No Loss Martingale Strategy | Page 7 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Big Loss Martingale Strategy! 12 replies

Martingale, Anti-martingale, and Compounding 40 replies

Martingale, Reverse Martingale, Modified Martingale, Maths 1 reply

How to change this Martingale to Reverse Martingale ? 3 replies

Martingale vs. Non Martingale (Simplified RoR vs Profit) 1 reply

Attachments: No Loss Martingale Strategy
Exit Attachments

No Loss Martingale Strategy

  • #121
  • Sep 18, 2015 8:55am Sep 18, 2015 8:55am
Quoting edyraper
{image}
Either contribute to this thread or stop trolling with your narcissism.
  • #122
  • Sep 18, 2015 9:00am Sep 18, 2015 9:00am
Quoting ollaya
{quote}The expectancy will be always 0 in the long run for 50/50 system (coin tossing) .
Only that coin toss 50/50 rule can't be applied to trading when you factor in broker commission.
  • #123
  • Sep 18, 2015 9:58am Sep 18, 2015 9:58am
Quoting Nabz
{quote} Only that coin toss 50/50 rule can't be applied to trading when you factor in broker commission.
correct , you need a system with >1:1 RR. and trading should not be random and as Hanover said you need to put the odds in your side by using some sort of analysis ( tech or fund)....and again i did not test this. i am just translating what i have read in french here... this method worth at least the effort of verification . Hanover calculated the success factor at 90% but the 10% will make you lose everything you won... logically you can increase the success of this method by using a high probability trading method.
  • #124
  • Sep 18, 2015 11:58am Sep 18, 2015 11:58am
i found this searching in google : So cool , this is how to calculate the probality using Chrome :

" Here's a way to get an approximation for the answer via a simulation. It's kinda fun to tweak the numbers and see the probability change.
Run this in Tools -> Javascript Console in Chrome or Tools -> Web Developer -> Web Consolein FireFox:
function experiment(streak, tosses) { var i, ctr = 0; for (i = 0; i < tosses; i++) { if (Math.random() < 0.5) ctr = 0; else if (++ctr === streak) return 1; } return 0; } function run(streak, tosses, runs) { var i, total = 0; for (i = 0; i < runs; i++) { total += experiment(streak, tosses); } return total / runs; } console.log('\nP(A) = %s\n', run(2, 12, 1e6));
I get the following result:
P(A) = 0.908336 the same results as Hanover`s spreadsheet.

i did this with 2 wins in a row and 24 trades:

function experiment(streak, tosses) {
var i, ctr = 0;
for (i = 0; i < tosses; i++) {
if (Math.random() < 0.5) ctr = 0;
else if (++ctr === streak) return 1;
}
return 0;
}
function run(streak, tosses, runs) {
var i, total = 0;
for (i = 0; i < runs; i++) {
total += experiment(streak, tosses);
}
return total / runs;
}
console.log('\nP(A) = %s\n',
run(2, 24, 1e6));
VM1100:19

P(A) = 0.992815

what do you guys think?
  • #125
  • Edited 3:44pm Sep 18, 2015 3:03pm | Edited 3:44pm
Quoting ollaya
the advantage that I see in Alembex is that , we are not adding to losers and the drawdown is much less than other (anti-martingale )systems.
I see no advantage in using Alembex (or any other progressive pos sizing method) over flat betting, because -- as the math has demonstrated -- the long term expectancy is exactly the same.

Quoting ollaya
And I believe that coupling this betting method with system that can generate 2 winners in a row in less trades will make the system more profitable .
Yes it will, but if I found an entry/exit system "that can generate 2 winners in a row in less trades", then I would simply flat bet the system, because the long term expectancy is exactly the same, and the risk of ruin (by avoiding Alembex's higher average pos sizes) is exponentially less.

Quoting ollaya
We would love to see and EA using this system and compare the result of the same EA with fixed betting ( % or fixed lot/$).
I am convinced by what the math has demonstrated, hence IMHO writing an EA would be a waste of time.

Now I will disappear, and let the OP continue with his line of discussion.
  • #126
  • Sep 18, 2015 7:39pm Sep 18, 2015 7:39pm
the math is correct if every card is played, but we have limited life span that it is not evenly gives all the chance to rise.
so , to beat the math, we need to lessen the trade.
  • #127
  • Sep 18, 2015 8:14pm Sep 18, 2015 8:14pm
I'm back with another idea, my dudes:

there is a mathematical certainty that the price can not keep moving in one direction without a retrace, what do you think if we can scan the market in a special pair and achieve to some information.

for example: since January 1 2015 until now, the AUDUSD pair has never moved more than 240 pips in short without having an at least 25 pips movement in long, suppose our martingale strategy can withstand 150 pips if market goes in the opposite direction, so wee need an indicator that send a signal when the market has moved 100 pips without retrace(or even more, to being cautious) in short, then we conclude its time to start a martingale trading in long with the TP of 20 pips.
so, what do you think?
  • #128
  • Sep 19, 2015 2:38am Sep 19, 2015 2:38am
your blow-out is just a matter of time, it will happen sooner than later wich such a simple strategy ...
  • #129
  • Sep 19, 2015 3:04am Sep 19, 2015 3:04am
Quoting cvds16
your blow-out is just a matter of time, it will happen sooner than later wich such a simple strategy ...
plain martingale strategies are more simple than mine, however their still profitable at least in a period of time, but let's see what the others say about it, and,
can you give me a better idea?
  • #130
  • Sep 19, 2015 3:31am Sep 19, 2015 3:31am
Hi everybody
please help me... what is wrong...?
Attached Image (click to enlarge)
Click to Enlarge Name: StrategyTester.PNG Size: 31 KB
  • #131
  • Sep 19, 2015 5:54am Sep 19, 2015 5:54am
Quoting hanover
{quote} I see no advantage in using Alembex (or any other progressive pos sizing method) over flat betting, because -- as the math has demonstrated -- the long term expectancy is exactly the same. {quote} Yes it will, but if I found an entry/exit system "that can generate 2 winners in a row in less trades", then I would simply flat bet the system, because the long term expectancy is exactly the same, and the risk of ruin (by avoiding Alembex's higher average pos sizes) is exponentially less. {quote} I am convinced by what the math has demonstrated,...
Say for example that we will cut the trade size in half after a loss and keep the same betting tha Alembex use after a winner ( Cum +1).
will that make a difference , will that reduce the DD?
  • #132
  • Edited 7:33am Sep 19, 2015 6:30am | Edited 7:33am
Quoting ollaya
{quote} Say for example that we will cut the trade size in half after a loss and keep the same betting tha Alembex use after a winner ( Cum +1). will that make a difference , will that reduce the DD?
t think this will do the trick . i did some change manually to the file you made and already the postive expectancy start to show.

i am not very good at excel ...can anyone redo the simulations based on the last rules ( reduce the trade size by half after a loss and add 1 to the cumulative loos after a win).

let`s see the final expectancy !

here is a good one too . we can reduce the DD more by betting the min lot size after a loss and keep the Alembex betting method after a win :
if the 1st trade is for 1 lot and we have a streak of 7 losses , the lot size for those 7 trades should be 0.1 i.e if we have a win after those losses , the next trade size will be 2.7 and if the 2.7 trade is a loser we go back to 0.1.

the idea is bet the min until the we have a win , then go big .

anyone can do the test with these conditions
Attached File(s)
File Type: zip Belkhayate expectancy debunked (3).zip   1.3 MB | 868 downloads
  • #133
  • Sep 19, 2015 6:55am Sep 19, 2015 6:55am
Quoting Fxentropy
Here it is.
HERE it is!
Money Can't Buy Happiness. Poverty Can't Buy SHIT! You Choose!
  • #134
  • Sep 19, 2015 11:28am Sep 19, 2015 11:28am
Quoting ollaya
{quote} Say for example that we will cut the trade size in half after a loss and keep the same betting tha Alembex use after a winner ( Cum +1). will that make a difference , will that reduce the DD?
It makes no difference. Overall expectancy remains 0, exactly as I expected. See XLS in attached ZIP.

Here is s statement that will probably shock many people, but I believe that it's 100% true:

You can take absolutely ANY entry/exit system, and then no matter what sizing system you apply to it, then (providing each trade is an independent event, i.e. there is no causal link between trades), the total long term expectancy will be exactly the same as if you use consistent sizing (i.e. 1 unit for EVERY trade).

If you think about it, this is amazing because there are a potentially infinite number of entry/exit systems that can be devised, and a potentially infinite number of sizing systems also. Yet it applies to every possible situation.

If the statement in blue is true, here are some obvious implications:
 Any 'edge' comes entirely from entries and exits.
 Pos sizing can not provide an edge. It can not turn a negative expectancy entry/exit system into a winning one.
 And no pos sizing system can make an existing positive expectancy entry/exit system more profitable than it already is.
 Experimenting with different pos sizing ideas is a complete waste of time.

I'm not going to spend the rest of my life testing all of these possibilities. I have other projects to work on. I've seen enough to know that the statement in blue typeface is true.

There are two objectives that a trader must juggle with: (1) maximizing profit, and (2) minimizing risk of ruin. Larger pos sizes increase profit, but also increase risk of ruin. If the statement in blue is true, then flat betting (1 unit per trade) does as good a job as any other sizing system of maximizing profit, and it reduces risk of ruin to the highest possible degree. Therefore flat betting will always be better than any kind of progressive sizing.
Attached File(s)
File Type: zip Belkhayate expectancy debunked (half following loss).zip   1.4 MB | 835 downloads
  • #135
  • Sep 19, 2015 12:26pm Sep 19, 2015 12:26pm
Quoting hanover
{quote} It makes no difference. Overall expectancy remains 0, exactly as I expected. See XLS in attached ZIP. Here is s statement that will probably shock many people, but I believe that it's 100% true: You can take absolutely ANY entry/exit system, and then no matter what sizing system you apply to it, then (providing each trade is an independent event, i.e. there is no causal link between trades), the total long term expectancy will be exactly the same as if you use consistent sizing (i.e. 1 unit for EVERY trade). If you think about it, this is...
thanks Hanover for the file ...
i did some change by betting the min lot size after a loss and keep the Alembex betting method after a win :
if the 1st trade is for 1 lot and we have a streak of 7 losses , the lot size for those 7 trades should be 0.1 i.e if we have a win after those losses , the next trade size will be 2.7 and if the 2.7 trade is a loser we go back to 0.1.

the expectency went up drasticly ... not sure if i did a mistake . but maybe the blue statement is wrong , at least for now

let me know what do you think?

I am wrong , you are right i did a mistake in the file
Attached File(s)
File Type: zip Copy of Belkhayate expectancy debunked (0.1 following loss).zip   2.2 MB | 935 downloads
  • #136
  • Sep 19, 2015 2:04pm Sep 19, 2015 2:04pm
Quoting ollaya
{quote} thanks Hanover for the file ...
Ollaya,

The reason I'm sure that the 'blue statement' is correct, is that I've performed the same mathematical analysis on hundreds of different betting systems. I attached 2 additional XLSs in post #106. You can vary their parameter settings to test many different systems, i.e. rules for changing the bet sizes following a win, or a loss, and when to quit from a sequence. In all cases the total expectancy is zero.

However, one of the XLSs allows you to change the probability of a win, following a loss. That means that consecutive trials (trades) are no longer independent events, which (if you read the blue statement carefully) means that you can gain an increase in expectancy. This is the same idea that blackjack pros use to beat the casinos: they bet more when the cards remaining in the deck are favorable, less when they're not. Note that their bets are related to the probabilities that the game is offering them, i.e. it is not merely to recover recent losses. Applying this to trading, there are situations where we could argue that consecutive trades have are not completely independent. A couple of examples:
1. The further price moves away from a given level, the greater the probability that there will be a reversion toward that level at some future point (as traders take their profits).
2. The longer price remains in congestion, the higher the probability of a breakout (that is the principle being used by artem_t back in post #26, and the mgrid EA here).

In these cases you could justify increasing your pos size following a loss, because the probability of the next trade being a winner has increased. But (as I explained in post #133) a trader must also consider risk of ruin. Increasing pos size recklessly to exploit a small improvement in probability will almost certainly result in wipeout, at some future point. The increase in size is no longer commensurate with the improvement in the probability of a winning trade. For example, if you use martingale doubling (1,2,4,8,16...) then after four consecutive losses (1,2,4,8) the next trade needs have approximately 16 times the chance of winning as the first trade in that sequence, to justify that kind of increase.

One last point: the best edges in trading come from applying robust, broadbrush principles. Looking for accuracy in numbers/parameters can easily lead to curve fitting, which future markets take no cognizance of. When developing trading systems, maintaining objectivity is critical.

Anyway, I must end here, as I have other projects to work on. Hope I've given you some ideas to try. Good luck.

David
  • #137
  • Sep 19, 2015 2:52pm Sep 19, 2015 2:52pm
Quoting hanover
{quote} Ollaya, The reason I'm sure that the 'blue statement' is correct, is that I've performed the same mathematical analysis on hundreds of different betting systems. I attached 2 additional XLSs in post #106. You can vary their parameter settings to test many different systems, i.e. rules for changing the bet sizes following a win, or a loss, and when to quit from a sequence. In all cases the total expectancy is zero. However, one of the XLSs allows you to change the probability of a win, following a loss. That means that consecutive trials...
thanks David...
i watched the MIT blackjack team few weeks ago ... good watch ( if any on e intested . here is the video :
https://www.google.com/url?sa=t&rct=...03073922,d.cWw

and what you said is correct as i stated before , we can use this method in our advantage if we have a high probability system .... and from excel file we can see that ranging market is a killer (WLWLWLWLW..).

thanks for your time and contributions and I wish you all the success and Millions $$$$$$$$$

thanks for the files , i am sure they will be very helpful.
  • #138
  • Edited 7:45pm Sep 19, 2015 7:14pm | Edited 7:45pm
Quoting hanover
{quote} If the statement in blue is true, here are some obvious implications:  Any 'edge' comes entirely from entries and exits.  Pos sizing can not provide an edge. It can not turn a negative expectancy entry/exit system into a winning one.  And no pos sizing system can make an existing positive expectancy entry/exit system more profitable than it already is.  Experimenting with different pos sizing ideas is a complete waste of time. I'm not going to spend the rest of my life testing all of these possibilities.{file}...
It is subtle, but I actually don't think I am on the same page as you with this. I am going to have to think on this some more.......

It appears that you are a strict adherent to EMH based on these assumptions which is great in theory but an appalling replicator of reality. I am not so sure that each trade event is strictly an independent event unlike say the roulette table.

Contingent events are a fact of life in trading such as the contingent events associated with a hedge fund progressively reducing their exposure over a stated timeframe by the release of trades. The very fact that traders scale into a trade suggests that contingent events are important.

Furthermore position sizing is a critical method applied in the CTA space. They may not play around with Martingale Progressions, but many of them adhere to the principle that position sizing with positive expectancy is a critical method of improving performance efficiency.

....anyway, I am 100% on board with you that there is no way to use MM as a way to turn negative expectancy into positive expectancy. Where I differ is the notion that MM cannot be used to enhance performance having achieved positive expectancy.........interesting stuff Hanover. :-)
  • #139
  • Sep 19, 2015 7:51pm Sep 19, 2015 7:51pm
Quoting hanover
{quote} It makes no difference. Overall expectancy remains 0, exactly as I expected. See XLS in attached ZIP. Here is s statement that will probably shock many people, but I believe that it's 100% true: You can take absolutely ANY entry/exit system, and then no matter what sizing system you apply to it, then (providing each trade is an independent event, i.e. there is no causal link between trades), the total long term expectancy will be exactly the same as if you use consistent sizing (i.e. 1 unit for EVERY trade). If you think about it, this is...
Let's take the case of two traders, A and B.

Trader A flips a coin to determine entries and exits. His reward:risk ratio is 1:1 and win% is 50% and his trade series is independent because there is no dependence between market moves and coin flips. There is no predictable advantage to be had for trader A.

Trader A never has any chance to come out a winner based on anything other than a lucky streak. In the long run transaction costs will wipe out A.

Trader B uses a trading system that has some degree of dependence in the trade series. Trader B also has a 1:1 system with 50% wins.

Trader B determines that based on the dependence in his system's trade series that at times his win% increases to 55-60%, and at other times it decreases to 40-45%, with the rest of the time B's system is around 50% wins. So how does B use this to his advantage?

Quote
In these cases you could justify increasing your pos size following a loss, because the probability of the next trade being a winner has increased.
B doesn't need to take every trade, as a blackjack card counter would have to do in a casino to gain an advantage. B can ignore trades where the win% is around 50% and simply take trades when the odds are tilted away from 50% by a significant amount. In this case B is able to gain an advantage simply from the dependence in his trade series.

So I suppose the underlying question is whether system trade series are really independent or if some trading systems exist that have some degree of persistent dependence greater than the underlying market? I have found the latter to generally be the case though I've created systems with trade series that fit into both categories.
  • #140
  • Sep 19, 2015 8:19pm Sep 19, 2015 8:19pm
Great link, Copernicus!
1 trader viewing now