What is the optimum depth of history for identifying a useful signal? - page 12

 
ZaPutina:
How about the depth, if you write 27 points, I do not see where it is - this calculated depth in your posts. Do you see it?
27 points is the profit obtained (on the minutes at your request, once again I repeat on the minutes) by the method of depth selection for today
 
azfaraon:

"So you think that any Expert Advisor can work profitably if the depth of history on which the optimization is performed is defined correctly? "Yes, according to the depth method it is.

"What's the criterion used for defining the time for new optimization? The depth method itself.

"If you choose the optimal history, you don't need periodic optimization, and the Expert Advisor will work profitably for many years in a row"? You don't need periodic optimization, but it will work as long as the depth selection method determines.

Well, if by "works so long" you mean the final period of time, then you probably need a new optimization after that.
 

So you are saying that in this case to get the following

" optimization parameters are changing slower than at the moment, on M1 period, with intraday trading"?

was the current day's history enough for you, just half a day's worth of minute history?

Was the optimization performed on this half-day history? If so, the sense of such optimization, if you have to do it on every bar, and the maximum profit will jump much, stronger than the possible profit or loss when changing the stop and take obtained during the optimization and applied in the current trade here and now.

 
Yes...I'll write down the stops and takes and give you an exact answer for today's profitability
 
khorosh:
If by "works as long as" you mean a finite period of time, then after that you probably need a new optimization after all, don't you?
Do you want perpetual motion?)))
 
ZaPutina:

So you are saying that in this case to get the following

" optimization parameters are changing slower than at the moment, on M1 period, with intraday trading"?

was the current day's history enough for you, just half a day's worth of minute history?

Was the optimization performed on this half-day history? If so, the sense of such optimization, if it has to be done on every bar, and the maximum profit will jump much, stronger than the possible profit or loss when changing the stop and take obtained during the optimization and applied in the current trade here and now.

What a strange habit )))), first you write an answer, then you add your own ...))). Of course everyone has a choice ... I wrote it to find out if someone works in this direction or not ...

I checked the system in all markets, but I was more surprised by the FB of the method...

 
ZaPutina:


So what is the optimum depth of history to analyse. I have my own opinion, but would like to hear it.

If for quotes, two observations in history (one of them is the last price of the current bar) will be enough to predict the future price direction with a positive expected payoff.

See Theorem about memory presence in random sequences

 
<br / translate="no">
khorosh:
Well if by "works for so long" you mean a finite period of time, then after that a new optimization is probably needed after all?
You can take the whole history since 1999 of the Eurobucks and choose the depth according to the method. Then it will probably work for a long time. But you have to understand the risks (I mean the size of stops and takes, takes are always bigger than stops) and whether you have the patience to wait...
 
Reshetov:

If for quotes, two observations in history (one of them is the last price of the current bar) are enough to predict the direction of quotes in the future with a positive expectation.

See Theorem on memory of random sequences

This is more of a postulate of technical analysis ...always ask two questions where the price is and why ...
 
ZaPutina:

In short, we probably do not understand each other. I just saw an analogy with volatility prediction, on the history parameters of take and stop are selected, and at certain calculations are predicted in such a way that at any arbitrary entry of the trading system my trade will be more than 50/50, that is, the volatility here and now will be less than the predicted size of the stop, Thus, volatility here and now will change slower than the predicted volatility, hence stops will expand (respond to changes) faster, thereby in places where there was a loss on the stop will be a drawdown and the final plus, or the final loss is smaller...

I do not understand the depth of history and I have not heard the answer.

As for the time being I haven't found any method of setting stops and takes without a chart.