Martingale experience - page 2

To add comments, please log in or register
Sky L
1269
Sky L  
Prabir Paul:
Soon you'll destroy your account in real time market.....

Thank you, Prabir. 

I know martingale is dangerous. Could you share any experience to lower the destroy risk? 

Sky L
1269
Sky L  
Kubilay Kaletaş:


 it blows the accounts eventually.

Thank you, Kubilay. 

Yes, the blow event is very difficult to avoid.

Back test will not show the blow event, because every author will design it away from hit any blow event in the past, but it won't work in the future.  

The past is like a pit which could be seen, I can easily jump over it, but I can’t skip the pit that I can’t see...
Amir Yacoby
1445
Amir Yacoby  
Sky L:

Thank you, Prabir. 

I know martingale is dangerous. Could you share any experience to lower the destroy risk? 

You can make it somewhat more reasonable if you comply with solid mm (martingale is anti mm) principles, likewise:
Trade 1 - 0.5 percent risk
Trade 2 - 1 percent risk
Trade 3 - 2 percent risk (or 1.5 to be)

And thats it. Next trade back to step 1.
amando
2886
amando  
With an unlimited account equity an martingale isnt dangerous
el_looto
56
el_looto  

Ive seen a mathematical paper (y'know, academic with proofs) that shows martingale systems, mathematically, have the exact same chance of doubling your stake as the probability of any single outcome.

ie if your system has a 53% win rate, the chance of martingale doubling your account is 53%.

This hold true no matter how you 'tweak' the martingale; all that changes by tweaking is the time it takes, on average, to double or lose. 

Making the doubling time take longer can lead to the impression that the system is "working."


Martingale, simply put, is a red herring at best and outright fallacy at worst.

Sky L
1269
Sky L  
Amir Yacoby:
You can make it somewhat more reasonable if you comply with solid mm (martingale is anti mm) principles, likewise:
Trade 1 - 0.5 percent risk
Trade 2 - 1 percent risk
Trade 3 - 2 percent risk (or 1.5 to be)

And thats it. Next trade back to step 1.

Thank you so much, Amir. 

Very good idea. 

Sky L
1269
Sky L  
amando:
With an unlimited account equity an martingale isnt dangerous

Thank you, amando. 

Sky L
1269
Sky L  
el_looto:

Ive seen a mathematical paper (y'know, academic with proofs) that shows martingale systems, mathematically, have the exact same chance of doubling your stake as the probability of any single outcome.

ie if your system has a 53% win rate, the chance of martingale doubling your account is 53%.

This hold true no matter how you 'tweak' the martingale; all that changes by tweaking is the time it takes, on average, to double or lose. 

Making the doubling time take longer can lead to the impression that the system is "working."


Martingale, simply put, is a red herring at best and outright fallacy at worst.

Thank you for your continuing support, el. 

Marco vd Heijden
Moderator
13012
Marco vd Heijden  

It also depends on what you aim for or the Risk you are willing to take (or how greedy you are).

More greed means you will have to rely more on Luck and The Luck Factor is the real unknown factor in these systems.

Whatever it is, it needs to be eliminated to stabilize the final results.

Personally i do not believe in Luck.

I believe in Probability and Statistics.

So we can say that every Candle has a value, first let's assume a value of the entire candle in the base currency on a one Lot Position.

  

Here is the plot in Candle values.

Suppose you trade the entire candle, from the time it opens, until the time it closes, what will your target be ? 

A lower target, statistically gives you a higher chance on a successful outcome because the lower you go, the more candles will hit your target.

But the target has to make up for the costs of doing business like spread and commissions and swaps, and also to cover the costs of all of your other losing trades.

Set it to low and you will have a lot of small profits, but risk not going to be able to make up a positive outcome in the long run.

Set it to high and you risk having too little big hits to eventually cover all the costs of all other failed attempts.

Smaller targets have a higher Probability of being hit.

This example here only deals with the value of the entire candle, so it's unidirectional, it does not deal with direction (yet), its just to illustrate a basic principle.

Because a Candle is made up of the Long and Short fluctuations in Price, over a certain time span, you can never really win the entire candle but we will neglect this fact in this example.

The results of this calculation can theoretically be divided by 50% to include for a random direction, and then another element appears that drives final results, which is your ability to establish the trend (and to do so at the right time.)

These are however separate calculations which are not included in this example. 

So let's put it in perspective.

We randomly pick a direction and open a 1 Lot position the moment a new H1 candle opens.

If the direction we choose is right we will end up with the target profit - minus costs of doing business.

If the direction we chose was wrong we will end up in a loss.

So what are the odds?

A Picture speaks a thousand words.

The sample size used in this calculation is 100000 bars.

We set our target at $400 such a candle only happens once in around Ten Candles.

This means we need to be very Lucky on this one.


When we set our target at $300 we will hit it twice as many times as the previous target but there is still an awful lot of Luck involved.


 

We set our target to $200, this reveals the unseen because again we see a doubling in the number of times we hit our target, while the need to rely on Luck keeps shrinking.

The Probability of hitting our target increases as we get less greedy, but keep an eye on those Spreads.


Whoa ! What happen here ? We set our target to $100 but the times we hit out target increases many fold ! Is this what they call Exponential ? 

The Luck involved has also shrunken dramatically, There exists a Probability here of hitting our target almost 8 out of 10 times, of course we would still need to be right on the direction...and at the right time on the right instrument with the correct formula which takes n years to find it and get just right...


Less is More ! This is more like a sure thing. Bye Bye Luck. We are going to be wealthy soon. 

Statistically spoken, if your target is fine tuned you should be able to hit it 9 out of 10 times, while only the tenth time would you need to rely on Luck...

Of course it's not that easy because you also need to be right on the direction, but those particular odds can also be > above 50%.

And I'm Positive that it is possible to work something out.


So there you have it a small example on my understanding of Probability and Statistics.

Of course these calculations are based on an ideal situation, far from reality, but it's possible to refine the formula in a way that it starts to produce very accurate and usable numbers.

Also there are other elements involved like market activity in terms of time the stats are different for H1[0] then they are for H1[16],...

And of course a EURUSD Candle differs a lot in performance then a GBPUSD Candle so these values have to be established for each instrument independently.

And as a footnote, Please don't trade these numbers, they are for informational purpose only !

Sky L
1269
Sky L  
Marco vd Heijden:

It also depends on what you aim for or the Risk you are willing to take (or how greedy you are).

More greed means you will have to rely more on Luck and The Luck Factor is the real unknown factor in these systems.

Whatever it is, it needs to be eliminated to stabilize the final results.

Personally i do not believe in Luck.

I believe in Probability and Statistics.

So we can say that every Candle has a value, first let's assume a value of the entire candle in the base currency on a one Lot Position.

Thank you so much for your continuing support, Marco.

It seems like scalping is very good to combine with martingale. 

To add comments, please log in or register