Inconsistent results before/after ending of DST in Brazil.

[wemersonrv]

Hi,

First of all... Sorry for my poor english... it's no my language... use Google translate to help!

I always see that when there is DST changing here on Brazil the backtest results changes... then i wait for the end of our daylight saving time here in Brazil. Last february 20th was the last day of our daylight saving time.

Then at february 20th i did my last download and the result was the same as the other 10 times i've make the download and test, from january 1st till now.

- Download date: 02/20/2016 ( still in DST till midnight)
- Timeframe: H4
- Period: 01/01/2015 - 12/31/2015
- Results of my backtest: from $ 10.000 to $ 73.479,69 in all 2015.


After at february 21th eu deletei todos os dados e repeti o download. And for my surprise the result was totally different

- Download date: 02/21/2016 ( after DST end )
- Timeframe: H4
- Period: 01/01/2015 - 12/31/2015
- Results of my backtest: from $ 10.000 to $ 34.789,12 in the same period.


Can anybody tell me why this happens ? The same data downloaded before/after DST end produced very different results

[tickstory]

Hi Wemersonrv,

Some of the translation wasn't fully clear so apologies if I've missed something. Unless you deleted your Dukascopy data between the 20th Feb and 21st Feb, then the data should be the same. This means the only variables would be the parameters you used (spread can always cause marked differences).
If you are using two separate copies of the data (i.e. two separate Dukascopy store folders), then you may want to do a chart comparison so see if the data has changed. Dukascopy of course have the capacity to change their historical data but I imagine they would only do this if they were improving the accuracy of the data.

Hope this helps.

[wemersonrv]

Hi Wemersonrv,

Some of the translation wasn't fully clear so apologies if I've missed something. Unless you deleted your Dukascopy data between the 20th Feb and 21st Feb, then the data should be the same. This means the only variables would be the parameters you used (spread can always cause marked differences).
If you are using two separate copies of the data (i.e. two separate Dukascopy store folders), then you may want to do a chart comparison so see if the data has changed. Dukascopy of course have the capacity to change their historical data but I imagine they would only do this if they were improving the accuracy of the data.

Hope this helps.

No i do not change parameters. It's the same! The parameters are saved as default... and i always check them with the model below to see if are fine!



The only thing i can imagine is difference because DST changing..

See an example on the result below:


Previously it's gave me a net profit of $ 23688.11 and now it's only $ 15411.11... It's good too but far away from the older result. And i don't know in what result can i trust!

[tickstory]

Hi Wemersonrv,

Are you saying that you did the following:

1) Back-test with 2015 data.
2) Delete your Dukascopy store folder.
3) Download all of 2015 again.
4) Back-test again with 2015 data and get different results?

Are you stll able to reproduce the problem and if so, could you confirm what setting/differences you are making to get the two different EA results?

Thanks.