|
PureBytes Links
Trading Reference Links
|
I've been spending a little time reading some of the trading "classics",
and one of them is Welles Wilder's "New Concepts in Technical Trading
Systems" (c) 1978. Wilder advocated the following money management
rules:
0. Trade the 6 commodities that have the highest Commodity Selection
Index (descrbed below).
1. Don't margin more than 15% of total capital on a any one commodity.
2. Don't margin more than 60% of total capital at a time.
Although 1. sets the limit of 15% of total capital per commodity, Wilder
recommends not committing more than 10% of capital per commodity.
The commodity selection index (CSI) is calculated as follows:
CSI = ADXR * AvgTrueRange(14) * K
where:
K = (V / SquareRoot(M)) * (1 / (150 + C)))
which simplifies to:
K = V / (SquareRoot(M) * (150 + C))
ADXR => Average Directional Movement, which is given by:
ADXR = 0.5 * (ADX + ADX[14])
V = Value of $1 move in the commodity (Wilder says, or the "basic
basic increment of ATR(14) in dollars")
M = Value of Margin in dollars
C = Commission in Dollars
I'm a little hazy on the definiton of V. I think, for example, on
the S&P the value of V would be $250 (BigPointValue). I'm not sure
what Wilder had in mind though when recommends dividing the value
of V by the square root of the Margin.
The value of 150 above is some sort of approximation of slippage, since
it is added to C, the Commission cost.
The basic idea is to combine "trendiness" measured by ADXR, with
range of movement (ATR), and then to somehow take advantage of
the "lag" in adjusting margin to current volatility. Here's what
Wilder had to say (p. 115):
"I realize that what we're dealing with is not an exact science. The
margin requirements will not be set, nor could they be set to maintain
a constant relationship with volatiility, or directional movement,
nor any other variable, for that matter. The is however a direct -- though
not constant -- relationship between margin requirement, volatility,
and directional movement. The CSI constantly all of these factors
and points out the most advantageous situations."
"As a rule, margin requirements lag market action. They are slow to
go up and slow to go down. The Commodity Selection Index also
enables the trader to take advantage of this lag and obtain
the best return on invested capital."
Wilder goes on to mention that trend following systems will likely
be right only 30% of the time, and wrong the other 70% of the time
and therefore it is important to be in the right (trending) commodities.
A high ADXR rating will point out commodities that have been trending
recently, and the theory is that they will continue to trend in the
near future.
Wilder's method, method, as described above has a weakness in that
he sets no bottom limit on the value of ADXR, so it is possible that
it could select commodities that are not currently trending. This
might be easily fixed by only looking commodities with an ADXR of, say,
20 or above. Also, it appears that Wilder assumes that the underlying
system used to trade the commodities is always in the market. This
might cause a problem (or at least an inconsitency), if the system
used trade trade commodity C, is currently trading the opposite
direction indicated by ADXR. This could be easily fixed by setting
up the system so that it never trades opposite the direction
currently indicated by ADXR.
The point is that the CSI seems to have some of the attributes
recommended by Chuck LeBeau, but for the rules to "make sense",
the trading systems employed have to be in agreement with the
selection criteria.
It would be difficult to backtest the CSI selection method, because
one would need to have historical data on how margin was changed
overtime. Perhaps, a proxy for M (the margin requirement) would be
to use a longer term volatility measure. Some sort of liquidity
factor might need to be plugged in, instead of the fixed "150"
value used above, as a way of approximating bid/ask, or trading
slippage.
--
| Gary Funck, Intrepid Technology, gary@xxxxxxxxxxxx, (650) 964-8135
|