Old Dog with New Tricks | Page 34 | 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

  • #661
  • Aug 5, 2010 9:26am Aug 5, 2010 9:26am
ok sorry guys i need to sleep this off
Time hides Nothing
  • #662
  • Aug 5, 2010 9:28am Aug 5, 2010 9:28am
Quoting MorningMania
I think the old refers to the basic two strategies: breakout and countertrend. The new trick is that the ea decides between the two based on back testing. Or I could be totally wrong
Ding ding ding, we have a winner!
  • #663
  • Aug 5, 2010 9:41am Aug 5, 2010 9:41am
Need your expert guidance on which system to choose for analysing forex trades.

I am in love with P&F and just read the book of Thomas Dorsey.

Regards
  • #664
  • Aug 5, 2010 10:14am Aug 5, 2010 10:14am
Quoting Ronald Raygun
Boolean values when converted to integer are 0 and 1.
So which file should we be using? Steve's updated one or Ron's updated one? Or either b/c they both do the same?
  • #665
  • Aug 5, 2010 10:15am Aug 5, 2010 10:15am
Quoting unknown2u
So which file should we be using? Steve's updated one or Ron's updated one? Or either b/c they both do the same?
Either.

Steve's separates the OrderModify and OrderSend functions. Mine doesn't.
  • #666
  • Aug 5, 2010 4:19pm Aug 5, 2010 4:19pm
Ronald,

I tried to log on to your webinar, but I only got the message "Has Not Begun," plus 'on-hold' music. Were you able to make it? Is there a way to attend in the future?

Thank You,

Jay Allen
Dallas, Texas
  • #667
  • Aug 5, 2010 4:41pm Aug 5, 2010 4:41pm
9PM EDT. I'll be getting on in about 3 hrs or so.
  • #668
  • Aug 5, 2010 5:05pm Aug 5, 2010 5:05pm
Quoting Ronald Raygun
9PM EDT. I'll be getting on in about 3 hrs or so.

Ronald, after reading your posts on this thread I realized that your settings must be much more tight or with more restrictions than the default settings and this is because you have said that your EA have taken only 8 or 9 trades a week, well I m getting hundreds of trades per week.
  • #669
  • Aug 5, 2010 5:11pm Aug 5, 2010 5:11pm
Quoting Forexaccount
Ronald, after reading your posts on this thread I realized that your settings must be much more tight or with more restrictions than the default settings and this is because you have said that your EA have taken only 8 or 9 trades a week, well I m getting hundreds of trades per week.
Let's start with the obvious one. My winrate is not 50. :-)
  • #670
  • Aug 5, 2010 5:25pm Aug 5, 2010 5:25pm
Quoting Ronald Raygun
Let's start with the obvious one. My winrate is not 50. :-)
LOL. I knew it.
  • #671
  • Aug 5, 2010 5:26pm Aug 5, 2010 5:26pm
A 50 winrate is a little low. You might as well flip a coin.

It is impressive that the bot is still profitable though.

I wonder if you could make it be profitable with a 25 winrate
  • #672
  • Aug 5, 2010 5:29pm Aug 5, 2010 5:29pm
Quoting rfking2
A 50 winrate is a little low. You might as well flip a coin.

It is impressive that the bot is still profitable though.

I wonder if you could make it be profitable with a 25 winrate

you meant 75 win rate. Right!!!
  • #673
  • Aug 5, 2010 5:35pm Aug 5, 2010 5:35pm
No, I mean I would like to see if you can profit from 25. Maybe the TP would be so high that it would cover the losses.
  • #674
  • Aug 5, 2010 5:41pm Aug 5, 2010 5:41pm
In theory that would work.

Buy why not have your cake and eat it too? High Reward/Risk and a High Winrate.
  • #675
  • Aug 5, 2010 6:06pm Aug 5, 2010 6:06pm
Quoting rfking2
A 50 winrate is a little low. You might as well flip a coin.

It is impressive that the bot is still profitable though.
Yeah, my 2 weeks of forward testing at 50% were quite promising.
You are in a maze of twisty little passages, all alike.
  • #676
  • Aug 5, 2010 7:47pm Aug 5, 2010 7:47pm
Correct me if I'm wrong, but the winrate is not really the important factor here. The success rate is what dictates whether or not the current trade setup should be profitable. i.e. it's a measure of the winrate multiplied by the r/r, so anything above 50% means a better return than flipping a coin. Hence why a 50% winrate can be profitable.

That's as long as we keep letting the EA optimize by success rate (as I'm guessing everyone is doing?)

By limiting the trades to a higher than 50% winrate we're just introducing an additional filter which would make the trades more likely to be successful, however we could just limit trades to a higher success rate with similar effect, could we not?

I notice there was discussion earlier about weighting more recent results more heavily. Perhaps one way of doing that is to apply a linear gradient, so that the most recent are weighted X times as much as the oldest results.

This backtest system could also be utilised for a number of other trading strategies. It would be interesting to see it applied to some other easy-to-code currently zero-edge approaches such as MA crosses, rsi trades and so on, to see if an edge can be found. More complex approaches might prove more fruitful, however.

I haven't had a chance to look through this code properly yet, but I'm wondering how difficult it would be to plug in different logic for the backtest system. Any tips on this, Ronald? Or is that not within the scope of what you intended to happen in this thread?
  • #677
  • Aug 5, 2010 8:00pm Aug 5, 2010 8:00pm
For everyone who wanted to join, the webinar is now starting, ask away.

Room Name:
ronaldraygun
Organizer:
ronaldraygun
Start Time:
August 5, 2010 8:51:45 PM EDT
Dial-in :
702-473-3463
Passcode :
766253
Meeting URL:
https://my.dimdim.com/ronaldraygun/
  • #678
  • Aug 5, 2010 9:55pm Aug 5, 2010 9:55pm
Quoting Ronald Raygun
For everyone who wanted to join, the webinar is now starting, ask away.

Room Name:
ronaldraygun
Organizer:
ronaldraygun
Start Time:
August 5, 2010 8:51:45 PM EDT
Dial-in :
702-473-3463
Passcode :
766253
Meeting URL:
https://my.dimdim.com/ronaldraygun/


Thanks ron, very well done
  • #679
  • Edited Aug 6, 2010 7:00am Aug 5, 2010 10:01pm | Edited Aug 6, 2010 7:00am
Quoting Mindsphere
By limiting the trades to a higher than 50% winrate we're just introducing an additional filter which would make the trades more likely to be successful, however we could just limit trades to a higher success rate with similar effect, could we not?
The relation between win rate and risk reward is inverse within the same success score. Success score of 50 could mean a win rate of 50 and a risk reward of 1. It could also mean a win rate of 90 and a risk reward of (5/9).

By adding one of the other two parameters--risk reward or win rate--you more finely tune the EA's optimization engine.


Quoting Mindsphere
I notice there was discussion earlier about weighting more recent results more heavily. Perhaps one way of doing that is to apply a linear gradient, so that the most recent are weighted X times as much as the oldest results.
As I explained on the webinar, this method may be tricky due to how the optimizer works.

Quoting Mindsphere
This backtest system could also be utilised for a number of other trading strategies. It would be interesting to see it applied to some other easy-to-code currently zero-edge approaches such as MA crosses, rsi trades and so on, to see if an edge can be found. More complex approaches might prove more fruitful, however.
More complex generally means longer backtest times and even longer optimization times.

Quoting Mindsphere
I haven't had a chance to look through this code properly yet, but I'm wondering how difficult it would be to plug in different logic for the backtest system. Any tips on this, Ronald? Or is that not within the scope of what you intended to happen in this thread?
The backtester is simple--conceptually. It just needs a few things:

 

  1. Way to track trades and profitability
  2. Way to model the price movement within a bar.

In terms of using the structure of this EA to run the backtests for other systems, that's mighty difficult. Every time I want to write an EA that optimizes itself, I have to write from scratch.

  • #680
  • Aug 5, 2010 10:08pm Aug 5, 2010 10:08pm
Quoting Ronald Raygun
As I explained on the webinar, this method may be tricky due to how the optimizer works.
This is what I was trying to explain on the mic, if you are able to make it look back x bars like with trader and still keep hourly arrayed like in optimizer the more recent results will have a greater weight. Not much though.
1 trader viewing now