Old Dog with New Tricks | Page 41 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Old Dog's Taming of the Beast? 34 replies

Old Dog Fib Trading - EA Development 18 replies

Old Dog Fib Trading - EURUSD 399 replies

5 Tricks of Trend Trading, by Rob Booker 25 replies

How To Work Around A Market Maker's Tricks 0 replies

Attachments: Old Dog with New Tricks
Exit Attachments

Old Dog with New Tricks

  • #801
  • Aug 13, 2010 11:33pm Aug 13, 2010 11:33pm
Quoting Ronald Raygun
Why are you using M5?
Ah I thought that was what you recommended at some point in order to minimise whipsaw. You recommend using m1 then?
  • #802
  • Aug 13, 2010 11:44pm Aug 13, 2010 11:44pm
Quoting Ronald Raygun
I gave you the profit factor.

Profit factor => Gross win / Gross loss, right?

My apologies if I seem like a scrooge. This EA was one of the private EAs which I trade with. I have never seen the underlying concept here on FF before, thus, I think it's safe to believe there hasn't been much thought given to this type of EA, and furthermore, given the 50,000 or so members, that's 50,000 potential new ways to improve.

I will not give the settings I use. I want you all to experiment with this EA just as I did 1.5 years ago. Maybe you will make some...
Just found the thread and am now reading if from the beginning.
  • #803
  • Aug 14, 2010 12:22am Aug 14, 2010 12:22am
Quoting Mindsphere
Ah I thought that was what you recommended at some point in order to minimise whipsaw. You recommend using m1 then?
1M or smaller.

Basically the more granular the timeframe the better. I'm trying to generate some sizable 5 tick bars. I have about a month's worth of data. There's some promise but nothing conclusive yet.
  • #804
  • Aug 14, 2010 12:22am Aug 14, 2010 12:22am
Quoting Freddyfx
Just found the thread and am now reading if from the beginning.
Pg 21 is the current EA...
  • #805
  • Aug 14, 2010 12:48am Aug 14, 2010 12:48am
I use M5 with my version on post #621 because it uses a parabolic when counter trading. I continue to test both time-frames on RR's to ensure functionality, and they trade at most only a few seconds apart.
  • #806
  • Aug 14, 2010 1:38am Aug 14, 2010 1:38am
Quoting rfking2
I use M5 with my version on post #621 because it uses a parabolic when counter trading. I continue to test both time-frames on RR's to ensure functionality, and they trade at most only a few seconds apart.
The issue I'm concerned with is the backtest. 5M bars are generally larger than 1M bars and thus an increased likelihood of unsure trades--trades where any two of the entry price, takeprofit, or stoploss are on the same bar.

And RFKing, did you do something to ensure the filenames in expert/files are different? For all we know, the 1M EA's optimize data is being used by the 5M EA
  • #807
  • Aug 14, 2010 2:40am Aug 14, 2010 2:40am
Quoting Ronald Raygun
1M or smaller.

Basically the more granular the timeframe the better. I'm trying to generate some sizable 5 tick bars. I have about a month's worth of data. There's some promise but nothing conclusive yet.
Most interesting! I'm guessing you've got that running separately to metatrader (sub m1 timeframes)...

Do you suggest we limit the EA to less than 10 simultaneous trades? With m1 were more likely to suffer a string of losses even with a high threshold for successrate & winrate, are we not? Is there anything else you can suggest to counter such an outcome?
  • #808
  • Aug 14, 2010 6:07am Aug 14, 2010 6:07am
Quoting Ronald Raygun
The column headings are the same as the optimized settings file. Difference is this is every possible permutation. If you see something there that you like, replace the corresponding hour with the new line.
I checked the all permutations file and saw some weird (negative) numbers for the take profit values. What's that about?

Also, when a SL value is 37600, how many pips is that? e.g. for EURUSD, what value does that map too?

In the areas where the numbers seem more sensible, I found why I didn't have trades. For trades with win rate > 65% the highest reward:risk was 1.0.

This was a real eye-opening exercise. Start up the EA, let it optimize, stop it and go through the all permutations file to find settings that are appropriate.

Here are 2 screen shots of the Excel file where I added my own heading line. The first screen shot just shows the very first lines of the file. The second shows a more normal-looking area.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Image 2.png Size: 36 KB
Click to Enlarge Name: Image 3.png Size: 38 KB
You are in a maze of twisty little passages, all alike.
  • #809
  • Aug 14, 2010 7:56am Aug 14, 2010 7:56am
Quoting Mindsphere
Most interesting! I'm guessing you've got that running separately to metatrader (sub m1 timeframes)...

Do you suggest we limit the EA to less than 10 simultaneous trades? With m1 were more likely to suffer a string of losses even with a high threshold for successrate & winrate, are we not? Is there anything else you can suggest to counter such an outcome?
You can if you want. I don't see too many bad strings, but then again, I use super-long optimization periods.

You can also consider the possibility of a string of wins. The EA is fairly accurate in predicting its own success.

You can also consider increasing the required sample size.
  • #810
  • Aug 14, 2010 8:06am Aug 14, 2010 8:06am
Quoting stevegee58
I checked the all permutations file and saw some weird (negative) numbers for the take profit values....
Your stoploss value is the gross stoploss and your takeprofit value is the gross takeprofit.

The profit column is the net profit.

I am curious to know why the gross profit is negative. I think those were the tested trades where the entry bar is the same bar as the takeprofit bar or the same as the stoploss bar, or the takeprofit bar is the same as the stoploss bar.

In the master copy file, have a look at the trades themselves. Specifically pay attention to start and stop shift.
  • #811
  • Aug 14, 2010 8:30am Aug 14, 2010 8:30am
Quoting Ronald Raygun
Your stoploss value is the gross stoploss and your takeprofit value is the gross takeprofit.

The profit column is the net profit.

I am curious to know why the gross profit is negative. I think those were the tested trades where the entry bar is the same bar as the takeprofit bar or the same as the stoploss bar, or the takeprofit bar is the same as the stoploss bar.

In the master copy file, have a look at the trades themselves. Specifically pay attention to start and stop shift.
The master file doesn't have fields for start and stop shift that I can see. Here's a screenshot of the first part of the file...
Attached Image
You are in a maze of twisty little passages, all alike.
  • #812
  • Aug 14, 2010 8:43am Aug 14, 2010 8:43am
Quoting stevegee58
The master file doesn't have fields for start and stop shift that I can see. Here's a screenshot of the first part of the file...
I don't see anything too abnormal...

The trades are ending either in 1440 bars, or in < 100 bars
  • #813
  • Aug 14, 2010 10:49am Aug 14, 2010 10:49am
Quoting Ronald Raygun
The issue I'm concerned with is the backtest. 5M bars are generally larger than 1M bars and thus an increased likelihood of unsure trades--trades where any two of the entry price, takeprofit, or stoploss are on the same bar.

And RFKing, did you do something to ensure the filenames in expert/files are different? For all we know, the 1M EA's optimize data is being used by the 5M EA

I am only using about 250,000 bars because I don't like waiting any more then a minute to change settings. I have separately installed instances of mt4 to ensure my multiple demo and live accounts don't interfere with each other. This also inadvertently keeps the data files separate. I mean M1 time-frame and M5 time-frame. Using M5 with parabolic prevents whipsaws. In my demo I test "RR SO" on M1 and M5 (separately installed with small starting balances) with all pairs to keep a good account of profitability and ensure time-frame functionality. I am live with "RFK SO" on M5 with optimized pairs, and have good results even threw this past week when most other systems failed due to the market shift.
  • #814
  • Aug 14, 2010 11:55am Aug 14, 2010 11:55am
Will the EA perform the optimisation when the markets are shut...i.e. can I run it over the weekend so I can run various potential settings so I can analyse the csv files or does the markets need to be open?

I tied running it today but the optimisation animation doesn't come up like it normally does?
  • #815
  • Aug 15, 2010 10:42am Aug 15, 2010 10:42am
Quoting lovejoy80
Will the EA perform the optimisation when the markets are shut...i.e. can I run it over the weekend so I can run various potential settings so I can analyse the csv files or does the markets need to be open?

I tied running it today but the optimisation animation doesn't come up like it normally does?
Somebody said last week that you have to wait till the market is open.
We are our own best indicator.
  • #816
  • Aug 15, 2010 11:22am Aug 15, 2010 11:22am
You can attach the EA to your chart, and than close MT4 and start it again.
The optimization works for me even when the market is closed.

Quoting lovejoy80
Will the EA perform the optimisation when the markets are shut...i.e. can I run it over the weekend so I can run various potential settings so I can analyse the csv files or does the markets need to be open?

I tied running it today but the optimisation animation doesn't come up like it normally does?
  • #817
  • Aug 15, 2010 1:11pm Aug 15, 2010 1:11pm
If you right click on the chart and click "refresh", what happens?
  • #818
  • Aug 15, 2010 1:47pm Aug 15, 2010 1:47pm
One major differance I have noticed between trades taken simultaniously. Is M5 trades tend to have a higher TP then M1. Im not really sure why, but winrate stays the same so it is more profitable from my tests.
  • #819
  • Aug 15, 2010 4:44pm Aug 15, 2010 4:44pm
Today's science experiment is to run an optimization and go through the data to come up with some sensible values for MinimumWinRate and MinimumRiskReward. Last week I shot from the hip on these values, not knowing that I'd set them so tight that there were no trades at all.

So I'm running the optimizations now and will be going through the csv files when it's done.
You are in a maze of twisty little passages, all alike.
  • #820
  • Aug 15, 2010 7:48pm Aug 15, 2010 7:48pm
Great Steve, can't wait to hear your findings. Thanks for taking the time to do this.
4 traders viewing now