Why martingale doesn't work in trading

When mentioning MG, most people immediately think of either a scenario as TJPLD describes or, more likely, the practice of doubling each consecutive losing bet until a final win. Then two separate arguments ensue and are challenged and defended in the same debate.

I agree with TJPLDs idea of building positions in increments as it utilizes the concept of "averaging in". Most stock traders would agree. Your positions will be paid for at neither the best or worst price.

On the other side of the proverbial coin, if the trading method has no edge, I go along with Hanover and don't believe any method of MM will turn it positive. A well defined MM combined with a losing method, will give you nicely managed losses.

In the end, you need a small edge, and need to exploit it over and over. And consistent MM will prevent you from going too far in on any single trade and possibly mucking up your small edge.

I wish I had known magicchance when I lived in Reno. There was a handful of people who made their living in the casinos, so it seems he may know something we don't.

Yep I have to agree here.

As for magicchance he is probably paid by the casino. They let him win so everyone see's one happy guy with a bag full of cash walk out which will encourage others to keep on gambling.

As I understand it, MM is straight math, and applies to any 'game', trading included. For example, if I have a system that uses TP=30 and SL=10 (after costs), then I've devised a game that offers a 3-to-1 payoff (R:R). If the game is played when probability of the TP being hit first exceeds 1/(30/10+1) = 0.25, i.e. suitable price setups, then I have positive expectancy. And so on.

[Albeit with these provisos:
1. With trading, the win probability is not fixed (varies with different chart setups), and moreover can only be calculated very roughly through historical testing. Whereas with games like Blackjack and Craps, once we know the rules, we can calculate precise odds, in advance, for every trial.
2. The above assumes simple TP/SL exits. Of course trading allows more complex exit strategies, e.g. trailing a SL, but we are discussing sizing strategies ('pure' MM) here].

@Magicchance: "There are standard casino wagers with a 62-63% win ratio"
I don't understand. It's universally understood that casinos only offer games that they're guaranteed to win in the long term (with the possible exception of Blackjack, which is why they ban card counters). I guess you must know something that (virtually) nobody else does. If enough punters played with that kind of win ratio, casinos would be forced to either ban them, change the game rules, or face bankruptcy. A 62-63% win ratio scenario (assuming a 1-to-1 payoff) defies credibility; if shown to be true it would create headline news.

@Billflet: "I agree with TJPLDs idea of building positions in increments as it utilizes the concept of 'averaging in'. Most stock traders would agree. Your positions will be paid for at neither the best or worst price".
Mathematically, that's the same as entering one position at the average price. But of course one doesn't know in advance whether price will reach the additional levels. Hence averaging down is adding risk (in terms of staking additional capital) for the privilege of getting a 'better' price: another example of risk vs return. Some might suggest: but what if I (premeditatively) limit my total risk to X%? I would argue that the drawback is that only some of the additional levels might end up being triggered, hence return is being reduced, offsetting the lower risk. I have torn my hair out trying to find approaches using averaging in or out that somehow slew the return/risk equation in my favor (i.e. without using some kind of TA). I have yet to discover one.

@Birdt: Agree 100%, at least that's exactly the way I (currently) understand it. MM can't be used to create an edge, but it can enhance an existing one, i.e. by placing additional/larger bets when the probabilities are in the player's/trader's favor. An edge can only come from exploiting situations where price behavior 'breaks through' the win rate/RR equilibrium. Given this, with averaging down the key must be whether or not the further price moves away from a mean, the greater the probability of its returning there. Exactly as you say.

@Craig: I have no problem with averaging down (I've used it myself), provided that one doesn't leave oneself open to unlimited loss. See my comment @Birdt: I've yet to find a way in which a betting/sizing system in itself can create an edge. In other words, for me the issue is whether the same long term expectancy could be achieved by averaging down AND using the same position size at each level.

@Tjpld: Many thanks for sharing this info. You've made me very curious to find out WHY this sizing system in itself can offer an edge (i.e. without any TA). I will conduct more research, and (with your permission) post the results. I'm always willing to eat humble pie, especially if I learn something that will potentially improve my bottom line. But until I prove otherwise to myself, I don’t believe that MM in itself can provide an edge. Can't help my nature, I'm a born skeptic: this little elf keeps whispering in my ear, "if trading profitably is as simple as MM, why isn't everybody doing it?" The big funds could save themselves a lot of money by dismissing their analysts, LOL.

[Footnote] Just thought of another way of expressing this: given three hypothetical tests:
Test #1: average down using random entry points, and same position size at each level.
Test #2: average down using random entry points, increasing pos size at each level.
Test #3: average down using TA-based entry points, and same pos size at each level.

If Test #1 yields +EV, then averaging down in itself provides an edge.
If Test #1 results in EV=0, but Test #2 yields +EV, then it's the MM that's providing the edge.
If Test #1 results in EV=0, but Test #3 yields +EV, then it's the TA that's providing the edge.

Sorry, I've probably repeated myself. Too tired to proof read this post, must be time for bed.

Sure, I'll be the last person to hold you back.

I've done various lives tests in the past and an optimized multiplier and that little partial close trick I talked about absolutly don't add any risk or have other negativ sideeffects. You just get more money with the sames trades and the same Risk.

Hanover: Actually no, it's 2-1 payoff! And everything I said is true. A little more time reading up on roulette and a little less time telling me how wrong I am, and anyone here can do exactly what I have done. And yes, you are correct, that if everyone did what I did the casinos would indeed change the rules! Lucky for me I guess that most everyone else believes the same as you do!

Aha!, given that it's Roulette, I guess it must involve an analysis of wheel bias (e.g. something like this (http://www.rouletteraptor.com/)), or some kind of past-posting sleight of hand in laying down your bet (ah no, on second thoughts, that would give you a better than 63% win ratio); all of this is assuming the widely held view the game can't be beaten mathematically. A 2:1 payoff means that you're either betting columns, or groups of 12 numbers (1-12, 13-24, 25-36). Am I getting close?

I'd do that job!

I've already said exactly what I do, and it takes a little education, a lot of patience, and correct MM. Oh yeah, and brass balls! LOL

Hi Hanover.

I agree with everything you have said above. With regard to the Roulette stuff posted by magic chance... all he is doing is picking a colour and hitting it till he wins. Martingaling each hit so as soon as good ol' mean reversion kicks in (not really mean reversion cos each event is mutually exclusive) you get a winner that overshadows the losers and a you take a step forward. Back to the beginning.

Definately need brass balls. I've tried it when i was younger. Got 9 blacks in a row (i was punting on red). Wow. Angry. Sometimes I would wait till i saw 3 or 4 colours in a row then fade that trend. Again... it all means nothing in actual probability terms cos each event is seperate and random... but hey... whatever. I gave up on that.

EDIT: Forgot to mention you can mix colours and tables but keep the martingale and just bet on the probability of NOT losing 4 times in a row...

MM

No, that's not what I do. And whether or not you believe me, I've lightened the load of quite a few casinos around the world, and I did it using a system no one here seems to believe can ever work. I won't waste my time teaching the unteachable, so don't even ask. You don't believe it can work, so fine with me, just keep on keeping on! It's better in the long run for me anyway. Because the casino is NOT like the Forex: if too many people catch wise, Forex will still be ther tomorrow same as today; casinos, not so much.

MAGICCHANCE,

Making money is making money mate.

If you can teach me something about playing roulette that i dont already know... i would GREATLY appreaciate you taking the time to teach me. Im always open... i have just lost money doing things my way... which was uneducated and unorganised.

Mate, dont take my post the wrong way. Im VERY keen to learn any way of taking money from the casinos. If you dont wanna post to the whole forum... a PM would be cool.

Cheers.
MM

PS: Just saw your profile. Cool.

I noticed that IF, but techincally, every gamble is winnable with one or more IFs and therefore that would have undermined my view.

This is not to argue on who is right and who is not, but I would humbly say that "martingle never worked, it doesn't work and will never work in longer term" is the conclusion I got. There maybe people who conclude differently but giving a second thought to my conclusion would mean revisiting three decades of work.

ps. It all depends how one balances risk and reward. Using Margingale puts time factor in one's favor at the cost of ROI but not the odds. The odds in fixed risk/reward games are always in favor of bookmakers - always.

I understand what you are saying here. And even if I didn't, I'd believe it as your ideas on gaming theory are my "Gold Standard".

You are correct in saying that averaging in only gives you the same as one single position entered at the average price. But we can never know the average price beforehand, so where is the best place to make that single entry? I'm talking about building a position in the direction of the trend. Say you are building a long position and building it by adding on smaller and smaller pieces as the trade goes in your favor (not averaging down). The last piece you buy will be the smallest and if price doesn't go far enough you won't even buy the last piece. If price goes against you it will stop out the smallest pieces first, giving you time to evaluate the first entries (pieces) of the position.

If the last entry points aren't hit, then your call to go long was wrong anyway. If they all fill, then price reverses, you can reduce your risk in increments, maybe moving your first few entries to B/E and suffering a smaller loss on the stopped increments than if you had entered one large postion all at once. Testing the water for sharks with your toe, rather than jumping right in.

Of course, David, you know I'm referring to my newfound fascination--the method we just talked about where a grid is traded in the same direction as a longer term trend and as the trend continues, the entire grid is moved along with it. Tiny entries are built into larger positions with each tiny entry having lots of breathing room.

So your grid is your playing field and as the trend continues, your playing field is pulled along with it. Multiple entries adds a layer of MM that should give your edge a larger sample size to prove itself. That is if each entry is based on a reason and not just an arbitrary time or price.
Entry with no Reason/Edge = no real reason to apply any sort of MM.

I like TLJPD's analogy of clinging to boards in the ocean, always pulling the one that falls behind to the front again.

As an ex-Director of Security for a now defunk (but popular) casino in Reno I can tell you that not only were Blackjack Card counters banned - ANYONE who consistantly won was banned.

While this was back in a time prior to all the closed circuit monitoring they now have (our security personnel watched from catwalks hidden from view of the customers) every once in awhile someone would win consistantly (meaning making a profit) at Roulette. Yes, all casinos had paid shills to play, but mostly to entice customers to sit at empty tables.

Point is, that there were, and are today people who can win at the tables, but they have to bounce from casino to casino (and country to country) and constantly disguise themselves. Magic chance was undoubtably one of these.
P.S. The Nevada Gaming Commission had very strict rules governing the use of paid shills. They could not be dealt winning hands by the dealer (or any other method or means deemed illegal) to give the appearance of winning.

Flamenco!

Nice one... love it. Hopefully we hear a little more from Magicchance on all that. Hows the trading going by the way? I used to follow your posts very closely around the time of Canadian Dude and the FX Prime thing. Still remember when you posted a picture of a Harley and said you paid for it with the last few weeks of trading profit.

Still trading hte same system/style?

Cheers,
MM

Wow- you won't believe this but I just left the old FX Prime (Canadian Dude's system) site. That was a good time when volatility was 3x's what it presently is. Can't scalp anymore - got too burned out and went to higher time frames.
Always good to run across someone from 'the old days'! Hope your trading's going well! And presently my Harley is in Manila under 3' of water.

Yeah mate, same here. Trading longer term price action. Often hourly setups in the context of daily setups. Some cheeky scalping mixed in there as well but only cos im 'forced' to find a way to make budget during my trading hours at work. Gay.

One fine evening in Lugano, Switzerland, I won the lottery, and I did it buying the 72. I only bought a ticket once before.

As a trader who has developed martingale EAs and strategies and who runs a small business trading for a small clientele (one of which is testing a newly developed martingale styled EA, on a live account by the way), I will agree with this to a point. I started my business with a small group of investors and right from the start they all knew my strategies and EAs were very high risk models using these techniques. Yes, I agree there is an extremely high amount of risk to a client and a larger bankroll is definitely needed to make these work. But as long as your client is fully aware of the risks and understands what is needed to make the strategy or EA profitable for them, there is no reason to say you can't use martingale and run this as a business. I'm proof it can work..

still waiting someone with knowkedge to show a way to have any positive expectance with it. As i can see even using a "range of s&R" is not enough. Total R:R is horrible like any martingale. You can have a fantastic hit rate, but a few stops at deeper levels will kill your account.