Signs of a REAL system - page 25

New comment
 
ForexTools писал(а) >>

there were no hints, i wrote - it was just a reason to think about what a wrong system is ;)

that's it, isn't it? :))) look at the stack of numbers and give me a guarantee that I'll make a lot of real money :))))

i did not in vain called the topic signs of a wrong system. there are things that are very obvious will ruin any system. but if none of them is not, it does not mean that the system will surely give at least 90% profit. besides, it's been discussed 100 times that the tester, demo and real - are three big differences, they only look the same (like all the chinese we look alike, but they are all very different people).

But from personal experience the sequence of actions is as follows:

1) test in the tester different options / currencies / timing to exhaustion - if the number of satisfactory results more than negative go to paragraph 2

2) Open a couple of demo accounts for different currencies / timeframes, connect an EA to all of them and check - if the results are not worse than in the tester, then continue with step 3

3) open a microreal. to check the possibility of losing faster - with an wrong MM you quickly run into a mistake, that will destroy your little deposit. this "bleeding" is absolutely necessary to learn to value your money. if there was no bleeding, or its lessons are taken into account and checked again along the whole chain 1-2-3, then - go to step 4

4) open a real, put it on the monitoring, bragging about the results and looking for investors persuaded by these results.

Good luck ;)

i agree! this is exactly the sequence i'm using now!

Just a little nuance from the 2nd point: my system is trading with no more than one order, so to test it with different pairs and different timeframes I'll need corresponding number of constantly working computers with internet ...

 

oleg_slk писал(а) >>..... понадобиться соответственное число постоянно работающих компов с инетом...

Can't the right number of computers be replaced by the same number of terminals on the same computer?

 

oleg_slk писал(а) >>

so to test on different pairs and different timeframes you need a corresponding number of constantly working computers with internet...

Why computers?! You can just run several different copies of the terminal on the same computer

 

Can we look at the code to make it all live in one terminal?

 
ForexTools >> :

From personal experience, the sequence of actions is as follows:

1) we test different variants/currencies/times in the tester till exhaustion - if the number of satisfactory results exceeds the number of negative ones, then proceed to point 2

2) Open a couple of demo accounts for different currencies / timeframes, connect an EA to all of them and check - if the results are not worse than in the tester, then continue with step 3

3) open a microreal. to check the possibility of losing faster - with an wrong MM you quickly make a mistake, that will lead to a loss of your small deposit. this "bleeding" is absolutely necessary to learn to value your money. if there is no bleeding, or its lessons are taken into account and checked again along the whole chain 1-2-3, then - go to step 4.

4) open a normal real, monitor it, brag about the results and look for investors, convincing them with these results.

Good luck ;)

signs of an incorrect system:

1) test different options in the tester / currencies(martingale) / times(never mind) to the point of exhaustion - if the number of satisfactory results more than negative - go to step 2

2) open a couple of demo accounts for different variants of currencies(martingale) and connect the EA to all of them and check - if the results are not worse than those in the tester, then continue with step 3

3)...

4)...

Indeed, one can only wish you good luck!

 

Thanks, appreciate it :)

We all really need it!!!

 

FOXXXi писал(а) >>

(martingale)....times(no longer important).....

don't get it :( what do the amendments mean???

 
Sergey Kravchuk:


Signs of fitting:

  • The tester radically changes the balance growth curve and the result obtained by changing just a little bit of a single parameter
What if the system is written in such a way that 1 parameter is decisive, i.e. SL/TP?
  • In the tester, the balance growth curve and the result changes radically when you change the test method (all ticks/checkpoints/opening prices).
It's a mystery why)
  • stop and take parameters are not calculated, they are hardcoded with specific numbers.
What if there is a trailing system and stop loss is fixed there?
  • Trading signals are calculated using formulas from external neural networks or AI (usually they are kilometre formulas with ratios of a dozen decimal places, like 1.28677263556*Open[i]).
That's right) AI will make chaos out of chaos
  • Uneven balance growth. I mean when 60%-80% of profit made by 2-5% of trades
Nobody cancelled the mechanism of probing the market
  • Svinozavr: dramatic difference in the results with different quotes history (from different brokerage companies).
How is it possible?
  • niko1312: Difference in results on different history intervals (for the same brokerage company).
It is undoubtedly) Taking into account that we may trade with trends and others with flat prices.
  • Goldtrader: the results on another timeframe are very different (for the worse) relative to the basics,
This is a good system for any market, timeframe, asset, but with subtleties on the volatility side.
  • Goldtrader: the results on another similar (for example GBPUSD after EURUSD) trade instrument is very different (worse) with respect to the base instrument,
We should dig into the correlation of pairs and it is definitely not a divergence
  • Thereis a large number (more than 2-3) of optimizable parameters in the Expert Advisor.
The code can be used to adjust or remove the settings. The main thing is that they fit together well


Purely my opinion, nothing more

1...1819202122232425
New comment