Martingale debunked by Gary

[garyfritz]

Martingale has never seduced me ......however "Elagnitram" has essentially the opposite of martingale with a twist!

That's more commonly known as "anti-Martingale," Jonathan.

It can work, if your strategy really does have long streaks of wins and, more importantly, long streaks of losses. It's a variant of the "trade the MA of the equity curve" idea. For most strategies, which don't have long streaks, you lose more than you save.

See http://www.stevehopwoodforex.com/phpBB3 ... 780#p28780 for an example.

[Jonathan2FI]

Martingale has never seduced me ......however "Elagnitram" has essentially the opposite of martingale with a twist!

That's more commonly known as "anti-Martingale," Jonathan.

It can work, if your strategy really does have long streaks of wins and, more importantly, long streaks of losses. It's a variant of the "trade the MA of the equity curve" idea. For most strategies, which don't have long streaks, you lose more than you save.

See http://www.stevehopwoodforex.com/phpBB3 ... 780#p28780 for an example.

Gary,
yes I can see that --- it does not work in the long run on many strategies it works on very few - at least the spread sheet attached gives a statistical basis to use this trade management technique or not Jonathan

[bfis108137]

First of all this is my first post on the forum. Thank you Steve and all of the contributors for everything.

This is not only related to Martingale but it certainly impacts it. I have written a small program outside of Empty4 that is basically a random coin flipper with a fixed percentage. In other words in simple terms, it flips a weighted coin therefore simulating a so called sinning percentage. It will then tell you how many streaks there were of wins and both losses for your result. This is relevant to martingale as well as anyone who thinks they have a winning system. It will tell you what your chances are of actually getting those results by just luck. I don't know if any of this makes sense but just to give you a taste I will post some results and you will see why Martingale will NEVER work over the long run. I will run 500 simulated trades with a 60% and 70% winning system. Here is a spreadsheet that I made with the results. Every time I see anything written about streaks, martingale, or winning percentages I break this little program out. As you can see, even in 70% win situation you can reasonably expect to have 1-3 losing streaks of 5 or more and I am pretty sure that most of us don't have a 70% winning system so at 60% you can reasonable expect 1-3 losing streaks of 6 or more. This should be enough to forget about Martingale. Of course it will come back I am sure.

60%

first run
4 losses in a row

[cjgovi]

http://www.myfxbook.com/members/govigov ... ing/545992

[garyfritz]

Nice results, cjgovi, as long as it works.

You can get beautiful results with a Mart. Until it blows up in your face.

Looks like in a month you've gotten six level-4 (0.16 lot) trades, and three of them went to level 5 (0.32 lots). How much drawdown do you sit through when a 0.16 goes against you? How much can you stand? What if it goes 6, 7, or 8 levels against you?

[danielv]

So the first thing we need to do is increase the chance of winning right, about 60% is okay? And secondly use higher lot ? 2% of our equity maybe

[SteveHopwood]

So the first thing we need to do is increase the chance of winning right, about 60% is okay? And secondly use higher lot ? 2% of our equity maybe

Do yourself a huge favour.

Open a live (i.e. nor demo) mini account with Cowboy IBFX. Deposit $100. Cowboy IBFX mini accounts are actually nano accounts, so it is perfectly safe to trade 0.01 lots per $100. Only an idiot blows such an account.

Then trade your Marty thingy until you blow your account and feel an idiot.

Then start taking some notice of guys like Gary here and stop losing money.

The lesson you will learn will only cost you $100. It cost a lot of the rest of us a lot more. You are one lucky dimwit. I wish Gary had been around when I started trading. And Bob, CJ, various Georges, the occasional Phil.........................................................

Stop being so stupid, fella. Or carry on being stupid but don't expect sympathy when the results of your stupidity do poop on your doorstep.

Good luck. You need it, but you will not get it. In Forex trading, if something that is bad for you can happen, then it will happen, and probably sooner rather than later. You are going to lose all your money.

[garyfritz]

Daniel, did you read ANY of the posts I made in this thread!?

1. Increasing your winning % reduces the chances of a blowup. It does not prevent them from happening, it just makes them happen less often. It WILL still blow up if you trade long enough.

2. I just did a quick simulation of a 60% winner. In a few-thousand-trade test, it hit several cases of 10 losses in a row. If you double your position size on each loss (bet 1, lose; bet 2, lose; bet 4, lose; bet 8, win; +1 profit total), that means a 10-loss series results in a bet size of 2^10 = 1024 TIMES your initial bet size. If you started out with 0.01 lot, you'd be risking 10.24 full lots. All to win a 0.01-lot-sized profit. Does that sound like a sensible Risk:Reward to you? And nothing prevents you from getting streaks of 11, or 12, or more losses. Either you trade with near-infinite funds (in which case your 0.01-size profits are a waste of your time) or you risk blowing up your account.

3. 6- and 7-loss streaks are common with 60% wins. 6-loss streaks mean you'll be betting 2^6 = 64 times your original bet size. I'll let you figure out how well that works if you start out with 2% risk.

Martingales make a system look unbeatable, but it's a mirage. A Martingale doesn't turn a losing system into a winner; it basically just delays the losses. And then those stored-up losses hit you all at once. Over the long haul a strategy traded with & without Martingale sizing will return roughly the same profits/losses, assuming you trade with finite funds and a bail-out point -- but the non-Martingale hands you the losses in manageable and predictable bites.

You can't limit the size of losses a Martingale can hand you. All you can do when it goes against you is to puke the position and take a huge loss.

You are MUCH better off to trade your higher win-% strategy with a simple non-Martingale fixed-size strategy. You won't have the pretty ruler-straight equity curve, but you won't risk getting bankrupted either.

Don't believe me? Here are some real numbers. This is a 5000-trade simulation of a 60% winning strategy. 60% of the time you win +1, 40% of the time you lose -1. The blue line is the result of trading 1 unit per trade, so after 3 wins and 1 loss you have a profit of +2. The red line is the result of trading with Martingale sizing, doubling the bet size after each loss.

Mart1.gif

Pretty, right? The Mart has a nice straight equity curve, 3x more profit than the non-Mart case. Sure, there were one or two uncomfortable moments, but it all worked out in the end.

But that assumes you have infinite money and you never have to crap out. Now let's say you have a more realistic account, and you can't stand to risk more than about 100x your original bet size. So after losing 7 in a row, instead of betting 128x on the next trade, you crap out and take the 1+2+4+8+16+32+64 = 127x loss from the previous 7 losing trades. That puts some dents in the pretty ruler-straight Martingale equity curve.

Mart2.gif

But hey, the Martingale still came out ahead, right? So it's worth sitting through those stomach-churning equity drops to get the higher profit, right?

WRONG. In order to get that higher profit, you're risking as much as 64x times your original bet size. Take a look at what happens if you trade a non-Martingale position, with only 2x the original bet size on EVERY bet:

Mart3.gif

So with 0.02-size bets, you make about as much profit as the Martingale makes with as much as 0.64-size bets. WITHOUT the gut-wrenching losses and ridiculous risks.

AND, since the 0.02-size non-Mart bets have a fixed and known risk size, you can actually scale them up much larger. If you determined the Martingale's 0.64-lot bet was your maximum tolerable risk, imagine what would happen to your profits if you traded a NON-Martingale strategy with, say, 0.10 or 0.20 or 0.50 lots per bet...

Let me say this very slowly and clearly: MARTINGALES DO NOT WORK.

Period. Full-stop. End of story. MARTINGALES DO NOT WORK.

If you don't believe me, do like Steve says. Go bloody your nose with a Martingale and THEN maybe you'll pay attention when somebody who's studied the thing tries to warn you.

[garyfritz]

Duh. I could have saved myself a lot of time building that simulation...

I forgot I posted a Martingale simulator in the first post of this thread. Daniel, go load up that simulator and play with any Martingale parameters you want. Change the win%, the win/loss size, the Martingale scale-up progression, the bail-out point, any brilliant idea you have. You will find that a modest-sized FIXED position will do FAR better than a Martingale in just about any situation you can think of, and it does it with far lower risk.

At the risk of repeating myself, MARTINGALES DO NOT WORK.

[gaheitman]

Gary,

You have have the patience of a saint. I still can't believe people post pro-martingale positions in a thread with this name. It's like they are shouting "Nuh-uh!" :>

George