back start next

[start] [1] [2] [3] [ 4 ] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54]

4

The other way to derive the average trade amount is to start with the percentage retum per trade or a geometric mean [35, 36]. In Table 1.3, lo compute the Average \$ Per Trade, multiply the Equity by the Allocation per position, then multiply by the "o Return Per Trade, and finally subtract the Trade Cost (commissions and shppage). The Net Income is the Monthly Gross minus Taxes minus the Portfolio Cost.

Tablel.3. Expect edMonthly Income

 Equity 100000 Portfolio Cost 1000 Trade Cost 200 # Trades 20 Allocation 500 Tax Rate 0.3 "oRetiuTi Per Trcde Averte S Per Trade Mmtlily Ooss Mt liicoiin Mollify "aRe/tmi 0.50"o 1000 -300 -0.30<*o 0.7»o 3500 1050 1450 1.4»o 1.0¹o 6000 1800 3200 3.20"o 1.2»o 8500 2550 4950 4.95°o 1.50>o 11000 3300 6700 670"o IJS-o t3500 4050 8t50 8.45°o 2.00°o t6000 4800 t0200 lQ2¹o

To receive real-time quotes, a trader must complete exchange agreements and pay monthly fees for the data feed. Standard trading tools are tjpically bundled by a direct access broker so that the trader pays one monthly fee for a certain level of ser\ice. In many cases, the monthly fee will be waived or rebated based on the mmiber of trades; the credit is usually applied the first week ofthe following month to your account.

I1r- - - mr.iii IS m.illiciit:il((.il ri-iiii loi (li-s< lilniif llirKi*<wl< lodc-"riii- i-M li.mfm-.(li%lii),-.iii.li lii-w((-ii />)"/< iii"h.f/.iiiil iiii, fl of, unit,.!/ lt.lili-i% < h-ik-mIIv, ilvmi .„. ........I..I .v.il, ll............ i„.l., Im ... I.....>. -Ill........I l>.l,.,ll ..I .1 .ii.m....

illCIl \<...........yi- ll I tVllll illi I yi ll.llil . .1 . llllli ..... Mlll.l lliin .Mill \tn .......

[fyour trade volimie is vary high, then negotiate with the broker for a lower commission rate. Commissions should be no greater than one cent per share, or \$10 per 1000 shares. Other fixed costs are:

a Technical analj-sis software,

a Real-tune news sources such Bloombeig or Dow Jones, and a Subscriptions to advisory sen-ices and other publications.

Depending upon the requirements of the trading sjstems, monthly costs will \aij from as Uttle as se-eral hundred doUais to se-eral thousand doUais. Paj-ing more for ad\anced trading tools such as stock screeners (e.g., FirstAlert) and ser\ices (e.g., a Bloombeig terminal) maybe worth the additional cost. Software costs can be expensive and have a significant impact on the bottom line for smaller accounts (review Table 1.3).

Think of margin as a length of rope, and recall the wall-known idiom about hanging. The tjpical investor with a brokerage account gets 2:1 margin, and the pattem day trader gets 4:1 intraday margin. The question is whether or not a trader with a great sjstem should use margin. First, frame the question in terms ofriskas apercentage of equity, i.e., how much one is willing to lose on a single trade. Suppose the trader has a \$100,000 account and is willing to lose no more than 2% ofequity on any single position. The maximum loss per trade is \$Z000

Now, siqipose the trader wants to le-ere the position on 2:1 margin. The position size is doubled but the percent risk is stiU 2" o. Ifthe trader has designed a stop loss based on this riak value, then positions wiU be stopped out more often because the maximum loss per trade has not been adjusted to refiect the doubled size ofthe position. To maintain the efficacy ofthe sjstem, the trader would have to increase the percent risk to 4" o, thereby increasing the maximum loss per trade to \$4,000. This change affects the integritj oftiie portfolio, as its past and future performance may not be able to bear 4" risk on e-erj trade.

Returning to the great system, suppose the maximum loss of our sjstem has been 1.5% ofequity for a series ofsewralhimdred trades, and the percent risk is initially set to 2°o. Given that the maximum loss has been only 1.5°o ofequity (but with no assurance as to future performance), the trader may decide to use margin in our theoretical account of \$100,000. The formulais:

Margin = Equity X (Risk "a / Maxinmm Loss "a) (1.1)

In our example, the traders margin would be \$100,000 X (2 / 1.5) = \$133,333. The expected highest loss would be \$ 133,333 X l.t, = \$2,000, or 2<*o ofequity.

Before using margin, however, be skeptical ofthe highest percentage loss numberandthinkof seen arioswh ere thatnumbercouldbeexceeded[30].Fur-

tJier, do not use margm on a sjstem with lunited historical data or a short back testing period (e.g., a relatively new issue or instrument). Finally, examine the maximum consecutive losers to determine whether or not the sjstem has an exceptional losing string.

Position Sizing

Position size for aH ofthe Acme Trading Sjstems is calculated from the models described inThaips book Trarfe Your Way toFinojicialFreedom [34]. The sizing models are as foUows:

a Equal Value Units Model

a Percent Risk Model

a Percent Volatihty Model

The Equal Value Units Model is simple. AJlocaie a fixed percentage ofequiiyto each position in the portfoUo. For example, if account equity of \$100,000 is to be spread equaUy among 4 positions, then \$25,000 is aUocated to each position, regardless ofprice. If Stock A is trading at a price of 10, then Stock As position size is 25000 / 10 = 2500 shares. If StockB is trading at aprice of25, then the position size of Stock is 25000 / 25 = 1000 shares. The problem wiOi this model is that it does not considervolatiUtjinthe equation, so Stock A may have amuchgreater impact on the portfolio than StockB, orvice versa.

The Percent RiskModel is based on the maximum number ofunits (e.g., points for stocks) one is willing to lose on any single trade. The fonnulais:

Position Size =Eqiuty XRlsk RiskUnits (1.2)

For example, if Equity is \$100,000 and the Risk Percentage is 2°o, then the trader may decide that a two-point stop loss is appropriate. The position size in this case is 100,000 X .02 / 2 = lOOO shares. As a practical consideration, the trader must select an appropriate stop loss per stock and not apply the same alue umversaUy to aportfoUo ofstocks. The weakness ofthis model is that it is unit-based and not percentage-based. Instead, the stop loss value should be derived from a standard percentage loss such as 4" o. StUl, even the use of a fixed percentage is not optimal.

The Percent VolatiUtj Model is the default model for the Acme sjstems. It is the only model to standardize across volatility The difference between this model and the Percent RiskModel is the calculation ofttie Average True Range (ATR) denominator. This model adjusts to the inherent volatility of each stock because it uses the ATR, in contrast to the Percent Risk Model where the trader selects risk. The formula is:

ie = Eq ry X Fjs

(1.3)

For example, siqipose atrading account has \$100,000, and the traderwishes to lose no more than 2% on any one trade. Ifthe stocks ATR is two points, then the number of shares is 100,000 X .02 = 2000 / 2 = 1000 shares. Ifthe ATR is fourpoints, then the number ofshares is 2000 / 4. or 500 shares. As volatility increases, the number ofshares decreases.

Example 1.1. Funcuon AcmeGetShares

AcmeGetShares: Calculate the number of shares based on risk model

RiskModel = 1, Equal Value Units Model RiskModel = 2, Percent Risk Model RiskModel = 3, Percent Volatility Model

Inputs:

EqiJity(Nunieric), RiskModel(Nuineric). RiskPercent(Nunieiic), RiskUnits(NumeTic);

Variables:

HiniinuinShares(200), Risk5tiares(0), ERP(O.O), Lengtli(20);

ERP = Equity * RiskPercent / 100;

If RiskModel = l and Close > 0 Then RiskShares = HaxList(HininiuniShares, 100 * IntPortion(Equity / (100 * Close)));

If RiskModel = 2 and RiskUnlts > 0 Then RiskShares = HaxList(HininiuniShares, 100 * IntPortion(ERP / (100 * RiskUnits)));

If RiskModel = 3 and Volatility(lerth) > 0 Then RiskShares = Max List (HinimurnShares, 100 » IntPortlon(ERP / (100 * Volatility(Length))));

AcmeGetShares = RiskShares;

Eadi of the position sizing models is encoded in a common function that can be called by all ofthe trading systems in the portfolio. The .-icmeGeiSfus-es function shovv-n in Example 1.1 is written in EasyLanage; it calculates fheposition size

,. i:J.,nl,/fi...............I,. ....I. \x.,.,y

based on the equity and the selected position-sizmg model. The number of shares is calculated andretumed to the trading sjstem calling the function.

By standardizing the number of shares traded across all equities, risk is spread evenly across the entire portfolio. Thus, the AcmeGetShares function is called by evary trading sjstem in the portfolio. An EasjLanguage example of calling the function is shown below in Example 12:

Esample 1.2. Cdiing the.4fwGeiS/ji!Wi Function

Inputs:

Equity(lOOOOO),

RiskModel(3), {Percent Volatility Model}

RiskPercent(2.0),

RiskATR(l.O); {Applies only to Percent Risk Model}

(Calculate shares based on risk model}

N = AcnieCet5tiares(Equity, RiskModel, RiskPercent, RiskATR); Buy N Shares;

The Trade Manager ishke an octopus; it is the brain of a trading operation with its arms in every trade. The trader must decide whether or not to use stop or limit orders, to use profit targets or not, how to implement stop losses, and how long to hold aposition

MLAST-Daily 03«1/2002 Mov 1 llnrlClose,50,0) 49.266

The Trade Manager halps the trader with visual cues, showmg the action pornts ofthe trade. Knowmg the profit targets and stop losses a priori gi-es a trader confidence and reinforces discipline when exiting a trade. Figure 1.2 shows an example ofthese visual cues.

Improper settings can tum a winning trading sjstem into a losing one; the strength of a trading sjstem depends not only on its design but also onabalance between the maximimi profit potential of a position and the holding period. For example, ifthe distribution of trades shows that the a-erage return of a winning trade is 2** in two trading dajs and 2.5% rn three trading days, then the trader should be taking profits after two days [32].

The trader should go through the exercise of experimenting with and without profit targets, testing different holding periods, and adjusting entrj and exit parameters. For example, the trader may decide not to enter long positions one tickabo\e the previous ds highbut instead wait for ahttle more confirmation based on a percentage ofthe average true range, e.g., 25 ofthe ATR

Naming Convention

Each of the Acme Trading Systems has a designated letter (SystemID) that is part ofthe naming convention for trading signals. Each signal name contains a two-letter identifier containing the order tjpe. Long (L ) or Short (S) combined with either Entrj (E) or Exit (X). Entries have a SystemID, and exit signals have an identifier appended to the order tj-pe speciijing either a profit target or a stop loss. An example of an entrj is kcme SE M, an Acme M short signal. An example of an exit is H-M, a multi-day profit target for a long entrj. Refer to Table 1.4 for the Ust of qualifiers used in a signal name.

Table 1.4. Signal QuaHfiers

 Symbol Description Long Entry Long Exit Short Entry Short Exit Daily Profit Target M..lti-l).iy Profit TarKt . 1,.. 1

[start] [1] [2] [3] [ 4 ] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54]