tickstory backtesting results are the opposite with broker data

General discussion about the Tickstory Lite software package.
Post Reply
tickwaiting
Posts: 72
Joined: Wed Sep 06, 2023 11:27 pm

tickstory backtesting results are the opposite with broker data

Post by tickwaiting »

Hi Tickstory

i integrated my expert advisors from mt4 to mt5 just to add final layer of backtesting results validation.

things were smooth on mt4 here are my approach:

1- download tickstory data (10 years)

2- backtest symbols on mt4 (3 years)

3- out of sample forward testing (7 years)

4- apply mt4 results settings on mt5 using broker 10 years data (100% history quality)

5- everything went south the results went from winning to losing


now i'm stuck and confused why is this happening? any explanation behind this? any solutions?


thanks
tickstory
Posts: 5025
Joined: Sun Jan 06, 2013 12:27 am

Re: tickstory backtesting results are the opposite with broker data

Post by tickstory »

Hi Tickwaiting,

There could be many areas where your transition could have differences. For example, was the EA converted from MT4 to MT5 completely and correctly? Are the settings the same in both terminals (starting equity, instrument digits, etc)? Did you test with fixed spread on MT4 versus variable spreads in MT5?

Of course, do remember that all EAs will be affected to a degree by any change in input data - especially if your EA is particularly sensitive to such changes (scalping strategies for example) (see here and here for some additional commentary).

Your best approach is to take the first few trades of your strategy in both MT4 and MT5 and do a comparison to see why there are differences and investigate from there. This process will also help you understand any weaknesses in your code and/or strategy before your go live.

Hope this helps.
tickwaiting
Posts: 72
Joined: Wed Sep 06, 2023 11:27 pm

Re: tickstory backtesting results are the opposite with broker data

Post by tickwaiting »

hi tickstory

sure i will, but how about a major difference between broker and tickstory which is the market data.

i decide to backtest based on broker data, but the question is broker data enough? in other words do i need tickstory tick data knowing broker provide tick data.

other thing i noticed the spread on mt5 is not fixed or optional on strategy tester like mt4 but more it's variable on mt5. so is there a way to control the spread?

thanks
tickstory
Posts: 5025
Joined: Sun Jan 06, 2013 12:27 am

Re: tickstory backtesting results are the opposite with broker data

Post by tickstory »

Hi Tickwaiting,
i decide to backtest based on broker data, but the question is broker data enough? in other words do i need tickstory tick data knowing broker provide tick data.
Historical broker data is usually not as complete as the data provided by Tickstory. That is, it could contain gaps or missing days. You would need to do a deeper analysis of the data if you wanted to know more. Tickstory data is more likely to represent the live data you are going to be getting from your broker. That said, you should always seek to understand why there are differences between your two back-tests as it represents a potential vulnerability in your trading strategy.
other thing i noticed the spread on mt5 is not fixed or optional on strategy tester like mt4 but more it's variable on mt5. so is there a way to control the spread?
You can manually create a fixed spread by simply exporting data from Tickstory with the bid and ask price set, for example:

Code: Select all

{Timestamp:yyyyMMdd},{Timestamp:HH\:mm\:ss},{BidPrice},{BidPrice+0.0001},{BidPrice},{BidVolume:F0}
There may be other ways to set the spread in MT5, however your best source for more information on that would be the MT5 manual and/or forums.

Hope this helps.
tickwaiting
Posts: 72
Joined: Wed Sep 06, 2023 11:27 pm

Re: tickstory backtesting results are the opposite with broker data

Post by tickwaiting »

ok thanks tickstory
Post Reply