Ultimate Algorithmic Trading System

Recommendations

Info

28STOCHASTICS AND AVERAGES AND RSI! OH, MY!The ADX index incorporates moving averages of true range and both +DMand −DM. Data averaging is also the same as data smoothing. A moving averagefunction cuts down the impact of abnormal data points and creates a more robustmodel of the data. Wilder created his indicators without the use of a computerand created a shortcut to average his data points. This Wilder’s smoothing is morelike an exponential calculation. Different trading/testing/charting platforms usedifferent smoothing functions for Wilder’s indicators, so be sure to check youruser’s manual and see which form is being utilized by your particular software.Assuming a 14-day ADX length, Wilder would sum up the first 14-day values andthen divide by 14. Once these initial averages were calculated the subsequent barvalues were calculated by first dividing the previous sum by 14 and subtracting itfrom the previous sum and then adding today’s values:previous plusDM14plusDM14 = previous plusDM14 − + plusDM today14This form of smoothing eliminates keeping track of the prior 14 bar values andaccomplishes the desired effect (reducing the impact of outlying data points) onplus DM, minus DM, and true range. Remember, Wilder probably had the use ofjust a calculator, and instead of inputting 42 numbers (summing three sets of 14numbers), he simply used the above formula and whipped out the values in a minuteor less. Most of today’s charting software utilizes an exponential moving average tosmooth the ADX. This smoothing method creates results that can be quite differentthan Wilder’s original method.Wilders:Exponential:ADX =((priorADX ∗ 13)+DX)∕14ADX = priorADX + 2∕15 ∗ (DX − priorADX)The smoothing factor (2/(N+1)) or in this case (2/15) can be changed to (1/N)to arrive at Wilder’s calculation. This becomes more like a 27 period exponentialmoving average than a 14. However, this exponential method is considered accurate,but, in some cases, different charting software will offer both versions: ADX andWilder ADX.Calculation for a 14-day ADX:1. Calculate the true range (TR), plus directional movement (+DM), and minusdirectional movement (−DM) for each bar.2. True range (TR ) = max(C[1],H) − min(C[1],L). In other words: Take thehigher of yesterday’s close and today’s high and subtract the lower of yesterday’sclose and today’s low. Basically, you are expanding today’s range to incorporatethe prior day’s close if it is outside of today’s range.3. Smooth these periodic values using a 14-day moving average. You will eitherhave to wait until 14 days have transpired or you can simply look at 14 days ofprior history before commencing the ADX calculation.www.rasabourse.com
4. Divide the 14-day smoothed plus directional movement (+DM) by the 14-daysmoothed true range to find the 14-day plus directional indicator (+DI14).Multiply by 100 to move the decimal point two places. This +DI14 is the plusdirectional indicator that is plotted along with ADX.5. Divide the 14-day smoothed minus directional movement (−DM) by the 14-daysmoothed true range to find the 14-day minus directional indicator (−DI14.Multiply by 100 to move the decimal point two places. This −DI14 is the minusdirectional indicator that is plotted along with ADX.6. The Directional Movement Index (DX) equals the absolute value of +DI14 less−DI14 divided by the sum of +DI14 and −DI14.7. After all these steps, it is time to calculate the Average Directional Index (ADX).The ADX value is initially simply a 14-day average of DX.8. Wilder went one step further and created the ADXR, which is simply the averageof an N-day differential of the ADX. Continuing with our 14-day example, theADXR would be calculated by applying the following formula:ADXR = ADX[1]+ADX[14]14Figure 2.2 illustrates how the ADX can be used to help determine the trendinessof a market. A rising ADX, as shown in the early part of the price chart, demonstrates29STOCHASTICS AND AVERAGES AND RSI! OH, MY!FIGURE 2.2 ADX as a trend detector.Source: TradeStationwww.rasabourse.com
Page 1 and 2: THE ULTIMATEALGORITHMICTRADING SYST
Page 3 and 4: THE ULTIMATEALGORITHMICTRADING SYST
Page 5 and 6: CONTENTSAbout the authorIntroductio
Page 7 and 8: Reproduction 238Mutation 240Using G
Page 9 and 10: ABOUT THE AUTHORIt was March of 198
Page 11 and 12: Well that was 27 years ago and I di
Page 13 and 14: xivINTRODUCTION TO THE ULTIMATE ALG
Page 15 and 16: xviINTRODUCTION TO THE ULTIMATE ALG
Page 17 and 18: CHAPTER 1Introductionto TradingAlgo
Page 19 and 20: However, at this introductory stage
Page 21 and 22: completely above the tableau. If th
Page 23 and 24: Don’t get bogged down trying to u
Page 25 and 26: checking account was down the $10K,
Page 27 and 28: Short Liquidation Condition—What
Page 29 and 30: One BarStrengthTwo BarStrengthThree
Page 31 and 32: and inexperienced trading system pr
Page 33 and 34: Input = 6StartInput = 4Input <> 6In
Page 35 and 36: The diagram in Figure 1.4 looks cle
Page 37 and 38: observed. If the latter criterion i
Page 39 and 40: Start or NewBarIs C >AVG(C,200)YesN
Page 41 and 42: CHAPTER 2Stochastics andAverages an
Page 43: The basic theory behind the ADX is
Page 47 and 48: if myADX < 35 thenif MarketPosition
Page 49 and 50: TABLE 2.1Performance of Directional
Page 51 and 52: FIGURE 2.4Source: TradeStationExamp
Page 53 and 54: Start orNew BarIs RSI < 30YesBuy Ma
Page 55 and 56: RSI DivergenceMarket Continuing to
Page 57 and 58: If marketPosition =-1 and close > e
Page 59 and 60: TABLE 2.3Performance of RSI Failure
Page 61 and 62: Start orNew BarIs Slow %K CrossingA
Page 63 and 64: TABLE 2.4Performance of Stochastic
Page 65 and 66: it is boundless extended readings a
Page 67 and 68: TABLE 2.5Performance of CCI algorit
Page 69 and 70: FIGURE 2.11Source: TradeStationThe
Page 71 and 72: We will discuss moving averages in
Page 73 and 74: MACDAvg = XAverage( MyMACD,9)MACDDi
Page 75 and 76: FIGURE 2.13Source: TradeStationThre
Page 77 and 78: LC0_E0B −25000 −41790 398 −62
Page 79 and 80: FIGURE 2.14Source: TradeStationA 60
Page 81 and 82: TABLE 2.8 Performance of Bollinger
Page 83 and 84: If filterTrade = false thenIf close
Page 85 and 86: Keltner ChannelChester Keltner brou
Page 87 and 88: FIGURE 2.15Source: TradeStationKelt
Page 89 and 90: 35000300002500020000150001000050000
Page 91 and 92: TABLE 2.12Best Parameter Sets of Ke
Page 93 and 94: CHAPTER 3Complete TradingAlgorithms
Page 95 and 96:
Box 3.1 Turtle System #1Buy one tic
Page 97 and 98:
TABLE 3.1Turtle System 1 without LT
Page 99 and 100:
TABLE 3.3Turtle System 1 with LTLF
Page 101 and 102:
TABLE 3.4Tandem Performance of Turt
Page 103 and 104:
Pause the Battle—Let’s Talk Ris
Page 105 and 106:
Box 3.3 Battle Royale—And the Com
Page 107 and 108:
TABLE 3.5Single Moving Average (SMA
Page 109 and 110:
TABLE 3.8Donchian Channel Breakout
Page 111 and 112:
we have, because who has the time t
Page 113 and 114:
%Winners ShortLen InterLen StopDoll
Page 115 and 116:
on a peak. This demonstrates a lack
Page 117 and 118:
The correlation coefficients range
Page 119 and 120:
TABLE 3.13The Best Results for the
Page 121 and 122:
TABLE 3.17Results for the Donchian
Page 123 and 124:
FIGURE 3.5AmiBroker’s walk-forwar
Page 125 and 126:
direction of the 200-day moving ave
Page 127 and 128:
range day and initially puts on two
Page 129 and 130:
CHAPTER 4Introduction toAmiBroker
Page 131 and 132:
Things All Programmers Need to Know
Page 133 and 134:
expressions deal with a string of c
Page 135 and 136:
AFL uses functions DateNum(), DateT
Page 137 and 138:
to Short and Cover as well. If you
Page 139 and 140:
FIGURE 4.2The AmiBroker Editor wind
Page 141 and 142:
to the user. Instead of having to r
Page 143 and 144:
This is where you can change the va
Page 145 and 146:
shortMav = MA(C,avgLen1);longMav =
Page 147 and 148:
code. This is a very difficult thin
Page 149 and 150:
the order of the arrays. I admit th
Page 151 and 152:
FIGURE 4.17The AFL Code Wizard ‘
Page 153 and 154:
■■■This time click on is grea
Page 155 and 156:
This should look very similar to th
Page 157 and 158:
before a single line of code is typ
Page 159 and 160:
COND3 = C > MA(C,100);COND1 = cente
Page 161 and 162:
’********************************
Page 163 and 164:
FIGURE 5.1trading.Once you define t
Page 165 and 166:
GetDataGET DATACOMPONENTLoad Data i
Page 167 and 168:
152USING MICROSOFT EXCEL TO BACKTES
Page 169 and 170:
performance of the trading system.
Page 171 and 172:
156 FIGURE 5.9 The Visual Basic Edi
Page 173 and 174:
FIGURE 5.12This error message appea
Page 175 and 176:
End IfIf marketPosition = -1 ThenCa
Page 177 and 178:
■■moc—Market On Close order.
Page 179 and 180:
Simple Moving Average Crossover usi
Page 181 and 182:
Call Trade(ExitShort, "S-ATR Prof",
Page 183 and 184:
168USING PYTHON TO BACKTEST YOUR AL
Page 185 and 186:
It works well on both Unix and Wind
Page 187 and 188:
function modules and some are a lit
Page 189 and 190:
inclusive, the number of iterations
Page 191 and 192:
176This indicator is programmed as
Page 193 and 194:
if prices[curBar - offset] < prices
Page 195 and 196:
All these files deal with opening a
Page 197 and 198:
entryPrice,fileList,entryPrice,entr
Page 199 and 200:
Python has a ton of reusable code a
Page 201 and 202:
186USING PYTHON TO BACKTEST YOUR AL
Page 203 and 204:
portfolioClass. These classes are n
Page 205 and 206:
percent wins, number of trades, and
Page 207 and 208:
>>> c = a + b>>> cAfter you hit ent
Page 209 and 210:
you don’t need to know exactly ho
Page 211 and 212:
Shorting occurs when the low of the
Page 213 and 214:
That was the long entry logic only
Page 215 and 216:
FIGURE 6.4brackets.A Python run-tim
Page 217 and 218:
prevAvg2 and today’s avg1 must be
Page 219 and 220:
■ TradeStation IDE204AN INTRODUCT
Page 221 and 222:
206AN INTRODUCTION TO EASYLANGUAGEt
Page 223 and 224:
FIGURE 7.5The algorithm with a sing
Page 225 and 226:
the Vars statement and initialize t
Page 227 and 228:
TABLE 7.1EasyLanguage Keywords and
Page 229 and 230:
214AN INTRODUCTION TO EASYLANGUAGEF
Page 231 and 232:
standard deviations. In our example
Page 233 and 234:
218AN INTRODUCTION TO EASYLANGUAGEt
Page 235 and 236:
into a project makes it easier to a
Page 237 and 238:
If marketPosition =-1 thenbeginend;
Page 239 and 240:
or program any trading idea. Practi
Page 241 and 242:
CHAPTER 8Genetic Optimization,Walk
Page 243 and 244:
superfast processors, searching a l
Page 245 and 246:
This equation involves four unknown
Page 247 and 248:
For generation = 1 To 100 ' total g
Page 249 and 250:
Converting Fitness in Terms of Slic
Page 251 and 252:
'the 6 darts landedRandomizeFor l =
Page 253 and 254:
roulette(i) = Rnd(1#)If roulette(i)
Page 255 and 256:
selection and reproduction may even
Page 257 and 258:
This is just one possible solution.
Page 259 and 260:
history of each market. This test w
Page 261 and 262:
Crossover RateThe rate that determi
Page 263 and 264:
TradeStation’s and AmiBroker’s
Page 265 and 266:
TABLE 8.3In-Sample Testing TA2Le Se
Page 267 and 268:
Here are the various calculations t
Page 269 and 270:
1. movAvgLen { 50 to 150 by 10 incr
Page 271 and 272:
TABLE 8.6The Out-of-Sample (OOS) Re
Page 273 and 274:
your algorithm are each written on
Page 275 and 276:
That’s it! These seven lines of c
Page 277 and 278:
[Equity]350003000025000200001500010
Page 279 and 280:
Calculating Start Trade Drawdown Wi
Page 281 and 282:
A loop is used to flow through each
Page 283 and 284:
The maximum drawdown of this algori
Page 285 and 286:
■ Fixed Fractional272PORTFOLIO AN
Page 287 and 288:
FIGURE 9.2Add a strategy to your st
Page 289 and 290:
276FIGURE 9.6The UGTATS markets and
Page 291 and 292:
278FIGURE 9.9 How to adjust the set
Page 293 and 294:
FIGURE 9.12 The Trade Analysis wind
Page 295 and 296:
282PORTFOLIO ANALYSISFIGURE 9.15thi
Page 297 and 298:
FIGURE 9.18 Profit and drawdown are
Page 299 and 300:
FIGURE 9.20 To set the portfolio st
Page 301 and 302:
FIGURE 9.24Source: TradeStationThe
Page 303 and 304:
count makes sense; trades are turne
Page 305 and 306:
APPENDIX AAmiBrokerHere is a brief
Page 307 and 308:
switchbreak (part of the switch sta
Page 309 and 310:
Here is the source code from Chapte
Page 311 and 312:
PctSize = PositionRisk * MarginDepo
Page 313 and 314:
■ KeywordsprevMarketPositionmarke
Page 315 and 316:
upSumAvg = upSum / lengthdnSumAvg =
Page 317 and 318:
Sub BollingerBand(dataList, length,
Page 319 and 320:
308EXCEL SYSTEM BACKTESTERIf (kVal
Page 321 and 322:
■ Keywords and Identifierscurrent
Page 323 and 324:
312PYTHON SYSTEM BACKTESTERdef calc
Page 325 and 326:
sysMark =self.systemMarkets[i]avgLo
Page 327 and 328:
class stochClass(object):class stoc
Page 329 and 330:
318PYTHON SYSTEM BACKTESTERfor i in
Page 331 and 332:
■ Python-Specific Keywordsandasas
Page 333 and 334:
Monte Carlo Analysis# Monte Carlo A
Page 335 and 336:
FIGURE D.1324TRADESTATION AND EASYL
Page 337 and 338:
Simple RSI with Profit Objective an
Page 339 and 340:
If condition1 and condition2 then b
Page 341 and 342:
330TRADESTATION AND EASYLANGUAGEIf(
Page 343 and 344:
if lastTradeLoser thenbeginif curre
Page 345 and 346:
Value1 = swingHighBar(1,h,pivotHiSt
Page 347 and 348:
■■■■■■■Statistically
Page 349 and 350:
INDEXNote: Page references followed
Page 351 and 352:
calculation2 standard deviations, u
Page 353 and 354:
ES. See Exhaustive searchESB. See E
Page 355 and 356:
Mini Russell day trading system, na
Page 357 and 358:
PPO. See Percentage price oscillato
Page 359 and 360:
Time-based market order (MOC), exit
show all

Ultimate Algorithmic Trading System

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?