|
PureBytes Links
Trading Reference Links
|
Thanks for the reply, Paul. One sentence really stands out for me...
"Testing needs to be done using methods that can really be implemented.".
This sounds simplistic, but it is sooo true in all aspects of system
development and money management, and IMHO, is often overlooked in quest for
the ultimate equity curve...
----- Original Message -----
From: "Paul M. Zislis" <pzislis@xxxxxxxxx>
To: <omega-list@xxxxxxxxxx>; "Ray Gurke" <Ray@xxxxxxxxx>
Sent: Sunday, August 25, 2002 6:24 PM
Subject: Re[2]: Geometric Capital Growth / Optimal-f
> Replying to your message of Sunday, August 25, 2002, 5:07:33 PM,
>
> > Okay... but what, if any, is the "practical" upshot of this if your
primary
> > trading vehicle is a portfolio of futures and you're a regular Joe with
an
> > account size of 100K or less? The hypothetical account size/growth
achieved
> > through Fixed Fractional or Fixed Ratio looks splendid on paper ...but
have
> > you ever tried placing an order to trade 1.6 futures contracts --
because
> > that's the current bet size/percentage your money management strategy
calls
> > for? It appears to me that to follow/utilize either a fixed fractional
or
> > fixed ratio approach your account size would have to be HUGE ...like
the
> > ability to trade 10 - 100 contracts of every item in your portfolio,
> > depending on how closely you intended to follow the bet size gradient.
This
> > just sounds like a statistical pissing match to me, unless your primary
> > trading vehicle is stocks, or you're a professional CPO, CTA or Hedge
Fund
> > manager with a fair amount of money under management. Am I missing
something
> > here?
>
> It isn't necessary to trade a huge account. You need to take as
> position size the floor (greatest integer number of contracts less
> than or equal) of the size calculated by the position sizing method.
> You use the floor, rather than rounding, to ensure that you never
> exceed the estimated risk level you have established for yourself.
> Testing needs to be done using methods that can really be implemented.
> Since you can only trade an integer number of contracts, you do that.
> Your equity grows a little slower because you traded 1 contract
> when your position sizing method came out to 1.6. If you manage
> to keep equity growing you will reach the point where your position
> sizing method tells you to trade 2.0 or 2.1 contracts and then you
> start trading 2 contracts. Etc.
>
> Another issue related to hypothetical account size/growth is worth
> commenting on. The equity growth that is frequently shown when
> position sizes increase with profits is absurd. Humor starts: At some
> point in the march to trillion dollar account equity, you are going to
> find out that trading liquidity dries up, causing slippage to increase
> dramatically or requiring special methods for getting in and out of
> positions. Or perhaps you'll decide that you only need a few billion
> so you won't continue trading at some point. Humor ends. If you have
> more realistic goals such as growing account equity at an achievable
> rate and an acceptable level of risk, being able at some time to live
> off your futures trading profits while continuing to (more slowly)
> grow account equity, then testing methods of equity withdrawal in
> order to understand how it impacts equity growth is worthwhile.
>
> Paul Zislis
>
> > ----- Original Message -----
> > From: "Paul M. Zislis" <pzislis@xxxxxxxxx>
> > To: <omega-list@xxxxxxxxxx>; "_Craig" <craigbud@xxxxxxxxxxx>
> > Sent: Sunday, August 25, 2002 11:24 AM
> > Subject: Re: Geometric Capital Growth / Optimal-f
>
>
> >> Replying to your message of Saturday, August 24, 2002, 12:16:32 PM,
> >>
> >> > I am trying to work out a good way to maximize the geometric growth
of
> >> > capital in a fixed portfolio of trading systems and stocks.
> > Traditionally,
> >> > I have traded using a fixed equity model and the profits were not
> >> > invested but withdrawn.
> >>
> >> > I could take the empirical optimal-f for each market in the portfolio
> >> > and then readjust the bet sizes as the capital grows.
> >>
> >> Optimal f gives optimal growth based on historical results. But,
> >> since the future is likely to be somewhat different from the past,
> >> trading at optimal f may well cause your account to go bust. To
> >> reduce the likelihood of going bust you should consider backing down
> >> from optimal f to a less "optimal" (but also less risky) level of
> >> trading. You might also want to consider the level of f (or Delta, in
> >> the case of Fixed Ratio) to trade at based on a portfolio level
> >> analysis.
> >>
> >> > Perhaps it's better to grow the portfolio as the equity grows by
> >> > adding new markets and systems.
> >>
> >> Diversification through adding systems and markets can be useful in
> >> reducing risk if your portfolio is not diverse enough. If you think
> >> you have sufficient diversity then I would study position sizing
> >> strategies such as Fixed Fractional and Fixed Ratio.
> >>
> >> > Is there a methodology that shows an optimal or best way to simply
> >> > allow the geometric growth of capital (re-investing profits) to reach
a
> >> > certain size and then withdraw a specified amount of capital to
reduce
> >> > it to a "minimum size", and do this over and over again every time
> >> > the capital reaches its "maximum size"?
> >>
> >> I have been testing position sizing strategies that look at this
> >> issue. I define the following portfolio money management parameters:
> >> 1. withdraw level (when profit reaches or exceeds this level I
> >> withdraw some equity from account)
> >> 2. withdraw amount (this is how much I withdraw when I reach the
> >> withdraw level)
> >> I keep track of and report the number of withdrawals I've made, their
> >> dates and the total amount withdrawn. I subtract the amount withdrawn
> >> from a global variable that represents current account profits.
> >> Position sizing decisions are partially based on this global variable
> >> so that I automatically start trading fewer contracts after an equity
> >> withdrawal.
> >>
> >> Since the testing platform I use allows me to run tests on a portfolio
> >> using iteration of money management parameters, I am able to test the
> >> effects of varying withdraw levels and withdraw amounts.
> >>
> >> Lest anybody misunderstand, I am a trader, not a vendor. I've put a
> >> lot of energy into building and/or buying tools that I find useful for
> >> my personal trading analysis but I do not sell any such tools.
> >>
> >> Paul Zislis
> >>
> >> > Thanks,
> >>
> >> > -c
> >>
>
|