Is there a logical explanation as to how this happens with my EA testing....

Hi

i have an indicator based EA that uses a maximum stop-loss size, and has to values/parameter lengths for the one indicator, this is all the parameters.


When i optimise the indicator parameters, i select/save the optimum/most profitable values, then run the backtest based on these values.


Without doing anything else, i re-optimise the new optimum indicator values. Sometimes, this leads to a new pair of optimum parameters, with a different profit/loss level over the H4 6 month backtest . I may then save these optimum values, run the backtest with these new optimum values, then re-optimise these values, and the latest optimisation may produce some new optimum values .
How can this be so, when i haven't changed any other setting in between . Once the indicator parameters have been optimised once, these should be the "optimum" parameters. Re-optimising these indicator values should return the same indicator values ALL the time.
Is there a logical/systematic/rational/scientific explanation of this problem i have encountered??

Many thanks.

What's your backtest quality?
Are you recalculating pip values or utilizing previously generated ticks?

Thanks tesla .

I simply optimise the 2 indicator lengths (1 indicator *2 lengths), set the optimised values, run the backtest, then re-optimise the indicator, and it sometimes comes up with a new set of optimum indicator parameters.

i generally backtest & optimise using "open prices only", as the "every tick" method produces the exact same results. I figure this is due to the simplicity of my EA, only using an SL and 1 indicator. Therefore the ea doesn't need to look inside each H4 bar (although i recognise that it would need to look at the lesser timeframes within the H4 bar if the ea used for example a trailing stop-loss)

How does the modelling quality affect things, and how would it come into play here?

Cheers.