@Paris – I agree with your %’s, a truly successful system is more down to robust scaling, risk and money management rules rather than just purely the entry conditions. For the time being any system that avoids a zero bar closing is a good start until PRT resolve the issues which as Nicolas mentioned earlier, they are working on which is great news. I am also working on adopting the strategy to 1s – 1m timeframe and will post some views as soon as I’ve finished testing.
my
| papertrade |
|
|
|
|
|
|
|
|
|
|
|
|
|
| Date |
Type |
Price |
Qty |
Value |
Brokerage fee |
Entry date |
Exit date |
Type |
Nbr Bars |
Abs Perf |
Relat Perf(%) |
Brokerage fee |
| 6/02/2017 9:05 |
Buy (trailing) |
11,596.50 |
1 |
57,982.50 |
0 |
|
6/02/2017 9:05 |
6/02/2017 9:05 |
Short |
0 |
54 |
0.09% |
0 |
| 6/02/2017 9:05 |
Sell (entry) |
11,607.30 |
1 |
58,036.50 |
0 |
|
|
|
|
|
|
|
|
| live trade |
|
|
|
|
|
|
|
|
|
|
|
|
|
| Date |
Type |
Price |
Qty |
Value |
|
|
Entry date |
Exit date |
Type |
Nbr Bars |
Abs Perf |
Relat Perf(%) |
| 6/02/2017 9:06 |
Buy (exit) |
11,601.30 |
1 |
58,006.50 |
|
|
6/02/2017 9:05 |
6/02/2017 9:06 |
Short |
1 |
25 |
0.04% |
|
| 6/02/2017 9:05 |
Sell (entry) |
11,606.30 |
1 |
58,031.50 |
|
|
|
|
|
|
|
|
|
results today, see the different between live and paper
I’m still trying out various strategies
@Jonjon on your old post
From the look at your screenshot, you are using the DAX Future contracts. That is not the same as DAX indices. There could be variation on these two markeds. I know it’s an old post but I just thought I would mention it.
Cheers Kasper
CNParticipant
Senior
@Raul, Wow, looking promising 🙂
Post as soon as you see fit.
AlcoParticipant
Senior
Wow raul, 1254 contracts!! Looks promising but also dangerous haha
CNParticipant
Senior
@alco, that’s just the limit of contracts for the autotrade
Hi all – providing an update on the issues discussed here. Nicolas, I have already sent a few bug reports to PRT per your email request and instructions yesterday and will continue to send if/when I find any more, am still going through comparisons. I have noticed though that some differences that I had between live and backtest last week are no longer present when I re-run the same backtest this week, so it could that the same day only issue mentioned earlier and/or maybe PRT have released a patch already ?
I also wanted to bring to people’s attention a limitation of using the manual code for trailing stop mentioned here #23495. If you are running this in a 5m TF then the trailing stop will only update once every 5mins as the conditions are checked once per bar – Nicolas, can you confirm that my understanding is correct ?
Hence it is probably not a suitable trailing stop to use for scalping as it doesn’t update quickly enough. What you really need is to run the strategy in a shorter period ie 1m or 1s. I have coded this in a few strategies and am running these in parallel live this week and the results so far confirms my views. I will post the findings here when I have a week’s worth of data as two days is not enough to draw conclusions on the workings of the system. So we are making some progress on narrowing down what we can/cannot use for our strategies discussed here in this topic.
Anyway, this was just a heads up to anyone using that trailing stop code on a longer timeframe, please be aware of it’s uses and limitations.
Hey manel, I’m not aware of any patch already. I’ll ask PRT tomorrow and let you know.
About the trailing stop code, you are right, like any other part of the code, it is only read and executed only one time per bar. That’s why I talked about move the strategy to the lowest timeframe possible!
raul,
quatro de sucesion de gano hoy dia ! (sl 5 tp5)
maravilloso
@manel
No patch has been applied already. So the backtests behaviour you got is still the same as the first tick/tick backtest engine public version, released weeks ago.
Screenshot 1, backtest, screenshot 2, real.
Screenshot 1.1. Backtest, screenshot 2.2. real.
Similarities appear, in some days it differs. But more or less remain more or less the same
Thanks Nicolas, the change in results must have been due to the same day issue that has already been identified.
