Reprinted from Technical Analysis of STOCKS & COMMODITIES magazine. © 1998 Technical Analysis Inc., (800) 832-4642, http://www.traders.

com
MONEY MANAGEMENT

Secure Fractional
Money Management
Here’s how to find a new fractional value of capital to invest p = Probability of a winning bet
in every trade to maximize returns subject to a constraint on
drawdown, using a variation of the optimal f money manage- A simple example would be if you had three bets — two
ment strategy. winners and one loser (1, 1, zero) — and you made or lost
equal amounts:
by Leo J. Zamansky, Ph.D., and David C. Stendahl
f = ((1+1)0.666-1)1 = 0.33333
ver the past four years, we have
developed and applied evalua- This formula solves for f. This formula is applicable when
tion software to trading sys- there are only two outcomes. For traders, there are many
tems that use money manage- outcomes. Vince introduces optimal f, and to find the value of

O ment techniques and have stud-

ied the impact of these tech-
niques on trading systems. A
new money management strat-
egy called secure f is one of the
optimal f, we need to maximize what Vince calls terminal
wealth relative (TWR). The problem can be formulated thus:

TWR(f) -> max

outcomes of that work, and where TWR(f)=(HPR1(f))((HPR2(f))(...(HPRn(f))

here, we will demonstrate how HPRi(f)=1+(f((-Return on the trade i)/(Return on the
to find a new fractional value of capital to invest in every trade worst losing trade))
to maximize returns subject to a constraint on drawdown.
Secure f can be a conservative strategy or an aggressive HPR = Holding period return
one, depending on the level of acceptable maximum draw-
down selected by the trader in question. It is a modification Figure 1 illustrates the optimization problem solved by
of the optimal f strategy that was introduced by Ralph Vince optimal f. As we can see from the description of optimal f and
in his Portfolio Management Formulas. Secure f differs from Figure 1, optimal f maximizes the final equity by investing
optimal f because it takes historical drawdowns into account the right amount in every trade. This amount is f% of the
and uses information about the prices of the underlying existing equity at the time the trade is initiated. To find the
security. value of optimal f, the calculations are applied to a set of

OPTIMAL F
Optimal f is a money management strategy that can be used
to improve and maximize system performance by finding the
best percent of capital to invest in each trade. This strategy
determines which percent of equity invested in a trade would
have yielded the highest return based on a sequence of past
trades. Because traders are able to employ a variety of money
management strategies, it can be useful to know what would TRADING
have been the optimal amount to invest in each case. SYSTEM
The concept of percent or fractional strategy itself comes
from the Kelly formula, which estimates the percentage of
your capital to trade when the amounts won and lost are not
equal:

f = ((b+1)p - 1))/b

b = Ratio of the size won on a winning bet to the size lost on FIGURE 1: OPTIMAL F. The goal is to use optimal f to manage the trading capital
a losing bet so that the terminal wealth relative is maximized.
Reprinted from Technical Analysis of STOCKS & COMMODITIES magazine. © 1998 Technical Analysis Inc., (800) 832-4642, http://www.traders.com

historical trades. Trade history

should be profitable; otherwise,
neither optimal f nor any other
strategy will turn a losing strat-
egy into a winning one. The
longer optimal f is used, the more
final equity will result from its
application.
As an example, imagine play-
ing the following game. This
game can be viewed as simpli-
fied futures trading. You toss a
coin in three series, with a total of
11 times each. You pay or get
paid only at the end of each se-
ries.
To play the game, you need to
buy one or more contracts. One
contract price is, say, $10,000. A
coin-toss is equivalent to a one-
point move of the contract. A
contract move of one point is,
say, $500. If you paid $10,000 for
one contract and won eight times
and lost three, you made $2,500.
Or if you paid $20,000 for two
contracts and won seven times
and lost four, you made $3,000. In
the game you played, for each
contract purchased you won six
times out of 11 in the first and
second series and lost six times in
the third. This gives you the re-
sults of 1, 1, zero — you won
twice and lost once.
The following are the results
in each series of your game:

Series 1={1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1}
Series 2={0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1}
Series 3={1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 0}

Now, let’s say you have

$100,000. Let us find the optimal
f based on the outcomes 1, 1, and
zero. You had trades of $500,
$500, and -$500:

HPR1=(1+f(500/500))
HPR2=(1+f(500/500))
HPR3=(1-f(500/500))

That gives us the expression for

JOSH GEORGE

TWR:

TWR=(1+f)(1+f)(1-f)=1+f-f 2-f3
Reprinted from Technical Analysis of STOCKS & COMMODITIES magazine. © 1998 Technical Analysis Inc., (800) 832-4642, http://www.traders.com

DRAWDOWN VS. TWR

120 25000

100
20000

Max Drawdown ($)

80

TWR ($1000)
TRADING 15000
60
SYSTEM

SECURE F CALCULATOR (RINA SYSTEMS)

10000
40
Maximum Acceptable Drawdown
5000
20

0 0
0 10 20 30 40 50 60 70 80 90
f Value (%)

FIGURE 2: SECURE F. The formulation of the problem that includes drawdown is FIGURE 3: RELATIVE COMPARISONS. Here are the terminal wealth relative and
set to a limitation while maximizing terminal wealth relative. maximum drawdown versus f value. As the f value increases, the maximum
drawdown also increases. The optimal f peaks at 33.3%. With the maximum
drawdown set to $5,000, the secure f values peak at just over 20% of capital.

This function reaches the maximum at f = 1/3. This means that Finally, optimal f does not take into account the se-
optimal f tells you to invest in every trade a third of your quence in which the trades take place; it does not consider
money, which for the first trade would be $33,333. equity drawdown.
Calculate your maximum loss during the game using These issues lead to the very limited use of the optimal f
optimal f. That $33,333 will let you buy three contracts, and strategy in the trading community. This is the key: How much
should you have invested in the game if you had had a limit
on your maximum loss? It could be in absolute value (in
Secure f money management strategy dollars) or relative value (in percentage). If your limit is
offers traders the ability to maximize higher than $7,500 — say, $10,000 — nothing has to
the return subject to the level of risk change, and optimal f is the solution. However, if your limit
on your maximum loss during the game were, say, $5,000,
they are willing to assume. then you could have only two contracts instead of three. The
same is true if we assume that the limit on your acceptable
in the first series, say you win $500 per contract for a total of maximum drawdown is 5%. That means that for your
$1,500. Your total capital after the first series is now $101,500. preference in this game, the optimal percent of the capital to
Again, you use a third of your capital, which is now $33,833, invest would be 20%.
which lets you buy three contracts yet again. This fraction is what we call secure f. Every trader uses
Because you lose $500 for every zero in the series, during some value of secure f because every trader has some sensi-
the second series, you will be losing $2,500 after the first five tivity to drawdown.
moves of the contract. That means that at this moment you
have a loss of $7,500 for this series, or a $6,000 total loss of SECURE F
capital. This is 7,500 divided by 101,500 = 0.0739 or 7.39% To formulate the problem solved by secure f, we add a
of the equity at the beginning of the second series. This is the constraint into the calculation of optimal f. The constraint
maximum trade drawdown. It is also 6% of the total capital may reflect the acceptable maximum drawdown (and/or
available ($100,000). other characteristics). This is a more conservative strategy
As we can see from our calculations, optimal f has several that has the benefit of finding the percent of equity invested
drawbacks. First of all, the strategy leads to a rapid expansion in every trade that would have yielded the highest possible
in the number of contracts traded. As a rule, this causes return subject to the acceptable maximum drawdown.
unacceptably large drawdowns. Let us reformulate the problem of finding optimal f by
The second limitation of optimal f is that it is not psycho- adding the drawdown constraint:
logically desirable for the average trader concerned with risk.
It may also be an unrealistic assumption, depending on the TWR(f) —> max
market traded, that the liquidity exists or that the trading subject to
system would still be viable given a geometric increase in the Maximum drawdown(f) is less than or equal to acceptable
number of contracts traded. maximum drawdown set by trader
 Reprinted from Technical Analysis of STOCKS & COMMODITIES magazine. © 1998 Technical Analysis Inc., (800) 832-4642, http://www.traders.com

The solution that maximizes TWR(f) is COMPARISON TABLE STRATEGY COMPARISON TABLE
the secure f. The formulation of the Characteristics Optimal f Secure f
f Contracts TWR MaxDD
problem that includes drawdown can
be seen in Figure 2. 0 0 100 0 Max. # of contracts
5 0 104.74 0 traded 75 4
This formulation of the problem is 10 1 108.9 2500 Net profit
such that its solution will maximize % increase 310% 56%
15 1 112.41 2500
TWR and guarantee that the draw- 20 2 115.2 5000 Max. drawdown
down when running the system on 25 2 117.19 5000 % increase 3962.50% 112.50%
past data does not exceed the amount Net profit/maximum
30 3 118.3 7500 Drawdown ratio 1.14 :1 8.27 : 1
defined by the trader — value D. This 33 3 118.52 7500
35 3 118.46 7500 Average trade (in $) 3,008.15 1,146.74
value is referred to as acceptable Standard deviation of
maximum drawdown. (This could be 40 4 117.6 10000 Average trade (in $) 48,976 5,296
45 4 115.64 10000
a trade or equity drawdown or other 50 5 112.5 12500 RINA index 13.92 34.32
risk measure, or several risk mea- Maximum drawdown
55 5 108.11 12500 (in %) 65.8 24.4
sures at the same time. There could 60 6 102.4 15000
be a constraint that includes a mea- 65 6 95.29 15000 FIGURE 5: The simple breakout system that trades the mark
sure of volatility. Other constraints 70 7 86.7 17500 buys on a 20-day price breakout and exits the position on a
75 7 76.56 17500 10-day price reversal breakout. Using optimal f, the maxi-
can be taken into account in the same mum number of contracts is 75 and the maximum drawdown
80 8 64.8 20000
way as shown in Figure 2 just by is 65.8%. Using secure f and setting the maximum drawdown
85 8 51.34 20000 to $7,000, the maximum number of contracts is 4.
adding more constraints.) Figure 2 90 9 36.1 22500
has the prices as an additional input 95 9 19.01 22500
necessary for calculations. FIGURE 4: Increasing f values for percent-
If the acceptable maximum draw- age of capital add aditional contracts, but
down is smaller than the maximum the terminal wealth relative is maximum at
drawdown during the period being con- 118.52.
sidered, the secure f value will be
smaller than the optimal f value and the secure f strategy will ALGORITHM
yield more conservative returns that will satisfy the limita- Let f s be the value of secure f, D the acceptable maximum
tions on the drawdown. drawdown, TWR (f, D) the terminal wealth relative value for
The equity curve for secure f will also have less variability. given values of f and D; delta is the desired accuracy for the
If the acceptable maximum drawdown is equal to or greater secure f calculation. Below is a simplified form of the
than the maximum drawdown during the period being con- algorithm for the calculation of secure f.
sidered, the secure f value will be equal to the optimal f value Step 1 Calculate optimal f
and the secure f strategy will yield the same results as the Step 2 Set f = optimal f
optimal f strategy. Figure 3 shows a graph of the TWR and
maximum drawdown versus f value, and Figure 4 is a com- If the acceptable maximum drawdown is greater than or equal
parison table for optimal f and secure f. to the historical maximum drawdown, then stop; otherwise,
go to step 3.
MONEY MANAGER (RINA SYSTEMS)

MONEY MANAGER (RINA SYSTEMS)

FIGURE 6: PERFORMANCE STATISTICS. The DM breakout system performance FIGURE 7: PERFORMANCE STATISTICS. The DM breakout system performance
statistics using optimal f can be found here. The first column is a standard test statistics using secure f are in this table. The column descriptions are the same as
trading one contract. The second column applies optimal f over the last five trades. Figure 6.
The third column presents the difference between the two strategies.
Reprinted from Technical Analysis of STOCKS & COMMODITIES magazine. © 1998 Technical Analysis Inc., (800) 832-4642, http://www.traders.com

FIGURE 8: UNDERWATER EQUITY CURVE. Using optimal f, this chart shows the FIGURE 9: UNDERWATER EQUITY CURVE. Using secure f, this chart shows the
percent of drawdown as of the end of each month as measured from the previous percent of drawdown as of the end of each month as measured from the previous
equity peak and reflects the maximum possible equity retracement in percent at equity peak and reflects the maximum possible equity retracement in percent at
each point. each point.

Step 3 f = f - delta
APPLICATION
If f > 0, then
AND COMPARISON
• Calculate MDD(f) - the maximum drawdown for the f
For comparison purposes, con-
portion of the capital to be invested in every trade using the
sider a simple breakout system
information about trades and prices. If MDD(f) < D, then
that trades the German mark.
calculate TWR(f, D) and store the values of f and TWR(f, D).
The system buys on a 20-day
• Return to the beginning of step 3. price breakout and exits the po-
sition on a 10-day price reversal
If f < or = 0, then breakout. The TradeStation code
• Find the value of f such that corresponds to the maximum of for this breakout system is listed below:
all TWR(f, D) stored. This value is the secure f, namely f s.
Breakout system
• Stop.
Input: BLen(20), XLen(10);
IF CurrentBar > 1 and Close > Highest(High,BLen)[1] Then Buy on
The same logic could be applied to find the secure f for Close;
constraints on other than maximum drawdown. IF CurrentBar > 1 and Close < Lowest(Low,XLen)[1] Then ExitLong;

FIGURE 10: EQUITY CURVE. Here is the equity curve using optimal f. The equity FIGURE 11: EQUITY CURVES. Here are the equity curves by bar for secure f.
growth, which is very erratic, peaks in 1998. Note the steady rise using the DM breakout system.
 Reprinted from Technical Analysis of STOCKS & COMMODITIES magazine. © 1998 Technical Analysis Inc., (800) 832-4642, http://www.traders.com

Both optimal f and secure f strategies increased the net profit Leo Zamansky, Ph.D., is president of RINA Systems in Cincin-
but at a cost to the maximum drawdown. The increase of nati, OH. The company specializes in software development
maximum drawdown relative to net profit clearly indicates for the serious trader. David Stendahl is vice president of
the dramatic differences between these two strategies. Al- financial services with RINA Systems and a professional
though the optimal f strategy generated a much larger net trader.
profit, the net profit/maximum drawdown ratio for secure f is RINA Systems is the developer of Money Manager, Perfor-
more than seven times higher than the same ratio for optimal mance Summary Plus, Portfolio Evaluator, 3D SmartView
f. The strategy comparison table in Figure 5 outlines the and Dynamic Zones software for evaluating and improving
differences between the strategies. trading systems. RINA Systems is also a codeveloper of
During the 12-year test period (1985-97), the optimal f Portfolio Maximizer.
strategy would have attempted to trade 75 German mark Zamansky and Stendahl can be reached at RINA Systems,
contracts in a single trade. The secure f strategy, on the other 7854 Weavers Lane, Maineville, OH 45039, phone 513 772-
hand, would have traded a maximum of four German mark 7462, or via Web site at http://www.rinasystems.com. Secure
contracts at a single time. f Calculator is available on the Web site under “Visual tours
The breakout system performance using optimal f and and downloads.”
secure f can be seen in Figures 6 and 7. Figures 8 and 9
illustrate the corresponding underwater equity curves for RELATED READING
both optimal f and secure f. Figure 9 shows the percent of Vince, Ralph [1990]. Portfolio Management Formulas, John
drawdown as of the end of each month as measured from the Wiley & Sons.
previous equity peak and reflects the maximum possible Zamansky, Leo J., and David Stendahl [1997]. “Dynamic
equity retracement in percent at each point. These figures zones,” Technical Analysis of STOCKS & COMMODITIES,
show the big difference in the equity drawdowns for optimal Volume 15: July.
f and secure f. Finally, Figures 10 and 11 depict the equity _____ [1997]. “Evaluating system efficiency,” Technical
curves by bar for optimal f and secure f correspondingly. Analysis of STOCKS & COMMODITIES, Volume 15: Octo-
ber. S&C
CONCLUSIONS
Secure f money management strategy offers traders the
ability to maximize the return subject to the level of risk they
are willing to assume. Secure f can be a conservative or
aggressive strategy, depending on the level of acceptable
maximum drawdown selected by the trader. This level can be
set by each individual trader to match the trader’s prefer-
ences. The secure f strategy can be applied to both mechanical
and nonmechanical trading and to increase profitability in
any market.

RINA Systems holds intensive three-day and Extreme MACD. Traders learn how to
seminars entitled “Effective Methods for systematically evaluate trading performance
Evaluating and Improving Trading Perfor- using RINA Systems’ Portfolio Maximizer,
mance” that cover the six stages of creating Money Manager, 3D SmartView and our
and trading profitable systems. The stages proprietary Price Simulator. The seminar
are: Design, Development, Evaluation, Im- also covers the use of statistical measures,
provement, Application and Monitoring. A charts, money and risk management tech-
variety of trading systems that trade several niques, to build robust trading systems and
markets and instruments, including S&P fu- to evaluate and improve trading performance.
tures, are discussed and their code is dis- Seminar participants receive a seminar work-
closed. The seminar covers several money book, trading rules for the systems ana-
and risk management strategies applied to lyzed, a free copy of the Dynamic Zones
both mechanical and discretionary trading. Indicator with a reference guide, an evalua-
Specific systems in consideration include: tion of the participants’ trading systems, and
CCI Spike, Variable Detrend, DZ %R, VIX copies of articles written by RINA Systems.