Some signs of the right TCs - page 33

New comment
 
fxsaber:

Different, but probably close in value. I can only guess now, as the data is no longer available. I am not prepared to repeat the experiment.

If different, the accuracy to the error is not clear to me too. If they are the same, everything is correct - a symmetrical system finds the same input parameters in all three passes. And correctly understood,BestProfit_OnlyBuy and BestProfit_OnlySell are not equal and should not be equal?

 
Valeriy Yastremskiy:

And correctly understoodBestProfit_OnlyBuy and BestProfit_OnlySell are not equal, and should not be equal?

Of course, they should not be equal. At least due to the fact that GA was made.

 
fxsaber:

Certainly not equal. At least for the reason that GA was done.

The GA will find the same parameters on the parameter space if the function or dependence of the function on the parameters is sufficiently uniform. And with correct mathematical logic. If GA will find different best points in the parameter space depending on the behavior of the series or the direction of the trade, it is not good for the Expert Advisor.

 
Valeriy Yastremskiy:

The GA will find the same parameters on the parameter space if the function or the dependence of the function on the parameters is sufficiently uniform. And with correct mathematical logic. If GA will find different best points in the parameter space depending on row behaviour or trade direction, it is not good for the EA.

Optimal is not a mathematically correct TS. Because it was expected to always opt for each direction separately.

Even GA re-runs do not have to coincide. Imagine a function with two strongly pronounced local maxima. Then either one or the other result will be obtained depending on how the randomness lies.

 
fxsaber:

Opted not mate the correct TS. As it was stated that each direction must always be opted for separately.

Even GA re-runs do not have to coincide. Imagine a function with two strongly pronounced local maxima. Then either one or the other result will be obtained depending on how the randomness lies.

GA is an algorithm. For the asymmetric logic of TC by direction of transaction it is necessary to opt for each direction. But I prefer the symmetric logic and the result of identical passes depends more on symmetry than on mathematical correctness. A lot of highs or their brightness indicates that the Expert Advisor is unstable.
 
fxsaber:

Opted not mate the correct TS. As it was stated that each direction must always be opted for separately.

Even GA re-runs do not have to coincide. Imagine a function with two strongly pronounced local maxima. Then depending on how the randomness lies, we will arrive at either one or the other result.

Do not forget that it was said about the general case, if you run on other instruments, onli_bye and onli_sell pairs will most likely often be quite different and therefore the results will not coincide. Even on such a TS with a claim to universality.

 
Aleksey Mavrin:

Don't forget that it was also said that this is a general case, if you run it on other instruments, the onli_bye and onli_sell pairs are likely to be often quite different and therefore the results will not coincide. This is why the results will not coincide even in such a TS with a claim to be universal.

Forum on trading, automated trading systems and strategy testing

Some signs of a proper TS

fxsaber, 2020.03.06 22:12

In general, in this particular case there is no point in setting up in each direction.

 
You used to use the zigzag in some strategies. Do I understand correctly that you are now trying to abandon it, as it uses the pips step?
 
traveller00:
In the past you have used a zigzag in some strategies. Am I correct in assuming that you are now trying to abandon it as it uses the pips step?

Forum on trading, automated trading systems and testing trading strategies

Some signs of good TS

fxsaber, 2020.03.02 08:09

TS input data: two numeric discrete series. Not one, but TWO: bid/ask.

ZigZag is a mathematical transformation with these properties:

  • Local extrema are retained.
  • Repeated application does not change the numerical series.
Also, ZigZag is good in that it is able to convert TWO series into one. If a series consists only of transcendental numbers, the ZigZag will work on it as it does on others. It is a purely mathematical conversion.
 
I was just confused by the fact that in the next thread, where an example of a correct strategy is given, the half-sum of bid and ask is taken rather than the zigzag, so I thought I'd clarify.
1...262728293031323334353637
New comment