Help with backtester reports
Ron, the v176 simply does not work with backtesting the grid.. i have the same thing, as do several others who are testing different strategies.
I have looked but have no idea where the problem is.. it *may* be with limit orders. The back tester works better on shorts than longs.
All this is a bit of a mystery!
I have looked but have no idea where the problem is.. it *may* be with limit orders. The back tester works better on shorts than longs.
All this is a bit of a mystery!
Backtesting doesn't work with longer-term timeframes on v176. I know it doesn't work with H4 and it most likely does not work with Daily either. I guess that's why they hid W1 and MN from the "Period" options in backtesting.
The shorter the timeframe you go, the more accurate, but more riskier you get. Then, there's the lack of adequate data for shorter timeframes. Adequate data for backtesting means a MINIMUM of 5 years of data.
The whole scenario has put me into a quagmire. I spent a month using MT4 around the clock, and I have yet to open a real forex trading account because of the numerous bugs I see.
Yes, they can say it's free and all... and I love that they're doing it... But... If I'm going to risk my non-free (earned) money on it, I will just not do it. I make too much in the stock market to be willing to take a risk in forex when I see problems. I'm better off making adequate gains in the stock market.
Overall, I love what MT4 is trying to do.... They just haven't gotten there yet for me to go live with one of their brokers.
The shorter the timeframe you go, the more accurate, but more riskier you get. Then, there's the lack of adequate data for shorter timeframes. Adequate data for backtesting means a MINIMUM of 5 years of data.
The whole scenario has put me into a quagmire. I spent a month using MT4 around the clock, and I have yet to open a real forex trading account because of the numerous bugs I see.
Yes, they can say it's free and all... and I love that they're doing it... But... If I'm going to risk my non-free (earned) money on it, I will just not do it. I make too much in the stock market to be willing to take a risk in forex when I see problems. I'm better off making adequate gains in the stock market.
Overall, I love what MT4 is trying to do.... They just haven't gotten there yet for me to go live with one of their brokers.
Hi, Hugues, Vooch,
Yes, it's a real bummer. I'd hoped to be able to get at least some idea (and I think I did, actually, in the previous version), but this new version seems to completely lose track of open orders at the end. I haven't closely inspected the orders executed to see if they're reasonable either, so that remains to be done.
As to short-timeframe risk, that's not necessarily a problem for grid-based systems, because they operate off price only. The only problem then is the length of data, which isn't really enough at that level. You can tell, though, that a grid system is ripping the guts out of your account in less than a month of data, so you can still strip out the really bad ones; the opposite isn't true. Survival past the data you have doesn't guarantee viability, as of course we all know.
If you're a serious trader, well capitalised, there's the Oanda API which is free after a certain monthly volume. You have to program your system yourself in C++ or Java, though, and spreads, while generally lower at Oanda, aren't constant, so backtesting what would have happened there can be a problem, especially if you use tight stops. Trailing stops of course might behave differently than expected as well.
RW
Yes, it's a real bummer. I'd hoped to be able to get at least some idea (and I think I did, actually, in the previous version), but this new version seems to completely lose track of open orders at the end. I haven't closely inspected the orders executed to see if they're reasonable either, so that remains to be done.
As to short-timeframe risk, that's not necessarily a problem for grid-based systems, because they operate off price only. The only problem then is the length of data, which isn't really enough at that level. You can tell, though, that a grid system is ripping the guts out of your account in less than a month of data, so you can still strip out the really bad ones; the opposite isn't true. Survival past the data you have doesn't guarantee viability, as of course we all know.
If you're a serious trader, well capitalised, there's the Oanda API which is free after a certain monthly volume. You have to program your system yourself in C++ or Java, though, and spreads, while generally lower at Oanda, aren't constant, so backtesting what would have happened there can be a problem, especially if you use tight stops. Trailing stops of course might behave differently than expected as well.
RW
Just thought I would throw in my 2 cents worth on this topic.
Since Aug last year I have had the same expert running on demo mode on one account and since feb of this year I have had the same expert running live on a real account, Thus I have 100% accurrate results of what my expert does........The expert uses a weekly chart...
I cannot simulate these results on backtesting, the closest I can get is using a 30min chart using the arraycopyrates function in my expert for my weekly levels....
All higher time frames are way out, even if you use the same start date.....
Maybe it is my lack of coding skills but it appears the only true way to get accurate results is to demo test live..........
GT to all.........................
Since Aug last year I have had the same expert running on demo mode on one account and since feb of this year I have had the same expert running live on a real account, Thus I have 100% accurrate results of what my expert does........The expert uses a weekly chart...
I cannot simulate these results on backtesting, the closest I can get is using a 30min chart using the arraycopyrates function in my expert for my weekly levels....
All higher time frames are way out, even if you use the same start date.....
Maybe it is my lack of coding skills but it appears the only true way to get accurate results is to demo test live..........
GT to all.........................
yes mpfx, u are basically right... but one hopes the backtesting is sufficiently close to what really happens for the basic conclusions to be valid.
this saves a lot of time.. forward testing is slow.. even if fun!
this saves a lot of time.. forward testing is slow.. even if fun!
mpfx,
Yes, that's interesting. I'm with hdb, though, in that I don't really mind if the results are off by a bit. I'm not going to be investing in a system that's only marginally profitable anyhow, so +- a bit wouldn't hurt me.
I just want to be able to say, with some certainty that this system is either a big loser, marginal or good.
(MQuotes - n.b. without open equity I have no way of telling that.)
Yes, that's interesting. I'm with hdb, though, in that I don't really mind if the results are off by a bit. I'm not going to be investing in a system that's only marginally profitable anyhow, so +- a bit wouldn't hurt me.
I just want to be able to say, with some certainty that this system is either a big loser, marginal or good.
(MQuotes - n.b. without open equity I have no way of telling that.)
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
At least before in the initial version you could see the almighty plunge at the end when the backtester closed out all open trades, however now it seems to just silently forget them. This is disastrous for backtesting systems that always have a number of open trades. The 'grid' style systems always look wonderful in closed equity, but when you add in the open trades (which are mostly losses) they're usually _big_ losers. I'd really rather not have to write my own script to parse and make sense of the backtester reports - couldn't we have something that tells about open equity, or at least the sum of open trades?
Or am I missing it somehow?
RW