The Dangers of Martingale strategies

[FetteredChinos]

been tinkering with position-sizing on my basic method...

attached is an equity curve trading 1 position normally, and then doubling up on the next trade if the former was a loser.

as you can see the equity curve was nice and smooth for years, but then wabam, we hit that Black Swan as it crosses the road..

ended up trading 256 times normal stake. yes it recovered, but the drawdown could/would have wiped you out

this curve is based on my normal method which is usually over 70% accurate, so you can see what trouble trading a trend-following method could result in, with its hit rate <40%.


just wanted to highlight the dangers of martingaling.

it is big, but its certainly not clever.

FC

 

[FetteredChinos]

and this is what happens with a strat nearer 50% accuracy....

anyone still fancy it??

 

[juanbyte]

What happens to the equity curve if you limit the number of times that you double up say to 2,3, 5 etc.

You will limit your drawdowns, but how will it affect your profits?

I do a binary method and double up only once then stop this has won 86 times out of the last 100 trades.

 

[FetteredChinos]

hmm interesting.. i will have a look..


curse you juan, i was trying to prove to myself that disproportionately increasing stake size after a loser was a bad idea.

will have a closer look.

i'll try it on both methods...

fc

 

[Roberto]

i was trying to prove to myself that disproportionately increasing stake size after a loser was a bad idea.

Might depend what you mean by "disproportionately"? If your overall success-rate is high enough, it seems fairly harmless (famous last words!). I sometimes do the same thing myself. Thanks for the *.

 

[FetteredChinos]

ok, having tested with the 76% strategy, unsurprisingly the position size doesnt affect things too much. i tinkered with all sorts of sizing (eg only doubling/trebling-up after a losing run of a few trades etc)

the net effect on the equity curve was negligible.

as for the 50% strategy - this aint pretty, J-ster..

 

[juanbyte]

The higher the number of the losing run the less you can stake per trade.

Hope you can make sense of the spreadsheet.

http://www.trade2win.com/boards/attachment.php?attachmentid=12278&stc=1

 

[FetteredChinos]

hmm yes, but the equity curve is going bananas though isnt it?

 

[juanbyte]

Added a chart

The longer the losing run the larger the drawdowns

http://www.trade2win.com/boards/attachment.php?attachmentid=12280&stc=1

 

[FetteredChinos]

so it kinda backs up my idea, that martingaling is ok for strats which have a very favourable hit rate (as the losing run will be a lot lot shorter)

but for those that are below 70% (perhaps higher) then it is Wipeout time sooner or later...

still not going to do it myself though!

fc

 

[shogun]

one thing you should consider if using martingale type management is your winning position should be at least the same as your loosing position , and if you had a strategy which was 70% win rate that could give you a winning position the same size as your loosing position you would not need to use martingale,

i spent a bit of time last year applying this to several strategies and my conclusion was , to risky.

 

[shogun]

just read my post , talk about fussy,

what i mean is if your max loss per trade is £100 then your minimum profit per trade must be at least £100 in order for martingale to work.