Discussing How to Use Backtesting? article

[guangho]

The article is here:
viewtopic.php?f=31&t=3036

Sorry, because I can't send a new post, so only the questions here.
Excuse me, does not affect the strategy of running condition can be used in calculating the parameter "MVA" is how much? I want to try to quote "MVA" ---1000 average, can? I tried, failed.But Backtest Strategy is successful. Why? Is there any way to solve this BUG?

Thank you!!

[Valeria]

Hi guangho,

It is not a bug. Let me explain. Initially a strategy has 300 bars to work at. But your strategy has to use 1000 bars. As to Backtester, your strategy starts working after 1000 bars. The only difference between the work in Backtester and on the real chart is that in real time you have to wait while 1000 bars are formed (actually 700 (1000 - 300 loaded bars)), and Backtester loads them quickly. Unfortunately, there is no easy way to fix it.

[guangho]

Hi guangho,

It is not a bug. Let me explain. Initially a strategy has 300 bars to work at. But your strategy has to use 1000 bars. As to Backtester, your strategy starts working after 1000 bars. The only difference between the work in Backtester and on the real chart is that in real time you have to wait while 1000 bars are loaded (actually 700 (1000 - 300)), and Backtester loads them quickly. Unfortunately, there is no easy way to fix it.

Thank you very much for your answer.
I know what you mean. Whether through the establishment of "temporary data flow" to achieve policy reference beyond 300 K data outside the function?
Indicators and strategies similar to consider for me?
Thank you very much!

[boss_hogg]

Hello and have a nice trading week everyone

there is an issue after the latest TSII upgrade to ver. 01.13.092613: in strategy backtesting, there now is a maximum of 10.000.000 on the total parameters combinations.

Is there a serious reason behind this? The strategies I'm backtesting often reach 100s of billions combinations, so this new restriction actually prohibits proper backtesting (I can manually run a few variations of 10.000.000 chunks, but not 100s or 1000s variations!!!)

Honestly, what I need is not an answer but a fix to remove the restriction!

Regards,
Alkis

[Valeria]

Hi boss_hogg,

Thank you for the suggestion. I have forwarded your request to the developers.

[boss_hogg]

Hi again,

are there any news regarding the 10.000.000 limit? This issue is now more than annoying, it's important and it's urgent since it really makes it impossible for me to do proper backtesting!

Hi boss_hogg,

Thank you for the suggestion. I have forwarded your request to the developers.

[Valeria]

Hi boss_hogg,

are there any news regarding the 10.000.000 limit? This issue is now more than annoying, it's important and it's urgent since it really makes it impossible for me to do proper backtesting!

The restriction will be removed or changed in the future release.

[Cuchulain]

Hello all.

I backtest one of my strategy with the date between 05/2014 ans last week in 2016.
I have this amazing result following.

Even if there was strange market moving, and maybe not happend in the futur,
i am really surpise of this O_o
Does someone have something like this in this backtesting?

Regards
(Sorry i am french)



STATISTIQUES DE COMPTE
Gain/Perte total 8 227,45
Profit total de tous les trades 9 132,08
Perte totale pour tous les trades 904,63

Statistiques de solde :

Solde de depart 100,00
Solde final 8 327,45
Solde maximum 8 869,63
Solde minimum 95,20
Chute maximale de solde 542,96 (543,0%)

Statistiques de capital :

Capital final 8 323,84
Capital maximum 12 773,88
Capital minimum 86,43
Chute maximale de capital 4 468,33 (4 468,3%)

[Julia CJ]

Hi Cuchulain,

Please let me know which project and parameters you use, so we can check whether you have correct results.

Thank you

[Theresa]

Hi everyone,

Recently i am doing a backtesting using real ac backtest with FXCM TS2 platform. I was happy to see the result as it's very promising. So i started trading in real ac but unfortunately result wasn't as expected and also difference from the backtest result i have on the same date. I check and the time of entry and is mostly a like with some trade miss but entry pricing is a few pips away.
I found the simulator and try on simulator and was extremely sad as the result is worse than expected and also having the same problem in pricing.
Can anyone advice what could be the issue.. Is it the strategy or the backtest or the simulator or FXCM pricing..
Thanks