Please help: Spread

[SkystheLimit 986]

Hello,

I downloaded and imported the 1-minute data from FXDD and use it for my backtests.

I realized that no matter what I define for "spread" on the strategy tester, results are the same.

So I wonder whether all backtest results will be flawed and why exactly (fixed spread from the data?), so even I imported and use 1-minute tick data, this is not representative at all?

thx so much

[SkystheLimit 986]

Note:

I checked all trades visually via the respective option, the 1-minute candles look realistic and correct, the signals work correctly and the EA trades as it is supposed to be.

[Marco vd Heijden]

fractalfreak:

Note:

I checked all trades visually via the respective option, the 1-minute candles look realistic and correct, the signals work correctly and the EA trades as it is supposed to be.

Of course it will never be representative.

It's just good to see if the logic fires appropriately.

[[Deleted]]

fractalfreak:

Hello,

I downloaded and imported the 1-minute data from FXDD and use it for my backtests.

I realized that no matter what I define for "spread" on the strategy tester, results are the same.

So I wonder whether all backtest results will be flawed and why exactly (fixed spread from the data?), so even I imported and use 1-minute tick data, this is not representative at all?

thx so much

If EA works on first tick of bar than representative.

[SkystheLimit 986]

ok so.......

the imported price data is displayed correctly...................so let us also say that this is the "real price"........now.........the indis of the EA and the EA strategy itself work correctly...........

my clear question:

1. since the imported data has a fixed spread, changing the spread as an input to the strategy tester does not change anything, right?

2. what would be the spread?

3. does the strategy calculate all profits / losses based on one spread? if yes, which one?

So if I have a fixed stop of 5.5p, and I trade 1m per trade, the loss shown is USD 550..........so seems to be right.....

Thx for helping with precise info.

[Marco vd Heijden]

fractalfreak:

ok so.......

the imported price data is displayed correctly...................so let us also say that this is the "real price"........now.........the indis of the EA and the EA strategy itself work correctly...........

my clear question:

1. since the imported data has a fixed spread, changing the spread as an input to the strategy tester does not change anything, right?

2. what would be the spread?

3. does the strategy calculate all profits / losses based on one spread? if yes, which one?

So if I have a fixed stop of 5.5p, and I trade 1m per trade, the loss shown is USD 550..........so seems to be right.....

Thx for helping with precise info.

Only if you use bar open prices mode and you can simply Comment() the spread to check it's value while it's running.

[SkystheLimit 986]

the entry signals are intrabar.....

so I guess I cannot check it?

[Marco vd Heijden]

fractalfreak:

the entry signals are intrabar.....

so I guess I cannot check it?

Did you import every tick or just ohlc price data ?

If the latter then the bars are filled by the tick generating algorithm.

Please read: https://www.mql5.com/en/articles/75

[SkystheLimit 986]

Thx Marco...

I guess it is the latter when it comes from FXDD right?

[Marco vd Heijden]

fractalfreak:

Thx Marco...

I guess it is the latter when it comes from FXDD right?

I guess so.

Why not run it on a demo account to see what it does?

It's not as fast but the results will certainly be more reliable.

By the way you can do a few quick scans over all available bars in history in the OnInit() function to calculate best parameters.

It makes your robot adaptive, and it is a lot quicker then the endless hours of backtesting a specific group of people do because they are in believe that the result have some relationship to the results gained on real market (tick) feeds.

[SkystheLimit 986]

Marco vd Heijden:

I guess so.

Why not run it on a demo account to see what it does?

It's not as fast but the results will certainly be more reliable.

By the way you can do a few quick scans over all available bars in history in the OnInit() function to calculate best parameters.

It makes your robot adaptive, and it is a lot quicker then the endless hours of backtesting a specific group of people do because they are in believe that the result have some relationship to the results gained on real market (tick) feeds.

Thx Marco...

what do you mean with "By the way you can do a few quick scans over all available bars in history in the OnInit() function to calculate best parameters."?

Thx for explaining...