Martingale is not evil at all, it brings profits - page 10

New comment
 
lovova писал (а): But Feller is seemingly impossible to implement in MQL
You don't need to implement it there, no modelling is needed. Just take three parameters (p, q, r) and estimate by (7.7) the average number of deals, at which a series of losses of length r will be met for the first time. If it is comparable to the approximate number of deals the trader expects, then such a system should be discarded.
 

I liked the expression about him in the bookMathemat named after Larry Williams: "All my knowledge in mathematics is in a single fingernail", it's been a long time since I touched it.

 
Volodya, write me on the email, which is indicated in my profile.

2 Yura Reshetov: Testing MoneyRain from 14 August 2002 to 28 March 2008 (the only long piece, in which depo is not bumped to the floor, and the Expert Advisor does not swear because of a request to open an excessive lot) shows the following:


Total trades 751
Profit trades (% of total) 378 (50.33%)
Loss trades (% of total) 373 (49.67%)
Maximum
consecutive wins (profit in money) 10 (1115.40)
consecutive losses (loss in money) 10 (-954.22)


I left the parameters the same (but the initial depo is $0.5M, lots=0.1). Here the probabilities are almost the same as a symmetrical coin toss, and r=10. Looking at the table in my long post: 34.1 minutes, i.e. 2046 trades. You have this series of 10 consecutive losses appearing on 751 trades. So far no significant positive difference from the series of independent trades on the Bernoulli scheme has been noticed.


P.S. And your martingale is extremely aggressive, judging by your report. On the other hand, you can find such an r, at which the first occurrence of a series of losses of length r will be obtained on the average in a series of trades approximately equal to 751. This r equals approximately 8.5. But generally speaking, these estimations are of little importance for a very aggressive martingale, because a single trade not entering into any losing series can wipe out a considerable part of the deposit. The reason is that with this MM the trade is missing a reasonable value of m.o.

 
Neutron:
How many times can we talk about Martingale? It is obvious that Martingale is a variant of MM and has nothing to do with TS! By itself it will not make either loss or profit. But for a profitable TS it will increase the rate of profit, and for a losing one it will help to lose faster. In general, Martingale is a disguised reinvestment of funds.

IMHO it's not about Martingale at all, it's about the pattern of price movements

https://forum.mql4.com/ru/11661/page4

 
Reshetov >> :

Martingale combined with a trading strategy and portfolio trading, gives you a profit on the real.

See details: http://bigforex.biz/load/2-1-0-170

This, for me personally, is a very useful conclusion.

I'm checking out a roughly similar strategy with bounce signals:

Martingale, or averaging, is used because of the relatively large signal error.

It has been observed that its reasonable application actually improves system performance.


Different lot sizes.


You can clearly see that the character of the curve remains the same, but with the positive behaviour of the strategy the win with this tactic is higher!


 

It is extremely dangerous to overshoot using this tactic. It is as dangerous to sit too close to stoplosses as it is to place them too close.

Below are the tests with different values of stoplosses versus takeprofit.

Lots 1-2-5-11

 
There's been a lot ofcatching!
Creepy... ;)
1...345678910
New comment