Trust backtesting ?

Hi,

I run a ea that gives good result in live account ( Statement attached).
But this EA is not so good when backtesting with metatrader.

I had 90% modeling quality ( http://alansforexblog.com/2008/02/27...pert-advisors/ )

So I vote for patience and forward testing when try a new EA.

Of course Backtesting gives a hint but no real picture.

Henrik

Henrik, see my recent post here: http://www.forexfactory.com/showthread.php?p=2156275

Hope you find it helpful.

Re your equity plot, the RHS bears no resemblance to the majority of the curve. Looks very strange to me! I would definitely want to know what caused this, before considering to trade the EA.

Best wishes,
David

Hi,

Thanks I will check your post.

Henrik

Hi,

Probably subject off topic because I want to speak about backtesting or not

Can you explain what you mean by strange ?

The EA trade one currency and then if some positions do not go right the system create position in other direction for close the bad trades. ( So no stop loss used)

I think we need to check the profit / month and not balance graf.

Henrik, by "strange" I mean that the right side of the equity curve is very different from the rest of it. See my annotation in the attachment below. I would want to know why the system has performed well for a while, and then suffered a (supposedly) uncharacteristic drawdown. It might have something to do with unrealized losses being taken into account at the end of the testing. The shape also becomes much more jagged, which would also concern me.

The system has a high win rate (82%) but low win size (13.83) to loss size (51.42). That suggests to me that, for better or worse, it's probably taking profit early, and letting losses run. However, the hedging algorithm could be more complex than that.

Anyway, the bottom line is that I won't trade anything that I don't understand. There are plenty of other approaches to trading that have a clearer rationale.

David

Hi,

hanover :: Thanks for your comment.

No other have anything to say about back testing a ea or not ?

1- Imagine that the back test is very bad that live trading is very good ? Is this possible ?
2- If 1 is possible maybe there is a lot of good ea's in the trash bin

Have a nice day .....

It's very possible for bad backtesters to be good forward testers.

You just need to find out why.

my backtesting always agreeing by live trading.

if disagrreeing 4 u, u need using real tick data. it will hleping much.

Hi,

I talked to some MetaTrader progs and in some case back testing will not give the correct result. He talks more about good EA in back testing and bad live

I try to get the same result as my live account with back testing but without success ....

if your stratigie depending on scalping, then need tick data.
but if swing trading on bar closing/opening, etc., then backtesting on general data is perfect ok.
i get 100% agreeing, even very fast scalping methids.

Hi Folks

I have some experience and need help
1. I test my Scalping EA that uses just 4-5 pips only and SL=11.

2. In Backtesting it works fine.

3. When I start testing it with Live account, first ten orders it won then rest of the orders are hit SL with a pip or 2 pip difference.

4. When I exported the M1 data from my broker and from Dukas. I found the same strategy performs on my broker data strangly whereas on DUKASCOPY data it works fine.


could this be my Broker feed problem. That they dont have consistency in their feed whereas DUKASCOPY are more reputable and their feeds are consistent.

If someone can help me in this I would sincerley appreciate help and obliged.

Regards
ForexZAK