Download MetaTrader 5
To add comments, please log in or register
Sell your CPU capacity and earn money!
wouter
48
wouter 2014.03.06 02:40 
Hello,

How can i make backtests with 100% modeling quality ...?


Kind regards,
Wouter vermeersch
Simon Gniadkowski
Moderator
17963
Simon Gniadkowski 2014.03.06 09:20  
wouterv:
Hello,

How can i make backtests with 100% modeling quality ...?

Open the .fxt file in a binary text editor and change the modeling quality value to 100 . . . the values are pretty arbitrary, even the 99% in tick data generated .fxt files is just a meaningless number. Don't get too hung up about it . . .
wouter
48
wouter 2014.03.06 13:50  
RaptorUK:
Open the .fxt file in a binary text editor and change the modeling quality value to 100 . . . the values are pretty arbitrary, even the 99% in tick data generated .fxt files is just a meaningless number. Don't get too hung up about it . . .

Hello raptor, thank you for you're swift response. If I backtest i have mostly 2 different outcommes: 1M TF: 25% moddeling quality 5M TF true daily TF : 90% ... Is it correct of me to think the higher timeframes are more relaiable to test my strategie on and are more comparable with the reality. I know that backtesting can't be compared with real live trading environments, due to the lack of latancy, spreadfluctuations, and server processingspeads. But what would be the most relaiable in you're opinion 1MTF, 5MTF -> daily or are they just same relaiable in what we can call relaiable (excluding real trade environment factors)
Simon Gniadkowski
Moderator
17963
Simon Gniadkowski 2014.03.06 16:32  
wouterv:
Hello raptor, thank you for you're swift response. If I backtest i have mostly 2 different outcommes: 1M TF: 25% moddeling quality 5M TF true daily TF : 90% ... Is it correct of me to think the higher timeframes are more relaiable to test my strategie on and are more comparable with the reality. I know that backtesting can't be compared with real live trading environments, due to the lack of latancy, spreadfluctuations, and server processingspeads. But what would be the most relaiable in you're opinion 1MTF, 5MTF -> daily or are they just same relaiable in what we can call relaiable (excluding real trade environment factors)
It depends on how your strategy is meant to work, personally I do not like any timeframe, for each bar you get 4 values OHLC, what about all the rest between Open and Close ? you miss all that data, why is it less significant ?
Ian Venner
2583
Ian Venner 2014.03.06 16:47  

Quality value is just a reflection of how many ticks were modelled to fill in the gaps between the OHLC. The OHLC can be considered "real Prices" as issued from the broker, we know the prices in between were also issued by the broker because they had to be, we just dont have an explicit record of them so they are "modelled" by the tester. The Higher timeframes use OHLC prices from lower time frame bars. This means the 1hr timeframe has 240 "real" OHLC prices from the 1M bars to use so less ticks need to be modelled. That is the only reason the 1hr timeframe has a higher quality rating.

The all important factor is to have 1M chart data to cover the entire period of higher timeframe testing.

/
To add comments, please log in or register