BEJS: -3/-2/1
As you see the last parameter "1" has not a real function, infact with the last version of the ea, after the opening, and after the price reachs the value "-3" it fix the hidden s/l at "-2" and here it remains during the entire operation. So you could also set BEJS -3/-2/0 and nothing changes.
Third:
S/L: 5/-2/1
Due to the very short distance from the kumo a S/L needs to be set at the open at a minimum of 5 pips from the first level of the cloud and then, after the price moves, at a minimum of 1 pip from the second level. You can also set 5/-2/0 but in that case you risk to not activate the S/L on any next 3pips candle becouse the distance from the kumo will be 5-6 pips on average.
So why i'am using the value "-2" to activate the s/l? Because on any open position the spread starts always at-2 so in this way we can activate the s/l for sure also if the price moves backward.
An other possible set is 5/-3/1 in the case the backward movement is very fast.
With this method, any time the ea received a signal to open a position but, suddenly, the price inverts, we guarantie a double minimum s/l, an hidden one (very close to the price) and a visible one in the case the first doesn't work for some reason.
Thank you for your attention.
Second:
BEJS: -3/-2/1
As you see the last parameter "1" has not a real function, infact with the last version of the ea, after the opening, and after the price reachs the value "-3" it fix the hidden s/l at "-2" and here it remains during the entire operation. So you could also set BEJS -3/-2/0 and nothing changes.
Third:
S/L: 5/-2/1
Due to the very short distance from the kumo a S/L needs to be set at the open at a minimum of 5 pips from the first level of the cloud and then, after the price moves, at a minimum of 1 pip from the second level. You can also set 5/-2/0 but in that case you risk to not activate the S/L on any next 3pips candle becouse the distance from the kumo will be 5-6 pips on average.
So why i'am using the value "-2" to activate the s/l? Because on any open position the spread starts always at-2 so in this way we can activate the s/l for sure also if the price moves backward.
An other possible set is 5/-3/1 in the case the backward movement is very fast.
With this method, any time the ea received a signal to open a position but, suddenly, the price inverts, we guarantie a double minimum s/l, an hidden one (very close to the price) and a visible one in the case the first doesn't work for some reason.
Thank you for your attention.
Ubi maior minor cessat
- #976
- Apr 19, 2012 6:42am Apr 19, 2012 6:42am
- Joined Apr 2012 | Status: Trader | 8,327 Posts
Today - eurusd rb3 no wicks
as you see in the last 4 hours we have only 6 position <3candles with a total maximum loss of 3-4 pips x 6= 18-24 pips
and 33 good candles for a total of 33x3= 99pips
so a total gain of 99-24=75 pips
On any open position the price intercepts kumo at 0 or -1 under the last candle and it means that only the last two candle of any position (6 pips+2spread) are lost
as you see in the last 4 hours we have only 6 position <3candles with a total maximum loss of 3-4 pips x 6= 18-24 pips
and 33 good candles for a total of 33x3= 99pips
so a total gain of 99-24=75 pips
On any open position the price intercepts kumo at 0 or -1 under the last candle and it means that only the last two candle of any position (6 pips+2spread) are lost
Ubi maior minor cessat
- #977
- Apr 19, 2012 7:04am Apr 19, 2012 7:04am
- Joined Apr 2012 | Status: Trader | 8,327 Posts
Ubi maior minor cessat
- #979
- Apr 19, 2012 8:31am Apr 19, 2012 8:31am
- Joined Apr 2012 | Status: Trader | 8,327 Posts
Ubi maior minor cessat
- #980
- Apr 19, 2012 9:11am Apr 19, 2012 9:11am
- Joined Apr 2012 | Status: Trader | 8,327 Posts
here there's a clear example:
as you see the ea opens 4 trades of 7: 
as you see the ea opens 4 trades of 7:
- the first one (long) opened by ea with setting: -3/-2/1 (closed at -2pips)
- the second one (long) opened by ea with setting: -2/-4/1 (and I lost 4 pips)
- the third one (long) ea does not open
- Than an other one (short) opened by the ea, same settings (but sell and than -2pips)
- the fifth one (long) ea does not open
- The sixth opened by ea (long), same setting=+14pips
- And here we are! As you see the ea fails again to reopen a long position and I've been forced to open it manually loosing an entire candle (3 pips)
The ea has to open any trade in any condition!
When the price enters and close inside the kumo and than comes out again and closes clean outside (olso in the same direction) this has to be seen by the ea as a good signal to open a trade.
Ubi maior minor cessat