How reliable is Tester?
Hello, I am quite new to this forum, but i've been here a lot in the last couple of weeks to read through the Documentations and other user post for help when I was writing my own EA code.
Well, here is the question. I wrote this code and like to ask you what is your experience with the backtesting feature in MT4? I mean, when you wrote yours and did the test and then run it live, was it working as expected?
Here are few printscreens of my test. Its basicaly developed for EURUSD 5min chart and backtested since Januare 1st till now. I tried to run the tast for the past five years but it didnt show any results and said NA.. So not sure why is that. I checked the history and have it since 99. Anyway, the lots are not optimazed so its keeps trading the same amout (1lot). I was even happy two achive write my strategy so it would work. I will probably spend few more hours trying to optimize it for the trading size so it gradually increase or decrease based on the account value.
I tested on $2000 account start. The report looks a bit confusing since my broker uses 5 decimals prices, so I had to write the code in 10times incremetal then normal.
Plase let me know what you think of the results and any commets would be much appriciated. I'm thinkink of opening live account and let the script do the magic.. Do you think that is a good or bad idea?
So, here are the screens:
Looks pretty good, however when you set up lot size management the drawdown could be a problem, consecutive losses looks high as well.
Tester results may or not be better than live testing results, live testing is never better than real trading, but can be pretty close depending on broker and ea.
In any event do not trade real money until you have run the EZ in real time on a demo. Most members here will tell you to run the ea for one month before switching to real money. Good advice.
Good Luck
Keith
JM
I second KM's post, all good comment
IME, EA's will always be less good on live accounts, some dramatically so.
Here we have a equity line that does not create confidence, low percentage of wins, drawdown & especially consecutive losses too high
Is the strategy <long only>?
Have you limited trading hours to the main session/s?
Also watch out for periods where the EA goes out of sync with market action
FWIW
-BB-
JM
I second KM's post, all good comment
IME, EA's will always be less good on live accounts, some dramatically so.
Here we have a equity line that does not create confidence, low percentage of wins, drawdown & especially consecutive losses too high
Is the strategy <long only>?
Have you limited trading hours to the main session/s?
Also watch out for periods where the EA goes out of sync with market action
FWIW
-BB-
> some trades were closed early
Was this by a trailing stop trailing too close, stop loss hit, EA signal-to-close - what does the Results tab show for these?
> and some were skipped altogether
Is this where the chart 'looks' to like a trade should have opened but the EA didnt?
> Please share how you transitioned into live trading
With much stress, pain & cost...
As a minimum the system has to backtest consistently to your expectations and then forward test on demo for 3 months if it is one of your first
Then, come the day, you just have to go with it on live - aint no guarantees, this is Forex..
In time you get to know the warning signs, the shape of the Balance/Equity curve, the ratios of Profit Factor, Drawdown, Win % and max concurrent loss
And, one day, you get to the point where you may backtest a bit, maybe optimise a bit then just start live on a micro account - no need to rush to do that though!
Good Luck and lets be careful in the Forex
-BB-
Hello, I am quite new to this forum, but i've been here a lot in the last couple of weeks to read through the Documentations and other user post for help when I was writing my own EA code.
Well, here is the question. I wrote this code and like to ask you what is your experience with the backtesting feature in MT4? I mean, when you wrote yours and did the test and then run it live, was it working as expected?
Here are few printscreens of my test. Its basicaly developed for EURUSD 5min chart and backtested since Januare 1st till now. I tried to run the tast for the past five years but it didnt show any results and said NA.. So not sure why is that. I checked the history and have it since 99. Anyway, the lots are not optimazed so its keeps trading the same amout (1lot). I was even happy two achive write my strategy so it would work. I will probably spend few more hours trying to optimize it for the trading size so it gradually increase or decrease based on the account value.
I tested on $2000 account start. The report looks a bit confusing since my broker uses 5 decimals prices, so I had to write the code in 10times incremetal then normal.
Plase let me know what you think of the results and any commets would be much appriciated. I'm thinkink of opening live account and let the script do the magic.. Do you think that is a good or bad idea?
So, here are the screens:
Don't forget about historical spreads! You can not simulate them in backtester unfortunately. If your ea is using 1lot and expected payoff is 80, then you have profit 8 pips per trade. This is when you backtest with constant 1.8 pips eurusd spread. Now try to add historical spread to every transaction... Profit will drop down :(
The EA going out of sync with market action concerns me. I noticed in my backtest that some trades were closed early and some were skipped altogether. How can I have any confidence in MT4 signals going live if it doesn't even trade properly in the demo? Please share how you trasitioned into live trading. Were these problems you had in demo as well?
I really recommend using visual mode for testing a new Ea, you can slow it down or speed it up and really see if your ea is behaving as you think it is, it is very common for an ea to deliver decent results and then you find out on closer inspection that it is not actually doing what you have coded it to do.
Another advantage of doing painstaking visual backtesting is it will often show weaks point in your Ea which you can then tweak(Or strong points to build on!!)
HTH
Keith
Thanks, BarrowBoy, et al. My trades that were closed early did not use a trailing stop. I did, however, have the stop move to breakeven once the position was favorable by a certain amount. Sometimes it worked. Other times, the stop would be moved to breakeven when the price has barely moved; and then I would be stopped out of a profitable trade. The very next trade it may work properly. Some trades would be skipped. Others would enter several bars off. I was not expecting a perfect experience, but my experience with MT4 so far makes the strategy untradeable because of lack of consistent execution.
I have used similar strategies for several years but I am new to MT4. I used Tradestation a couple of years ago and the signals worked as expected on futures. (Among other things, slippage was a problem there) I have been with FXCM since 2003 and they only recently implemented MT4, so I have been trying to get up to speed on the automation. I may have to try to connect using the API because MT4 does not seem reliable enough for me to work with. I would love to use it. I am okay with the spreads, the historical data and other issues. I just need it to consistently execute what is coded. How are your experiences with it live? Do your strategies do what they are coded to do 90% of the time?
Ok, thanks a lot for all the comments. I will definitely run it for couple of months before the real action.
I will try to optimize for better performance without the larger drawbacks. It has fixed profit and loss orders. Its trading from the whole market hours but there is a check for larger spreads so in case the spreads goes wild eg. bigger than 3pips it stops trading.
I will post new results again soon.
Once again thank you all for the comments, its really helpful and it gives me a good perspective.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Hello, I am quite new to this forum, but i've been here a lot in the last couple of weeks to read through the Documentations and other user post for help when I was writing my own EA code.
Well, here is the question. I wrote this code and like to ask you what is your experience with the backtesting feature in MT4? I mean, when you wrote yours and did the test and then run it live, was it working as expected?
Here are few printscreens of my test. Its basicaly developed for EURUSD 5min chart and backtested since Januare 1st till now. I tried to run the tast for the past five years but it didnt show any results and said NA.. So not sure why is that. I checked the history and have it since 99. Anyway, the lots are not optimazed so its keeps trading the same amout (1lot). I was even happy two achive write my strategy so it would work. I will probably spend few more hours trying to optimize it for the trading size so it gradually increase or decrease based on the account value.
I tested on $2000 account start. The report looks a bit confusing since my broker uses 5 decimals prices, so I had to write the code in 10times incremetal then normal.
Plase let me know what you think of the results and any commets would be much appriciated. I'm thinkink of opening live account and let the script do the magic.. Do you think that is a good or bad idea?
So, here are the screens: