Backtesting Mt4

[samalla]

I am hoping someone can help me here.

A lot of people say that Mt4 is not very good at backtesting, and you can't trust the results. Others say its accurate.

why do so many people disagree on this?

can ANYONE clarify this, please.

Thanks.

[jezzer1961]

Originally Posted by samalla

I am hoping someone can help me here.

A lot of people say that Mt4 is not very good at backtesting, and you can't trust the results. Others say its accurate.

why do so many people disagree on this?

can ANYONE clarify this, please.

Thanks.

MT4 uses a method called "Fractal Interpolation" to guess the tick sequence on the M1 timeframe. So any EA which has a SL or TP of less than 20 pips and enters mid-bar is likely to produce inaccurate backtest results.

Unfortunately a lot of EAs are scalpers which fall into this category.

Any EA which seeks a higher profit and which enters at close of a M5 price bar or slower is likely to produce more accurate and reliable results.

[geppyme]

Backtest on tradestation, much more accurate.

[alexvali]

I am doing backtesting on MT4 since a while, with various strategies, on data from different servers. I must say that if you repeat the test a few times you can get different results, even though you work on daily or 4h.
That leads to the conclusion of non reliability. Cause? I don't know. Does anyone know any other methods for backtesting?

[SquareRoot]

I can confirm this also. I've tested EAs in real time and then backtest them for the same time period and the results differ sometimes significantly when using scalping type EAs on short time frames.

[alexvali]

One interesting feature in MT4 backtesting is the optimization. This will give you the best parameters values to maximize your winnings.

[beaker]

you need at least to have 90% modelling quality. These 2 weblinks can be useful:

90% Setup Info
http://alansforexblog.com/2008/02/27...xpert-advisors

FREE Data
Forex Tester Software - Data Sources

[aft]

Indicators, EA, Systems are just trader tools and you must know how to use it.
If you don't know how to play chess can you win the game?
Same deal is trading.
Never pay attention for advertising tricks like" You don't need any experience..."
Certainly there are some useless soft on market.
Even the best EA or system not gonna help you if you don't know what you are doing.
For example back test and optimization.
Everybody use world "forward test" but how many of them know what that's mean?
Test and optimize your EA on 4/5 your testing time period.
For instance for 5 years testing take only 4 years and then run optimized EA on the last period.
Like i sad it is not so ease as seems to be.
Your tester doesn't recognize Fri-Sun or financial news and as a results is
difference between testing and live trading.
Different markets must use different settings and so on.
By our sell experience we are seen that people spend a lot of money on soft and never for learning.
For instance Forex Robot Reviver which helps maintain your EA.
Everybody blame only brokers but not their own ignorance.

[riki143]

One interesting feature in MT4 backtesting is the optimization. This will give you the best parameters values to maximize your winnings.

[vmusci87]

How to do a backtest, can anyone share a link?

[dangtrader]

There is an option in MT4, known as strategy history, here is a screenshot where its shown from my Hotforex platform:

[piplaser]

Originally Posted by beaker

you need at least to have 90% modelling quality. These 2 weblinks can be useful:

90% Setup Info
How to Achieve 90% Modeling Quality When Testing Expert Advisors | Alan's Forex Blog

FREE Data
Forex Tester Software - Data Sources

Also Check Logic Trading ? Let you money work for you

[dangtrader]

Originally Posted by samalla

I am hoping someone can help me here.

A lot of people say that Mt4 is not very good at backtesting, and you can't trust the results. Others say its accurate.

why do so many people disagree on this?

can ANYONE clarify this, please.

Thanks.

It's not a fault of MT4, the reason why backtesting results are unusable if quality of historical data from metaquotes. Brokers doesn't offer historical data so you are forced to test your algorithms using metaquotes data which can be significantly different from broker history, hence produce skewed results.