If trend following, can money management save you in chop?

[intradaybill]

Quote from oldtime:

that's true. Or better yet, just increase your lifestyle risk during drawdown and you wouldn't have to adjust your trading risk at all. If it all works out you could always quit smoking to keep everything in balance.

Reducing position size does nothing to eliminate risk of ruin if there is no cummulative exit. It only delays it. Certainly the risk of ruin at any given time is less but over the longer term is certain because the probability as getting a streak to wipe you out is 1.

Let the probability of the event {streak of losers to wipe you out} be z <1

You say that if you risk little then z <<1. True.

Now, the probability that the event {ruin} does not happen in n trials is 1 - z^n

As n goes to infinity, 1-z^n goes to 1, ruin is certain and the value of z does not matter. It will just take longer but it is the sure thing.

If you have a large win rate and very little risk it may be practically impossible to get ruined. But theoritically, if you continue trading for ever, you will get ruined. If you are lucky, you will get ruined too soon.

 

[oldtime]

Quote from intradaybill:

Reducing position size does nothing to eliminate risk of ruin if there is no cummulative exit. It only delays it. Certainly the risk of ruin at any given time is less but over the longer term is certain because the probability as getting a streak to wipe you out is 1.

Let the probability of the event {streak of losers to wipe you out} be z <1

You say that if you risk little then z <<1. True.

Now, the probability that the event {ruin} does not happen in n trials is 1 - z^n

As n goes to infinity, 1-z^n goes to 1, ruin is certain and the value of z does not matter. It will just take longer but it is the sure thing.

If you have a large win rate and very little risk it may be practically impossible to get ruined. But theoritically, if you continue trading for ever, you will get ruined. If you are lucky, you will get ruined too soon.

I'm just kidding with you. Knowing when to take a profit is the happiest but also most perplexing problem I face.

 

[kut2k2]

The crank troll known as intradaybill (or, as I prefer the more accurate name, imbecilebill) is all over the place regarding risk of ruin. First he posted a trainwreck of a thread in Strategy Design attempting to prove the following ridiculous notion:

"As number of trades tends to be very large, risk of ruin approaches certainty asymptotically 1" ... REGARDLESS OF SYSTEM EXPECTATION!

Then he was shown repeatedly how cranky his conclusion was, most notably here:

http://www.nuclearphynance.com/Show Post.aspx?PostIDKey=156954

http://www.elitetrader.com/vb/showthread.php?threadid=231633

Now he's off on a new tangent wrt win rate. Except his new crank theory is based on the Gambler's Ruin Theorem, which has the very specific restriction that W = |L| = 1.

Yes, for even-money payoffs, the win rate becomes the prime determining factor in risk of ruin. BUT for real-world trading systems where invariably W <> |L|, then focusing solely on the win rate becomes yet another of imbecilebill's crackpot obsessions.

And even in this thread, he has obliviously contradicted himself:

"Any system with win rate < 50% has a risk of ruin probability asymptotically equal to 1 as a function of time."

"I have given you a simple mathematical proof that ruin is certain under all circumstances if a target profit for quiting is not set."

Well, if ruin is certain under all circumstances (his original position, let us recall), then what difference does it make whether the win rate is greater than 1/2 or not?

We are looking at a world-class loonytoon here: imbecilebill, aka Crankus Maximus.

 

[The_Tourist]

Quote from N54_Fan:
Final conclusion is win rate effects drawdown and risk of ruin but does not effect profit. You can run these simulations any way you want,...I have and the conclusion is the same. So everyone is sort of correct.

The 2 simulation pictures of different win rates you posted showed the final profits numbers of same colored profit lines(comparing the 2 pics) to be quite different, how did you conclude that win rate which affects streak of losses 'does not affect profit'?

 

[N54_Fan]

Quote from The_Tourist:



The 2 simulation pictures of different win rates you posted showed the final profits numbers of same colored profit lines(comparing the 2 pics) to be quite different, how did you conclude that win rate which affects streak of losses 'does not affect profit'?

Because profit is largely based on expectancy of the system. It is true that you will profit more with a higher win rate. This is a function of have FEWER loser with higher win rate ,...so you lose less of your profit made on other trades. The MAIN effect of win rate has to do with draw down. If you run MC sims on varying win rates and keep expectancy the same you will see varying profits between the green and red lines in those sims. But the thing that is mostly effected is the draw down. Again the lottery is a perfect example of the extreme of this. People often say you can be "profitable" with a system that has a 10% win rate. While this is true you need very high expectancy (like the lottery) but the risk of ruin or massive drawdowns far outweighs the benefits of this system unless you are really lucky (like in the lottery). No one in their right mind would trade a system like that and expect to make a living at it.

So, I was probably a bit too simplistic but the point still stands that one should think of win rate as effecting draw down and risk of ruin. Similarly we should think of expectancy and profit factor as effecting your profit within a given win rate. Obviously, they are all interrelated,...but you get the point.

 

[Archin]

Quote from oldtime:

I'm just kidding with you. Knowing when to take a profit is the happiest but also most perplexing problem I face.

Same here, many times I take profit only to see the stock move big in the direction I was trading...

 

[deaddog]

Quote from intradaybill:

Do you know what is a binomial distribution?

Got me there Bill, I don’t understand
Maybe you’d be so kind as to explain it.

Do you know what starting capital means?

Got you there. I understand starting capital. I also realize that one reason traders fail is being undercapitalized.
But what does starting capital have to do with risk/reward or the % of capital at risk each trade?
If you only risk 1% of your capital on each trade you will never go broke.

Do you know what drawdown is?

Now here is something I really understand. If you can’t stand a little drawdown you shouldn’t be trading.

You are trading aren’t you?

 

[Wide Tailz]

Quote from MADASINHATTER:

Unless i'm wrong (happy to be corrected) weren't they massively overleveraged at the wrong time? My point is,money management can never become less important.Otherwise why do we have a global crisis?
Your argument appears to be the faster you drive the less important the brakes are

Yes, thanks. Please allow me to clarify. My assumption is that a trend following approach will be used, not selling puts, huge leverage spread bets, day trades in thin markets, or other systems that tend to be on the wrong side of a phat tail.

In other words, "money management rules get less and less important with higher win rates for a trend following strategy which trades liquid markets with reasonable leverage and a predefined limit on risk for any given position"

LTCM had no trade abortion strategy. Their model did not require it.

 

[Wide Tailz]

Quote from Handle123:

I have never heard someone say this before , LOL. Great money management rules can make any method much better, and by concentrating on lowering losing percentages, to me, is more important than increasing win rates.

I found the opposite to be true when testing my own trend following strategy.

 

[kut2k2]

Quote from intradaybill:

Reducing position size does nothing to eliminate risk of ruin if there is no cummulative exit. It only delays it. Certainly the risk of ruin at any given time is less but over the longer term is certain because the probability as getting a streak to wipe you out is 1.

Let the probability of the event {streak of losers to wipe you out} be z <1

You say that if you risk little then z <<1. True.

Now, the probability that the event {ruin} does not happen in n trials is 1 - z^n

As n goes to infinity, 1-z^n goes to 1, ruin is certain and the value of z does not matter. It will just take longer but it is the sure thing.

If you have a large win rate and very little risk it may be practically impossible to get ruined. But theoritically, if you continue trading for ever, you will get ruined. If you are lucky, you will get ruined too soon.

Would some kind soul point out to poor ol' Crankus Maximus what his Probability 101 mistake is here? I'm too busy laughing to do so myself.