Is martin so bad? Or do you have to know how to cook it? - page 41

New comment
 
lucky_teapot:
...

I do not doubt that someone already has something somewhere, I act as I know and as it is easier and quicker for me. And I certainly don't hope that something will appear on a mass scale.

P.S. The "contra" is not clear yet. Judging by the chart, red is better on average. If only it were not so noisy, it would be really good. I'm sure it can be improved.

No, it's not. I have never tried to use this kind of strategy, I've never tried to buy it on my real account.

And how do you manage to test below zero, if I understood correctly? In theory, there is a margin call and testing goes no further?

It's not free funds, but capital increase, mayorgincole has nothing to do with it. I place higher deposit during tests to see the whole curve. I also use equity indicators for the most part of tests, not a tester, it's faster and more representative with a price chart.

 
Alex_Bondar:

I can propose a contra (reverse proof). But first we need to agree on what is to be understood as a proof and what is a contra, so as not to waste time, I understand that people look at the exponential loss function positively. What can serve as arguments at all? Trying classical prediction strategies and comparing the result with and without martin?

It is impossible to enumerate all the strategies and even for one strategy, all combinations of parameters, timeframes, for all instruments and trading conditions. It is unrealistic, especially in the case of martin, which wobbles like an electron. The opponents will choose only profitable combinations, and who is on the other side of the barricades, claim that the MO is on their side.

For example Eurobucks, MACD 12-13 year, minute quotes from Dukas (spread ~0.7p, commission 5$ per 100 000$), fast MA runs from 0 to 1000 in 5 increments, slow is 1.5 times scaled relative to fast(slow=1.5*speed), signal is 2 times faster than fast MA. Horizontal months, vertical points (1 division 10 000), blue chart standard lot scaled in proportion to the MACD period to approximate its equivalence, red one - classic martin with the ratio of 2. Both do not take reinvestment into account.


The risul is quite impressive.

And what can I emphasise from it? Exactly nothing in the overall context. Have you yet to find a simpler strategy for your tests? Everything simple has long been forgotten and is only described in textbooks.

 
iModify:

Risulka is quite impressive.

And what can you emphasise from it? Exactly nothing in the overall context. Have you not yet found a simpler strategy for your tests? Everything simple has long been forgotten and is only described in textbooks.

This one is not simple either).
 
zfs:
It's not simple anymore either).
Did it ever work)?
 
iModify:
Did it ever work)?

It has always worked, such is the universe.

 
iModify:
Did it ever work)?
Mm-hmm)) It's always worked)) And tomorrow it didn't.
 
iModify:
mm-hmm)) It's always worked)) And tomorrow didn't.
You just thought tomorrow would be like yesterday).
 
iModify:

Risulka is quite impressive.

And what can you emphasise from it? Exactly nothing in the general context. Have you not yet found a simpler strategy for your tests? Everything simple is long forgotten and only described in textbooks.

Yes, there are tests with patha exhausts, I gave an example of the most usable combination. Although the real problem is not the slope of equity, but its volatility in all degrees of freedom (time, TS parameters).

Soon Pottson's promise to make the first version of the 3D multiple testing results interpreter according to my order and it will be possible to compare equity distributions depending on time and some basic strategy parameters. I will not reveal the profit algorithm, I will just show the annual surfaces of capital gains, with different MMs. And everyone will draw a conclusion about "robustness", riskiness, etc. based on the evenness of the surface.

ZZZ about simplicity/complexity))

iModify:

You don't have to use complex models for that. (As soon as you start applying them in the market they don't work anymore).

Any mathematical model can be so beautifully designed and its use is 0 in a real situation.

Everything should be simpler.

 

Something like that:

Blue is an annual trend system with standard lot, period runs from 1 to 1000, and crimson is the same system but with martin, as you can see the surface is extremely unpredictable in all degrees of freedom. That is, a reasonable person will not rely on such extreme luck, although it does not dive below zero, but it is still a great attraction...

Sorry this is still a demo version of what I wanted to see, there are a lot of comments, but overall it's easy to understand what it's about.

 
Alex_Bondar:

Something like that:

Blue is an annual trend system with standard lot, period runs from 1 to 1000, and crimson is the same system but with martin, as you can see the surface is extremely unpredictable in all degrees of freedom. That is, a reasonable person will not rely on such extreme luck, although it does not dive below zero, but it is still a great attraction...

Sorry this is still a demo version of what I wanted to see, there are a lot of comments, but overall it's easy to understand what it's about.

Cool. That's exactly what I was suggesting, as a standard feature in MT5. It would be very clear to see all results after optimization in such a view. A 3D chart is already available in MT5. Now we only need to do a small thing. ))
1...343536373839404142434445464748...63
New comment