Why martingale doesn't work in trading

I think the part that people miss when trying to martingale their trades is the fact that price is random..... by that I mean price hits different levels each day, one day it is 300 pips the next it hardly makes 100.

The other variable is the that price just doesn't move in one direction for the day, over the corse of the day price can oscillate or cycle up and down over a give range one or more times.

People confuse the erratic price fluctuations for the red and black of a roulette wheel, they only see the win loss aspect without taking into account the volatility of price movement, it is not a simple equal yes or no, red or black, up or down.

Martingale in roulette is a winnable system IF ....
1. you have the bankroll.
2. there were no table limits.....

Why do you think the casino puts a limit on the table?

Betting the martingale..... table limit $300 to $500, which is the norm except at high stakes tables.

there have been strings of one color that have well exceeded 10 times and more so with that in mind we will say that the sequence here is an example of such a string of loses.

table minimum bet is usually 5$ and the table limit is usually $500, there are some tables with a $2 min bet and those tables usually have a $300 limit.

so here is a $5 table example of a string of loses.
$5 loss
$10 loss
$20 loss
$40 loss
$80 loss
$160 loss
$320 loss
So with seven bets you have lost $635..... the eighth bet would exceed table limits.........

On a $2 dollar table you could bet 8 times at a loss of $512... the 9th bet would exceed table limits.

Just thought I would toss this out here.

What you say is true John but Martingale systems have their win/loss probabilities like any other system. For most of the people martingale directly implies starting at any point and average up. In fact martingale used with favorable setups give good results.
If market is random and efficient market hypothesis is true then no system should effectively work.

Martingale is just a form of Position Sizing nothing more nothing less.
If I wanted to Risk 5% on a LONG trade why shouldn't I start small
and set further Buy Limits with higher lotsizes lower and a stop level for all of
them so when all my levels are stopped out I end up with a 5% loss.
Why should I take one big position instead? There is no reason for that.

This way I can give my entry more range lets say 50 Pips with 5 Levels so on level every 10 Pips and a common SL of 65 Pips. Think about it.
You would just open one big position which would go against you and stop you out no way you could close that trade with profit. If you spread you Position size with Martingale you could exit on a retrace before it stops you out and still get out Break Even.

Gambling Rule #2: Disaster is the ultimate result of any kind of gambling. Because, there is a limit to how much one can lose, but no limit to how much one wants to win.

You neglected the "IF" in quoting me.... I said... martingale is winnable IF
you have the bankroll and there are no table limits..

One fine evening in Vienna, Austria I walked out with about 35K of the casino's money, and I did it with Martingale. It took about 8 hours of hard work, but I did it. My biggest DD was one series where it went maybe 7 or 8 passes against me before hitting -- but man, what a hit!

That was my best single night, but there were many others too, just with smaller wins.

Saying Martingale can't work is the same as saying the Forex can never be a proper business. With the right attitude, education and MM anything is possible.

You are missing very important points.
Who said you have to double up? You could sucessfully
place a martingale trade with a multiplier of 1.15.
Example: Buy every 10 Pips with 30 Pips Take Profit each 10 Levels.


Martingale IS MM.

Who are you replying to? You're quoting me without even reading my message. Nothing I said went against Martingale.

I was replying to johnedoe and read your post after I submitted mine.
Just wanted to add that to the bottom of my post.

martiangel need big equity. use SL and TP more profitable

No betting system, including Martingale, offers any advantage in expectancy over flat betting, over any number of trials (trades). Prove the laws of math wrong, by using the attached XLS. Go on, you know you want to! Type any combo of values into the yellow cells that result in the gray cell (AK8) having a non-zero value, and re-post the XLS here. Then ready yourself to accept the relevant Nobel prize. (all assuming there are no bugs in the XLS formulae )

Let's apply some common sense. If it were possible to gain any kind of edge at roulette by varying bet size, everybody would go down that path, and casinos would be forced to change the game mechanics. It's not just the table limits that restrain you. Sure, they're there to protect the casino against a freakishly lucky punter, but even without those limits Gambler's ruin (http://en.wikipedia.org/wiki/Gambler's_ruin) would ensure that virtually every punter who doubles up repeatedly will end up bankrupt long before the casino does.

Similarly, if it were possible to win at forex by using MM (i.e. varying position size or method) alone, then surely every trader would go down that path, and become a millionaire. The 'willing-buyer, willing-seller' and zero sum game constraints provide proof that this is an impossibility.

Averaging down will give you +EV to whatever extent the probability of price reverting to a mean improves, the further it trends away. Increasing one's bet size as probabilities improve is smart (ask any Blackjack pro). But increasing one's exposure without limit is guaranteed to eventually result in irretrievable drawdown.

Here's the rub. Let's say we average down, doubling our pos size every 10 pips. If price moves 80 pips against us, that's 1+2+4+8+16+32+64+128 = 255 units at risk. So we need odds of 255 to 1 in our favor that price will reverse 10 pips, before it falls another 10, to justify that bet size. The problem is that our bet size is based on our desired account balance, and is no longer commensurate with the probability that's being offered by the market.

And let's not forget that costs must be overcome. With a 2 pip spread, you'd need price to move 12 pips in your favor (TP) before it moves 8 pips against you (SL).

Ironically, Martingale guarantees a 100% win rate, until you lose everything. The only consolation is that the losing sequence that causes the 'death trade' might not happen during your lifetime. Irony #2 is that the longer you live, and the more often you trade, the greater the probability that you'll encounter it. There might be several hundred chambers in the gun, but you never know which one holds the bullet. Maybe it won't happen until the year 2030. Or maybe it will happen next week.

There are lots of ways to mitigate the effect of the Martingale:

1. You can use a less severe betting progression (e.g. Fibonacci steps, d'Alembert, etc), but then it takes more than 1 win to return to breakeven, i.e. you cripple your recovery rate. You need multiple wins without intervening losses, making recovery more difficult. So you're effectively offsetting one type of risk with another.

2. You can start with a nanolot pos size, but then your return is reduced in proportion with the risk. Not much point in having a $10k account and winning only 10 cents per trade. (Then, after (say) 10 losses, having to risk $204.80 just to make 10c!)

3. You can cut your losses and start over, e.g. after 5 losses, but then you're back to sizing at 1 unit, so you'll need 1+2+4+8+16 = 31 consecutive wins just to recover to breakeven. Or after 6 losses, you'll need 63 consecutive wins. And so on. Just another way of crippling the recovery rate.

4. You can withdraw winnings from the account periodically, but in doing so, there's less funds there to provide a buffer against a prolonged losing streak, thereby increasing the probability of a margin call. And there's no guarantee that you'll make the withdrawal before the death trade occurs.

The conclusion is: In any activity that involves uncertainty, return is always somehow commensurate with risk. To assure a 99.99999% win rate, you have to take an 0.00001% risk of losing everything. The innate, perfectly balanced equilibrium ensures that there's no exploitable loophole, no free lunch.

Anybody can get lucky in the short term. And maybe more than once. But I assume that most folk here aspire to be traders rather than gamblers. Having said that, there's no law against using forex as a vehicle to gamble, and with more leverage than any casino will ever offer you, should you choose to. Good luck!

A well constructed answer.

It seems like most of you just copy stuff about gamblers ruin and how you need unlimited capital in a casino and all that stuff from wikipedia which is fine. You seem to forget that we are talking about trading. In trading there are 2 clever ways to improve your probablities in a MG trade while keeping risk the same.

Most of you probably never used Martingale as MM which makes the discussion a bit obsolete. Hey I wouldn't talk about sex if I was still a virgin. For instance take my first post find me an argument against it you won't find any. Martingale 1 Traditional Position Sizing 0

EDIT: Ahh damn I commented alot in you quote but the forum software cut it out...

@Tjpld:

It seems like most of you just copy stuff about gamblers ruin and how you need unlimited capital in a casino and all that stuff from wikipedia which is fine. You seem to forget that we are talking about trading.
With respect, the MM math is the same with both casino games and trading. Math is math. I agree with you 100% when you say that Martingale is pure MM. We are on the same page.

It's true that the XLS effectively assumes a 1:1 RR trade. But it could be modified easily enough to simulate variable RR (e.g. average win amt, average loss amt). Provided the probability of the TP being reached before the SL in an X:1 RR trade is 1/(X+1) – which it generally is, everything else being equal – the result would be the same.


If I wanted to Risk 5% on a LONG trade why shouldn't I start small and set further Buy Limits with higher lotsizes lower and a stop level for all of them so when all my levels are stopped out I end up with a 5% loss.
If I understand this correctly, you limit your loss to 5% on any trade (sequence of entries), so you're protected against a 'death trade'.


Wrong. Just take my last post with 10 Levels 1.15 Multiplier buy every X Pips with a TP of 3X. If you lose on the 8th Level you can cover that loss with a 1 Level trade. It's all in the Multiplier if you do the math.
Yes, agreed, I can see that:
1.15^8 = ~3.05, and your TP = 3x your SL (3:1 RR)
So you're counting on a 30 pip retracement before price moves 9 levels or 90 pips against you, otherwise you bail out and lose 5% of your account..... that's if I understand you correctly.
If this price behavior occurs often enough, you have positive expectancy. Nice.

You could use a higher or lower number than 1.15, and/or vary the TP/SL ratio. (A lower RR improves the probability of the TP getting hit first, for example). That's somewhat akin to what I meant by "effectively offsetting one type of risk with another." Presumably you've arrived at the 1.15 and 3:1 RR by optimizing the numbers for the pairs/setups that you trade?


Hey I wouldn't talk about sex if I was still a virgin.
Now that I'm married, I hardly talk about sex at all, LOL.

With 1.15 and 10 levels with 3:1 RR your first 8 Levels turn out in profit the last two result in a small loss wich can be covered by one successful 1 Level trade. The Level that turns out with the most profit is Level 5 which is right in the middle of your 10 levels and which will occur the most often (did some statistics on that) if you use a 1.333333333... multiplier in that case every level will be profitable but the first levels get smaller in size because you want to keep the risk the same. So it makes sense the adjust your multiplier in that way that the level that occurs the most makes the most profit. So you can get more profit with the same risk just by optimizing your multiplier.
In addittion to that you could close your first level and partial close the rest of your open levels when you are in Level 8 which would end up with a small loss after its TP is hit. Now you can add one extra level on top because the little retrace compensated the loss of your first level and made a partial close of the rest possible without experiencing a loss but you gain one extra level that way and now you're back in Level 7 which will end with a profit. Think of it like crossing a sea with just 2 floating boards you always take the one behind you which you don't need anymore and put it infront of you. Again no extra risk just higher probability. And the most important thing when you compare martingale in Casino and Trading is that in Trading you can enter on allready HighProbability Setups which will help you alot.

People can argue all they want about whether or not it works. I just use it and it always works for me. There are standard casino wagers with a 62-63% win ratio, and that's just as a straight bet without MG added in. A little education (and a lot of patience) goes a long way.

Just wanted to add

With all due respect to Hanover, I'm with Tjpld on this one, It's easy to present static examples on why averaging down doesn't work, but IMHO it's not too hard to also think of ways to mitigate most of the pitfalls, however it is certainly no silver bullet against randomness. In the end it's about finding something that works, often this may fly in the face of popular opinion or what is obvious.

I can see how Martingale could work if the next trade following a loss had a higher probability of winning. So for instance, you identify a support area in an uptrend that you have deemed a good value buy. Your order is hit but quickly stopped out as price moves down to the next support level, which you also deem to to represent value, and now price is oversold in an uptrend so you weight your trade according to the greater probability of success which you have observed and identified. That is logical, and would yield a greater reward in the long run than a non-weighted MM policy assuming the observation that the next trade in a string following a loss has a greater probability of success is correct.

Increasing your bet-size because you need to make back the loss is not logical.

Glad we cleared that up.