Does the martingale system really work?

The win rate from the trading strategy itself essentially becomes irrelevant once martingale is applied. This is because, no matter what strategy is driving the entries and exits, martingale always guarantees recovery to the breakeven point. This is true whether the trading strategy has a 90% win rate or a 10% win rate, and is why every martingale-based equity curve moves upward at a 45° angle, with the telltale dips and recoveries all along the way. However, the sequence of losses that causes ruin is likely to occur much sooner, with the 10% winning system, as the dips in the equity curve are both deeper and much more frequent.

If a system has positive expectancy, then applying martingale is about the silliest thing anybody can do, since martingale does not alter long term expectancy, but merely increases the risk of ruin.

MM in itself can never provide an edge; all it does is alter the size/frequency distribution of wins and losses. Martingale provides an exchange of guaranteed recovery in the short term for guaranteed eventual ruin in the long term.

This is exactly right, if you've any doubts check this live equity curve https://openbook.etoro.com/aachen1968/stats/ and this guy is about to blow up his account.

No it's not irrelevant.Why do you think a roulette has only a maximum 47% win rate, no matter what numbers you pick on your initial bet, because everyone knows that once the numbers get above 50% things will go wild,so that many people who do roulette use martingale, and if they had an edge they could potentially make infinite money.And also its mathematically proven that martingale in some cases work, if certain criterias are met.Those criterias are like i said a must have 1:1 Risk/reward ratio and an edge over the 50% win rate,so atleast 51% recommended.

If you dont believe me do this little experiment, take a dice, you know a dice has 6 sides.Now there will be 2 groups representing the "heads & tails", 1,2,3,4 will be in the first group and 5,6 will be on the second group.Why because you clearly have inequal probabilities rolling the first group.So you have 4/6 =66% rolling the first group and 33% rolling the second group,so you have an edge.Now the risk reward is easily simulated you always bet 1$ in the start so you win and lose the same amount.Start with 100$ capital.

Now play this game with a friend, and apply martingale, always bet on the group 1 where you have 66% win rate, if you lose double the bet, and so on, if you win then reset the bet to 1$.The results should be, that you will not blow up that 100$ never, its a mathematical fact, why because you had an edge, of course without martingale you would have still won, but if you didnt had those criterias you would have still lost your account, sooner or later,with or without marty, but like i said in the previous post: martingale will only quicken up things,but it doesnt change the fate of things,thats the only thing it does, either be the succes of your account or the failiure, thats up to the original percentages. I hope i convinced you!

He blows his account up because he didnt had that >50% win rate and 1:1 Risk Reward.In trading its more tricky because you cant have a mechanical RR.I dont know why, but i also tried to make my system work, i tried all 1:1 RR settings and the maximum win rate that i could make was about 45% no matter how hard i tried (btw i did this with one of my EAs, to be quicker).So its like the market has some fundamental property that its not compatible with martingale, strange nonetheless.So even if martingale could theoretically work, that doesnt mean that the forex markets can handle them.I think that the spread is the cause, that is taking away our edge, but yea what can we do...
I hope i clarified a few stuff here, and sorry for my bad english!

I'm not convinced, unless you can at least quantify the risk of ruin for both cases, to back up your "mathematical fact". I suspect that the ROR is of a massively higher order when using the marty, than not. If so, then the marty has a HUGE bearing on success or failure.

For example, it takes only 7 consecutive losses, doubling from a starting stake of $1, to blow a $100 bankroll (1+2+4+8+16+32+64 > 100).
The probability of 7 consecutive losses over any sequence of 1,000 trials, given a 65% win rate, is approx 47% (see attached table).
Whereas if you don't use the marty, it should be self-evident the probability of blowing the account (number of losses > 2x number of wins AND number of losses > 100) is infinitesimally small.
Admittedly, the account will grow quickly over the 1,000 trials, making the 47% figure inaccurate (e.g. if the account reaches 500, it will take 9 consecutive losses), but I suspect that the comparison, in terms of orders of magnitude, will remain within the same ballpark.

Your next challenge is to find a forex methodology that delivers a 66% win rate at 1:1 RR.

i trade my own martingale idea,

but have you really really played this game ??? you don't have to play with your friend, just yourself

geez, you'll blow it up well likely within 100 rolls, try it, fool!

are you serious? are you dumb or trying to fool people?

about "martingale doesn't bring an edge", i used to believe it. but i found non-random edge using something with martingale nature.

and this non-random edge (i may well likely be wrong) doesn't really exist when running without the martingale MM.

i don't remember for sure, but i try to be accurate, my so-called 'non-random' edge:

eurusd, if touching last week's high/low, enter following the breakout direction;
if you have floating loss, then double and enter the opposite;
if price reverse again, you double again and enter the opposite;

from 2009~today, straight up, no DD almost. but this system cannot be that profitable if i use only 1 entry with stoploss.

similar ideas have been covered in the prior posts by the other members. you can apply martingale to a contrarian method like "price 100 pips higher than 100 EMA, you enter short", keep double position size, but quit after 3rd or 4th time.

these ideas sound similar, and i know mine cannot be as profitable if only trade step 1. i guess the other members' system (above) shares similar nature.

why? experienced, mathematical, traders? please enlighten me.

David, that's a great example, and 100% true to roulette betting. however, not so true for forex trading.

MathExtreme made 2 assumptions that many martingale system might easily override.
1. the flat bet has no edge -- not so for many forex systems; (i know death sequence still exists)
2. each bet is isolated, the 2nd bet doesn't have a higher winning % -- again, not true for some forex systems.

for the 2nd point, i cannot give an example as the systems i know are private, but i can say this:

the ultimate problem for our trading (not only for martingale), is that right after our entry, the market will reverses. ---- how to solve this problem?

now as to the systems i mentioned for 2nd point, for every prior going-sour trade, the winning chance of next trade grows. in fact, it is often because the prior trades going sour gives a bigger % for the following trades to win. Isn't that overriding the 2nd point above? no it is not necessarily a 20 pips one-way grid system......

it actually takes lots of backtests to examine this. the mathematicals often sound convincing, but not following reality.

i also have an analogy:

thru the U.S. 40k + ppl died in car accidents annually. Would ppl quit driving due to that fact? no! ppl keep driving and hoping that day never come. and the majority of folks do survive that 'death' incident and retire driving at 70's their age. i, trading martingale, use this analogy as trading mindset. even without martingale, a trader still have the % to death trades, doesn't she?

Maybe if everyone drove their cars the opposite of safe? Accelerated into corners instead of slowing down until after the apex? Drover faster in the wet and slower in the dry?

Oh happy day for you all... My posts go straight to the bin. Good one twee. This place never changes, and martingale never wins. Classic Factory, indeed.

Charvo,

I can hardly believe it myself, but it seems that I've made > 5,000 posts on FF (what a waste of my life, LOL). Probably > 1,000 of these posts have been on various types of money management, and all I seem to do is end up repeating myself. I have given my views on martingale many times. Perhaps I have not explained them well enough, but I have given it my best attempt. I've also posted many links to articles by mathematical experts and PhDs who have a better grasp of statistics and probability than I do. I've also posted spreadsheets and tables that I've used to satisfy myself that these experts are correct in their conclusions.

I believe that it is a mathematical fact, that MM, including all forms of martingale, can not change the long term expectancy of any system (entries+exits). That is regardless of whether it is a winning or a losing system, before martingale is applied. And regardless of the system, the recovery mechanism inherent in martingale will always bring your account back to breakeven, until the day finally occurs when a sequence of trades causes the stakes to suddenly and unexpectedly escalate to a level that creates irretrievable drawdown. It's true that all losing systems will eventually cause ruin, but martingale, with its escalating stakes, will bring it on much more quickly, possibly giving the trader little or no time to react before it occurs.

Some people seem to assume that, just because everything they've tried with entries and exits doesn't work profitably, that money management variants MUST be the answer. But that logic is flawed, hopefully the reason is self-evident. It is because people's knowledge is finite, and the number of possible solutions (profitable or otherwise) is potentially infinite.

Others seem to assume that the laws of expectancy apply only to casinos, and not to forex. (That seems to be your argument also, i.e. "100% true to roulette betting. however, not so true for forex trading".) But at the end of the day, forex is nothing but another win/loss generator, just like any other game of chance.

Re your car accidents analogy, a lot of people attempt to find analogies to forex (e.g. here). As a counterexample, I could liken martingale to playing Russian roulette. But there is no real causal link, and it doesn't really prove anything.

Maybe the math experts and PhDs are wrong. It really doesn't matter. People who want to use martingale will always find a reason that convinces them to do so. It's their money and their choice.

David

Hey nubs, good to see you back.

And your keyboard seems to have grown a caps lock key, LOL.

But anyway, you're right, the place never changes, for better or worse, diplomacy seems to always rule over truth here.

Take care, old buddy.
David

I'm not. Occasionally I drop by to cruise the bin, and then scan the main threads to see what's new. Nothing is ever new. Even the bin is boring these days when I look. It's like everyone has just given up. One time I was gone for aaaages, like a few months or something, and when I came back and had a look it was still all exactly the same. Oh, apart from the new toys on the site and the slow morphing into a fully-fledged signal service cluttered with more muppets and scammers than you can poke a stick at.


Y'know, I actually noticed that as well and almost went back to 'fix' that because it's just not right. I've started a habit of using caps from typing in other places. It makes it easier to blend-in with the other assholes on the net if I just type properly.


Yup. Please and thank you's will get you everywhere on this site. Clearly the result of great moderating.

People continue to want to implode their accounts and try to convince others that it's the way to go. And for some reason you still have the bloody patience of a Saint. I'm still hoping to see the day you rip some nugget a new asshole, but you keep holding-out and make my sporadic visits here fruitless. You placid bastard. I don't know how you manage to do it. It must be something in that water over there. Why must you break my heart this way?!?

It's still kind of amusing to see people fail to grasp basic concepts and regurgitate flawed 'systems' and 'proofs'. People in forex and roulette just seem to LOVE that martingale holy grail. Charvo wants to talk about risk and how everyone is always under a certain amount of risk... but you don't go out of your way to make losing your head or your money a guarantee. Charvo, that's just bad form. Wear your seatbelt. Brush your teeth. Don't throw your wallet into fireplaces.

Martingale, and everything that smells like it, is a true fools game. Even with a positive expectancy it's a fools game, for those special cases of intellect that still grasp to the martingrail under the presumption of a scenario involving an actual edge.

Nobody has ever shown any reasonable use of martingale other than fictitious 'unlimited funds' scenarios. If anyone has some ground-breaking martingale concept to share then go right ahead and impress the pants off of me and maybe I'll stick around a bit longer, or maybe just long enough to give you your earned Scouts badge of muppetry.

In the meantime Hanover will continue to explain to you, with diagrams and everything, why you guys are so damn wrong... and you'll all still fail to grasp it. I'm impressed with Hanover's patience and dedication as well as the dedication of others towards BAD ideas.

Hanover, you and your passiveness pisses me off, but in the best possible way. As for everyone else... the rest of you just piss me off. And thus the world continues to turn.

Adios amigos.

What if you scalp making money every week and one day you loose small micro account?

Nub, don't go yet. it is comforting to see old legs like you now and then. i'll post a backtest too tonight after work (NY time), just for fun, the backtests will be 5 yrs eurusd, with 3 level martingale MM......

i actually didn't believe martingale AT ALL. it is very recently that I've found seemingly limited martingale is helpful...... well, as Hanover suggested, i believe that i raise the position size with the winning % rising......

Edited with backtests of my EA:

testing software: forextester 2; testing period: 01/2009~09/2013, almost 5 yrs.
First Test: 1st image is its statistics, 2nd image is the balance/equity/DD;

Second Test: 3rd image is its statistics, 4th image is the balance/equity/DD;

Two tests used one same EA, same settings, except that, for first test, 2 more consecutive entries will be made, if the prior trade reaches a certain level of DD. That is, for each entry opportunity, a 3-step trading sequence in first test V.S a single trade used by second test.

Everything else is same, including tp, sl level.

although this EA is not really tradable (LIVE) in my opinion, a non-random edge is achieved by first test's martingale nature MM. According to second test, the EA is a typical random EA.

I in fact don't understand why as this is against my belief that was same as Nub and Hanover. All I can say is that I'm not that sure now......

Charvo, run some more tests.

The first one was only 100 trades, and the second wasn't much better.

The first set had a higher profit factor than the second... i.e. the set with less data points (the "hey guyz dis is so easy with da martingrail") showed a higher profit factor than the one with more data. Or to put it another way, the more data you threw at it the worse things got.

And look at the max drawdowns! The smaller set with the higher PF had a massive drawdown compared to the net profit, and the second with the worse PF had less profit than there was drawdown! How can any of that smell right to you?

And for all of that time spent wasted on martingrails there's only a handful of trades over all of that time, risked a butt-load of money to make chump-change, and got nowhere.

BAD.

yes, i think that's one major issue, too few samples to be significant. but that's a weekly trader and i didn't have larger data handy.

in terms of trading, do you guys actually think even a limited martingale application like my 3-level only position sizing is meaningless?

Yes.

You are under the assumption that your chances get better after 2 losses. IF... IF that were true then you'd be better-off just waiting for two fuckups before actually entering a trade.

"But nubcake, you fucking dickhead, then I miss all of those times I win on that first or second entry", in which my reply is "So then trade all entries". To this you ponder, and in return you respond with "But then there are times where I lose on the first, or the first and second entry, and that third entry being larger makes-up for it all", to which I say to you "So what you are actually saying is that sometimes you win on the first, or the second, or the third.... and so on. So really you don't actually know when you are going to win".

If you want to have a hard-cap on the levels of progression then consider this: If you traded this way and then looked-back on the data and categorized each trade under a section labelled "level 1", "level 2", and "level 3", and looked at the trades just within their own categories and crunched the numbers what are the results you get? If they are all following the same strategy theme for entries and exits then they should all give roughly similar data. They will either be, on their own merit, overall winners or losers.

By trading some progressive betting system based on losses all you are doing is abstracting away from a system that uses small bets and which loses with the same exact system that uses larger betting... which will sooner or later also become a losing system... which you then try to override again with the same losing system but with bigger bets again.... and bigger again... and bigger again....... annnnd so on.

Now that's not even talking about risk of ruin. This is just the system itself. Each betting level can be analyzed on its own, and when you do if you see that it is a failure on one level then going EVEN BIGGER with a new level of betting is just retarded. You can only win if you can keep increasing your betting level... i.e. if you have unlimited funds... otherwise if you cap your betting levels then you are just exposed to the failings of that highest bet level, plus the ones below it that got you to that larger level in the first place.

If you cap your betting level then you are minimizing your risk of ruin to an extent, but you've done nothing to magically create a winning system out of a losing system via a betting scheme. A bad system plus bad money management equals bad.

BAD.

Let me paint the picture this way instead:

Let's pretend that a failing system always magically starts-off seemingly a winner, but as you continue to make more trades the amount of profit per trade gets smaller and smaller until eventually it becomes losses, and then as you continue to trade it eventually starts to turn back upwards from losses back into profit.

It's a type of circle. It's actual a Sin wave if you plot it over time, but whatever.

In the end it all washes-out fairly even, and all you have lost overall is the spread costs associated with all of the trades it took to come full-circle in this breakeven-ish system.

Now, let's stop pretending, because this isn't actually some pretend system. This is 99.99999% of peoples system! People in general don't jump-in with a system that already shows itself to be a failure. They see a system that can win more often then not, so they start trading it (they are at the start of the upward curve in Sin wave). They make some money, but it starts to become less and less (getting closer to the top peak of the wave). It becomes break-even (reached the peak). It starts to lose more than it wins on average (on the downward slope). It's now lost all profits and the balance is now in negative (still sliding down below the zero line). It starts to stop losing so much (reaching the bottom trough). It stops losing (bottom reached). It begins to win again (back on the upward slope). The account finally comes back close to where it started (the cycle is complete, minus spread cost loss).

This is a simplified version of what happens irl. All you are trying to do each time you go up a level is hope to be positioned somewhere on that upward cycle for the system at that betting level, where each level has its own distinct cycle once you plot it all out over time.

But since irl a system doesn't perfectly have 'seasons' of profit and loss that you can selectively work around you just have to take it as it comes, and by the time you conclude that you are in a profit or loss season it's already too late to capitalize on and you might now be back on the other side of things, or not... you can never know the extent of the peak or trough.

If you could time catching an upward slope then you'd not need to stuff around with rubbish betting systems.

But if each time you went up a level you always caught the upward slope then eventually you'll start reaching that betting level and that level will have moved into the downward slope... and so you'll need to bet bigger AGAIN. And the cycle repeats over and over, larger and larger.

Each level, on its own, is an overall failure, and will always require larger levels on top plus some luck to overcome the losses. A hard-cap on levels just means you stop the progressive level silliness, but you are still exposed to the same failing entries and exits of your system. It'll all fall-apart sooner or later because the reason you don't just grind-out at one level and be satisfied with the system and always want to up the ante (i.e. the system sucks and loses money) will affect all levels you bet at eventually.

Again, split each level into their own categories and they'll all roughly show the same data when you analyze their results per category. It's the same system each time.

And once more just for fun... if you were to ignore trade 1 and 2, and just traded your level 3 trades then what would you get? The same results as just grinding at level 1. Failure. You can't pick and chose your trades, and chopping and changing when you do and don't trade just adds noise to the overall data that reveals a system to be failing. You don't trade just your level 3 entries because you know it fails in the end just as if you had traded all entries... so why are you bothering to ramp-up your losses into your level 3 bets which also eventually lose?

Martingfail. Cap or no cap.

Or to put it one more way before I go: if you were to flip a coin and write down the results for a buttload of coinflips you'd have a nice set of data to look at. If you found every instance of 2 tails and then looked at the next flip result you'd find equal amounts of heads and tails after 2 consecutive tails.

I.E. your chances don't get any better just because you had two fails (tails) in a row.

If you counted the result after consecutive tails for 49 flips, then you'd find just as many heads as tails for the 50'th flip. IF you are some sort of lucky bastard then you might get through life being the one fool who was fortunate enough to get more heads than tails on that last flip, just as some unlucky fool out there will always get more tails then heads on that last flip.

If a breakeven system always reverts back to breakeven on a long-enough timeline, then a less than breakeven (a losing system) will similarly revert back to a losing system on a long-enough timeline. You can't bet on that last flip being 'more likely to win', because it's not. There will be 'seasons' where it will win more than not, but that's true of any random system. You can't pick the winning sections from the losing sections in a random environment.

So, all that leaves you with is actually having a system that has good entries and exits that wins you more money over time than loses, and you put away childish toys like martingale.

Nub, i must thank you sincerely for taking time and rare patience (of yours) to elaborate this topic, you must be pretty bored, lol. it surely cleared up a lot of confusion that i'm having right now. i appreciate the comments from you, hanover, and pipmeup in last few pages. they are golden.

purely the result of being up all night writing code and needing a good break.

martingale usually comes down to believing that your odds get better, or worse, after a streak of some kind. Even if you flip the idea and martingale after wins then it's still exactly the same net loss system abstracted into different levels of betting. Whenever you manage to reach the higher levels you are still just as likely to win or lose as any other time, so either your system is a net positive system in general in which case there's no need for betting systems, or it's still a failure and will eventually fall-over.

you don't EXPECT a solid string of win or loss in something like a coin-toss 50/50 situation. if person A flips a coin 9 times and gets tails, and then more often than not the 10th flip is heads, they conclude that "you are more likely to win after a string of losers". but there is person B on the opposite side of the world flipping a coin 9 times and getting tails, and then more often than not getting tails on the 10th flip and so concluding that "a string of losers begets more losing".

and then you have person C who concludes that after a string of wins it's best to ramp-down the betting in anticipation of a loss. and you also have person D who concludes that after a string of losses it's best to ramp-up the betting in anticipation of a win.

and finally, you have everyone else who recognizes the nature of randomness and concludes that you can't pick the winners from the losers in advance and so it's best to just keep the betting at a reasonable level so as not to risk blowing the bankroll after a wave of inevitable losses. they conclude that there is no edge in trying to game randomness via bet sizing.

when you hear about someone who always wins after a string of losses, just know that somewhere there is a person who gets the opposite result and who doesn't spend their time on forums trying to mathematically prove their mathematically wrong conclusions to people who are more mathematically intelligent.

CAN YOU MAKE A LIVING WITH MARTINGALE? yes... but only if you are that one lucky fucker who catches a lifetime of lucky and unexpected breaks. everyone else gets to deal with the reality of not being 'lucky'.

this is the whole reason why someone who has made 10 perfect trades and quadrupled their account in their lifetime get shit-on by people who have made hundreds or thousands of losing trades. nobody cares for what that one lucky idiot has to say because they are the outlier, and their house of cards will soon fall down or they will continue to be the lucky outlier whose results are not grounded in the average persons reality of not living a charmed life.

if someone wants to bet on being a lucky person who manages to catch more wins after a string of losses, or more wins after a string of wins, or whatever combination of random luck-based betting, then good luck to them and they are better-off condensing their luck into a handful of bets... or just go to the casino and bet your grub-stake all on red. you will be lucky right now, or you will drag your luck out over a longer horizon. in the end it's still just a gamble on being lucky one way or another.

This is of course true as all flips are independent of each other, but why does a large number of flips follow a normal distribution? I have trouble reconciling the two ideas in my mind.