Old Dog with New Tricks | Page 42 | 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

  • #821
  • Aug 15, 2010 6:59pm Aug 15, 2010 6:59pm
Quoting unknown2u
Great Steve, can't wait to hear your findings. Thanks for taking the time to do this.
LOL I'm doing it to make money, no need for thanks.
You are in a maze of twisty little passages, all alike.
  • #822
  • Aug 15, 2010 7:01pm Aug 15, 2010 7:01pm
Quoting stevegee58
LOL I'm doing it to make money, no need for thanks.
LOL...of course.....but at the same time your helping others see what might be the best settings.
  • #823
  • Aug 16, 2010 12:28am Aug 16, 2010 12:28am
Has anyone been changing the min sample size setting for the default of 10?

I have a really good SS, WR and RR for a specific setting and specific hour on GU but the no. trades in the sample from the optimization over 1.5m 1min bars is only 29. I know the definition of min sample setting is:

The minimum number of trades in a sample (One sample is a specific hour, trading style, and takeprofit/stoploss combination) required to optimize.

But I'm not sure what that actually means in terms of what are the maximum number of possible trades within a given hourly sample over 1.5m bars? Thus is 29 really too low in terms of expected frequency of trades and too low to be a reliable statistical figure?
  • #824
  • Aug 16, 2010 3:00pm Aug 16, 2010 3:00pm
Today I ran settings of 500K bars to opto...just on UsdJpy
Had WR set to 65% and SS set to 0.6

No trades for 24 hours.

Guess I'll try something different now!
  • #825
  • Aug 16, 2010 3:13pm Aug 16, 2010 3:13pm
Hrmn.

I think for the next version of this EA I'll have it post the highest values gained for that permutation so you can manually compare against desired values.
  • #826
  • Aug 16, 2010 3:24pm Aug 16, 2010 3:24pm
Quoting unknown2u
Today I ran settings of 500K bars to opto...just on UsdJpy
Had WR set to 65% and SS set to 0.6

No trades for 24 hours.

Guess I'll try something different now!
Did you go through the csv files that are generated by the EA's optimizer?

The "All Permutation Settings" file has the info you need to see if your settings are too tight. Maybe there weren't any trades with those stats.
You are in a maze of twisty little passages, all alike.
  • #827
  • Aug 16, 2010 3:47pm Aug 16, 2010 3:47pm
Quoting stevegee58
Did you go through the csv files that are generated by the EA's optimizer?

The "All Permutation Settings" file has the info you need to see if your settings are too tight. Maybe there weren't any trades with those stats.
No, I guess I need to figure out how to do that lol
  • #828
  • Aug 16, 2010 10:03pm Aug 16, 2010 10:03pm
Quoting Ronald Raygun
Hrmn.

I think for the next version of this EA I'll have it post the highest values gained for that permutation so you can manually compare against desired values.
I thought's that what it does now? If I set min value of WR of 65 and optimise based on SS it will post the highest SS gained for a min WR of 65 within that 'hour'?
  • #829
  • Aug 16, 2010 10:25pm Aug 16, 2010 10:25pm
I did some optimization over the last couple days across the USD crosses and found that the for a WR > 65 RR falls to lower than 1 on average and vice versa for a low WR which you'd expect. The top SS's I found were in the 80 region apart from a few anomalies which were higher but the sample size was extremely low thus trades would be too few.

Are these types of SS's the same as what others are obtaining, I thought I remember Ron saying he gets above 105 on more pairs? (I know you don't want the EA to be trading all the time...i.e. just when it meets some picky criteria).

I'm optimizing based on 1.5m bars.
  • #830
  • Aug 16, 2010 11:49pm Aug 16, 2010 11:49pm
Quoting lovejoy80
I did some optimization over the last couple days across the USD crosses and found that the for a WR > 65 RR falls to lower than 1 on average and vice versa for a low WR which you'd expect. The top SS's I found were in the 80 region apart from a few anomalies which were higher but the sample size was extremely low thus trades would be too few.

Are these types of SS's the same as what others are obtaining, I thought I remember Ron saying he gets above 105 on more pairs? (I know you don't want the EA to be trading all the time...i.e. just when it...
I'm getting SS of between 80 and 90 with a wr of 50 on both eu and uj. I'm using 500,000 bars

Paul
  • #831
  • Aug 17, 2010 3:49am Aug 17, 2010 3:49am
Quoting lovejoy80
I thought's that what it does now? If I set min value of WR of 65 and optimise based on SS it will post the highest SS gained for a min WR of 65 within that 'hour'?
The EA doesn't quite do that now. It shows you the highest values that exceed your criteria.
  • #832
  • Aug 17, 2010 4:56am Aug 17, 2010 4:56am
Ronald, just a couple of quick questions:

- When backtesting does the EA presently take into account spread in its calculations?
- Is there any reason why the EA would close trades that haven't hit TP & SL? (default settings, other than modified minimum Winrate & Successrate).
  • #833
  • Aug 17, 2010 5:08am Aug 17, 2010 5:08am
Quoting Mindsphere
Ronald, just a couple of quick questions:

- When backtesting does the EA presently take into account spread in its calculations?
- Is there any reason why the EA would close trades that haven't hit TP & SL? (default settings, other than modified minimum Winrate & Successrate).
No
When backtesting, the EA holds trades open the maximum amount of bars, by default, 1440 (one day)
  • #834
  • Aug 17, 2010 8:15am Aug 17, 2010 8:15am
I have noticed that sometimes the EA closes the trade even if the SL or TP is not hit.
If a trade was taken based on a Breakout style at a certain hour, and let's say that in the next hour the EA detects a Counter setup, than the current trade will be closed and another one will be taken in the opposite direction.
Most of the time this trade reversal is beneficial (bank some pips and reverse the trade to gain more or cut the losses and reverse the trade to be in market direction).

I will let Ronald confirm!


Quoting Mindsphere
Ronald, just a couple of quick questions:

- When backtesting does the EA presently take into account spread in its calculations?
- Is there any reason why the EA would close trades that haven't hit TP & SL? (default settings, other than modified minimum Winrate & Successrate).
  • #835
  • Aug 17, 2010 8:17am Aug 17, 2010 8:17am
RR,

I started testing with money management last night, and all the trades started closing before hitting tp or sl. Any ideas why and how to avoid this?
  • #836
  • Aug 17, 2010 8:38am Aug 17, 2010 8:38am
Quoting networker75
I have noticed that sometimes the EA closes the trade even if the SL or TP is not hit.
If a trade was taken based on a Breakout style at a certain hour, and let's say that in the next hour the EA detects a Counter setup, than the current trade will be closed and another one will be taken in the opposite direction.
Most of the time this trade reversal is beneficial (bank some pips and reverse the trade to gain more or cut the losses and reverse the trade to be in market direction).

I will let Ronald confirm!
EA automatically reverses a long trade if there's a short signal and vice versa.
  • #837
  • Aug 17, 2010 12:47pm Aug 17, 2010 12:47pm
If a trade is closed during optimisation after 1440 bars does this trade count towards the overall win rate bearing in mind it neither hit sl or tp.

In addition to this what is the general view on how long trades should be kept open for. My tests opened 7 trades yesterday approx 11 am (GMT). These went about halfway to tp and are now in dd, although they are now moving in right direction. My point is that these trades in my view have now lost any relationship with the original trading time, ie whether they were part of a breakout or counter trend. As we have moved into another day they may as well just be random trades as the pair have now got new high and lows for today. Surely the relevance of why the particular trade opened becomes less the longer it is held for.

Cheers

Paul
  • #838
  • Aug 17, 2010 1:35pm Aug 17, 2010 1:35pm
Quoting paulbt38
If a trade is closed during optimisation after 1440 bars does this trade count towards the overall win rate bearing in mind it neither hit sl or tp....
The trade is counted as a win.

The way the optimizer works is (assume reverse trades is false)...

 

  1. An entry bar is found in the bar history.
  2. The EA counts the number of bars after the entry bar when the stoploss is hit. Or the maximum trade length is hit.
  3. The EA determines the highest takeprofit during that period and enters that takeprofit in the TP/SL optimizer.
  4. If the Maximum trade length is hit, the EA assumes the stoploss is hit at the end of that trade length and again calculates the highest takeprofit. In my initial development of this EA, there were some trades which were the 'perfect trades' where the trade is taken at the apex of a trend, which ultimately skews the optimization results.

If reverse trades is false...

The EA looks for the bar when the takeprofit is hit and checks for the largest stoploss in that period.

  • #839
  • Aug 17, 2010 6:25pm Aug 17, 2010 6:25pm
Thanks for the clarification Ronald, however I'm unclear as to whether you answered the first part of my question - does the EA take into account spread in its calculations at all? If it doesn't, aren't we getting somewhat skewed results from the optimizer?

Cheers!
  • #840
  • Aug 17, 2010 6:33pm Aug 17, 2010 6:33pm
Quoting Mindsphere
Thanks for the clarification Ronald, however I'm unclear as to whether you answered the first part of my question - does the EA take into account spread in its calculations at all? If it doesn't, aren't we getting somewhat skewed results from the optimizer?

Cheers!
No, the EA doesn't use spread. No it shouldn't skew the results as long as the sl/TP is greater than the spread.
1 trader viewing now