[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Bet sizing question: Scaling



PureBytes Links

Trading Reference Links

Yes, pretty much agree with everything you have stated. It has been conclusively proven by mathematicians that you can't turn a negative expectancy method into a profit in the long run by use of bet sizing.  It has also been proven mathematically, that you can go broke with a +ve expectancy method if you manage your bet sizing badly.  Mathematically, if you bet more than twice the optimal size bet, you will go broke in time, even with a winning method. This is one of the many reasons why the markets are very good at transferring the wealth from the many to the few.  

Adrian

-----Original Message-----
From: unicorn@xxxxxxxxxxxxxxx [mailto:unicorn@xxxxxxxxxxxxxxx] On Behalf Of Alex Matulich
Sent: Monday, 30 July 2007 3:27 PM
To: omega-list@xxxxxxxxxx
Subject: Re: Bet sizing question: Scaling

Adrian,

>What Van says is basically correct, but how would he comment on
>these situations:
>
>1.  A system that scaled in 1-3 longs and then scaled out of 1-3
>longs.  In other words, sometimes there would only be 1 buy signal
>and therefore only 1 exit signal.  Sometimes though the market
>keeps dropping and we accumulate 3 positions and have 3 different
>exit points.  Therefore, yes, we will have our biggest loss when
>holding 3 contracts. Van I assume would conclude this was a poor
>system.

To Van Tharp, a "poor system" is one that doesn't exhibit a positive
expectancy.  This is independent of the position sizing strategy.
He implies that a poor position sizing strategy can convert a
positive expectancy system to a negative expectancy system.

Based on the passage I quoted from the book, he would not conclude
this is a poor system; rather he would recommend testing the system
(which is assumed to have a positive expectancy) with all-or-nothing
trades (enter 3 contracts, exit 3 contracts) to determine which is
better, scaling out or getting out.

>2.  Exactly as above but now lets assume we have 3 trading models:	
>	a. Buy 1 contract on a dip and exit on a rally;
> 	b. Buy 1 contract on a larger dip and exit on a rally;
>	c. Buy 1 contract in a BIG dip and exit on a rally;
>
>So in 2 we actually have 3 trading models which can be assessed
>in their own right.  If all were profitable and acceptable, then
>would trading all 3 within your overall portfolio risk parameters
>be considered bad in Vans opinion?

Not at all.  Three models being run independently in the same market
would give the appearance of trades being scaled in and out, AS WELL
AS give the appearance of the same trade being reversed, depending
on the sequence of signals.  So what?

I believe Tharp would not consider this bad if all three were
profitable and treated independently.  No scaling out actually
occurs; the confluence of three independent sets of signals only
makes it appear that way.

What he objects to is taking a *single* profitable model and
applying a bet-sizing strategy that starts out with a large number
of contracts and scaling out as your profit increases.  It's pretty
obvious if you think about it (and perform simulations) that such
a position sizing strategy dramatically reduces your total return
due to having large holdings when stopped out at a loss, and
successively smaller holdings when intermediate targets are hit.  He
recommends getting out when your system says so, or instead tighten
your stop way down to let profits run further or stop out all the
way.

-Alex

28/07/2007 3:50 PM
 

28/07/2007 3:50 PM