Has anyone made Automatic Virtual Self-Optimization for their robot? - page 11

 
Petros Shatakhtsyan:

I haven't seen any concrete real examples from anyone.

I wasn't looking hard enough. Or do you have to do a tap dance to pay attention?
 
Petros Shatakhtsyan:

Real ticks on the tester are the same ticks that are fed to MT5. It depends on the broker and in some cases the tester result on real ticks is the same as the real trade.

Even if the "broken" ticks were real ticks that the broker traded on, it will not help in any way to study the possibilities of the investigated TS.

 
Petros Shatakhtsyan:

I understand that nobody can at least show the results (for all pairs) in the tester without self-optimisation and then with self-optimisation.

It is better to show the results in tabular form:

This will turn out to be a comparison of warm with soft. But doing self-optimization for any tst without sources is possible, of course.

As soon as the tst-format is opened, we will be able to see the equity graph and the trading history of the self-optimizing TS and other data of the Report. For now, only the final balance.

 
fxsaber:

Even if the "broken" quotes were real quotes that the broker traded on, it will not help to study the capabilities of the TS under study.

Actually, what difference does it make what quotes are? The robot should work everywhere and take into account any price movements. I think it doesn't even matter if the quotes are real or not.

 
Petros Shatakhtsyan:

The robot should work everywhere and take into account all price movements. I think it doesn't even matter if the quotes are real or not.

It cannot take them all into account, it is too complicated algorithm to take into account all possible quotes, even if they are not real. Rather, the robot should be stable on quotes that have a probability distribution similar to the real market, with a deviation of X%.

 
Petros Shatakhtsyan:

Actually, what does it matter what quotes are? The robot should work everywhere and take into account all price movements. I think it doesn't even matter if the quotes are real or not.

I am referring to the investigation of TS's possibilities.

Forum on trading, automated trading systems and testing trading strategies

Has anyone done Automatic Virtual self-optimization for their robot ?

fxsaber, 2019.12.01 10:50

With such proper TSs the same problem keeps coming up during research. Cosmic profits on the real history of some short sections. These profits are not systematic as there is a negative spread veiled there. As a result, you optimize such an Expert Advisor and see a perfect result. When you look, you see that half of the profit for the month is calculated within several hours.

I have to cunningly arrange filters for super profitable parts of the history to avoid this. I am inclined that these are broken quotes, because it is unlikely that lag arbitrage took place for real profits.

Apparently, I'm not making any sense at all.

 
Maxim Romanov:

Generally it is not possible to account for everything, it is too complicated an algorithm to account for all possible quotes, including non-real ones. Rather, the robot should be stable on quotes that have a probability distribution similar to the real market, with a deviation of X%.

This is the case, which is why I showed here how the results differ from each other. And that's why I think the only way out is to do self-optimisation frequently on different pairs.

But the frequency and depth of optimisation should depend on the trading strategy.

 
fxsaber:

It's about researching the TC's capabilities.

Apparently I'm not making any sense at all.

What does arbitrage have to do with it ? I only use trend and support/cop levels when trading. I'm not really bothered by negative spreads as well as their widening and slippage,execution time.

Each strategy has its own peculiarities.

 
Petros Shatakhtsyan:

What does arbitration have to do with it?

It has nothing to do with arbitration. We are from different worlds. I apologise for the off-topic.

 

fxsaber, 2019.12.01 10:50

With such correct TCs, the same problem keeps cropping up during research. Cosmic profits on the real history of some short sections. These profits are not systematic, because the negative spread is veiled there. As a result, you optimize such an Expert Advisor and see a perfect result. When you look, you see that half of the profit for the month is calculated within several hours.

I have to cunningly arrange filters for super profitable parts of the history to avoid this. I am inclined to think that these are broken quotes, because it is unlikely that lag arbitrage took place for real profits.

If you mean what's marked, that's fine too. That's what can happen with me when the lot was increased and after that there is a strong movement towards the plus.

In this case the position is not closed until the price shows clear signs of coming back. Is there anything unusual here?

However, in order to eliminate such results during self-optimization, I also take into account the number of trades. I choose the one with the most trades.