To use chat, please login.
Back to contacts

What is considered a great EA? :)

tbuitendyk (tbuitendyk)
Mar 24 2010 at 18:10
141 posts
paul posted:
    1. What % per month should it be averaging?
2. How long should it be forward tested?
3. Do you think back tested results are important?
4. What other factors should we look at before buying an EA?
5. Are most commercial EAs hyped?
6. Are you personally doing well using an EA?


Good questions Paul, thanks for asking...

1. Percent / month? I don't think that % / month is really relevant as it's too easy to skew. One can take 500 pips / month and make it turn 40% by over-trading with 1:500 leverage. It's also possible to take 2000 pips / month and make it turn 10% by under-trading too cautiously. In my view what should be measured is pips / month or pips / day on average. This is how I gauge the performance of our Volatility Trader. (See 'Volatility Trader Rel.3c - LIVE-A' discussion...)

2. Forward testing? Depends on the system. A static EA that only supports one trading mechanism should probably be forward tested for weeks or months before going live, but that depends too on one's risk tolerance. That being said, for an EA that is regularly adapted to current market conditions such as our Volatility Trader, forward testing does not really prove much -- market conditions change and the EA is continually being re-modeled to work with current conditions, so all forward testing proves is that the market conditions adaptation is currently good. The best forward test is with cautious use of real money.

3. Backtesting... That's a loaded question. 😉 It depends entirely on how it is being done and for what purpose. It probably also depends on the nature of the automation. For example, VT uses 'backtesting' so to speak as a modelling/programming phase that teaches the engine generally how the market is behaving in the recent past, with enough room for error that the past can be expected to project forward reasonably well enough to make trading decisions. In the case of our EA, modelling (i.e., 'backtesting') is critical to its designed function, and has less to do with generating fancy graphs than it has to do with teaching the pattern detection engine.

4. Considerations on buying an EA... I would probably not buy an EA again in the future, but if I did I would look for tight stops, use of hedging, attention to money management and account preservation, frequency of trading, good support for set-up details, etc. Also, I would want to know why someone would sell a successful EA in the first place. While I cannot really understand why anyone would sell a successful EA, I suppose that it's possible that there are conditions 'out there somewhere' that make selling one's EA the best way to market successful automation. Frankly I don't know. As many people have pointed out in the past, it's hard to imagine a reason to sell a successful EA -- to my mind it is better to just trade it or sell access to signals, managed accounts, etc.

5. Yes.

6. Yes. While rolling out and tuning the Volatility Trader over the past couple of months it up 15% since January with very low order sizing. VT Rel.3c is currently doing something like 150 pips per day on average. With the new money management guidelines we expect to see 20% - 30% per month going forward starting in April. Email to VTSS@homeandofficemicro.com if you would like to trade our signals.


Gear on the left...
MIDASUser
Mar 30 2010 at 00:02
84 posts
paul posted:
    1. What % per month should it be averaging?
2. How long should it be forward tested?
3. Do you think back tested results are important?
4. What other factors should we look at before buying an EA?
5. Are most commercial EAs hyped?
6. Are you personally doing well using an EA?


1) More than a bank could offer you in a year.
2) At least a month+
3) Only if they represent true history data and the broker supplying the data uses the same pip spread calcs. on history data being used for back testing as the account you intend to use it on. And I know for a fact of at least one very well known broker that does not, so back testing becomes pointless!
4) On-going support, free updates, the cost versus the returns achieved, money back guarantee etc. etc.
5) Err..yes.
6) Yes, very well thanks! 19.74% per month at time of writing!

When everything else out there is just not good enough, design your own! MIDAS
Fulltime247
Mar 30 2010 at 01:02
131 posts
donnalife posted:
1. What % per month should it be averaging? 10-25% per month constant
2. How long should it be forward tested? 2-3 months max.
3. Do you think back tested results are important? no
4. What other factors should we look at before buying an EA?
5. Are most commercial EAs hyped?
6. Are you personally doing well using an EA?


1. Make 5% to 10% a month and you'll do over 100% per year. George Soro's and Warren B would be Jealous.
2. A month plus or until you see the EA follow the strategy rules flawlessly every time.
3. No. I dont believe in them. Backtests are crap. (I only backtest to check code. I ignore the results.)
4. Forward Demo's with known brokers. LIVE Money Forward tests. (if a scalper)
5. It Slices. It Dices. It'll Grow Hair and Fill your Pockets with CASH!!...
6. Yep. 5-15% per month with about 9% DD is fine with me.

It ain't easy being Cheesy!
euphoria
Apr 20 2010 at 10:08
14 posts
Wow, first Post. I am currently 'playing' with EA development and am writing my own code at the moment. It does show promise. From my experience and perspective a great EA needs to obviously make money. It must do this:

a) Consistently
b) Without risking large % of account
c) Not opening 10 trades in one direction at the same time, just increase the lot size for F$% sake.
d) Low Drawdown, anything over 20% is too high in my opinion, and less than 10% commendable and anything less than 5% is great.
and finally, one which I cannot stress enough
d) Not make profit by trading with no stop loss or such a large stop loss so that you risk reward is terrible. I greatly disapprove of seeing EA backtests/forwardtests with the balance always increasing and it seems to be so consistent and awesome, until you realise that the account is actually in negative equity and floating on a number of losing positions. Just because you don't close the trade doesnt mean that it didnt happen.😡

And to the people bagging backtests, while a backtest is by no means an absolute proving ground and I wouldnt rely on it. If an EA cannot make good results on a backtest, it is not worth progressing with forward testing. And as already mentioned, backtesting is the only way to check the validity of coding routines and functions. I personally, ( as I believe most developers would be) would be completely lost without the ability to backtest.

Just my $0.02

jacklee090755
Aug 02 2016 at 06:35
2 posts
paul posted:
Jacky, thanks for your post... Are you currently using any EA? Any toa recommend?


aaaaa

Please login to comment .