-
Notifications
You must be signed in to change notification settings - Fork 382
/
bt.r
2195 lines (1808 loc) · 62 KB
/
bt.r
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
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
###############################################################################
# This software is provided 'as-is', without any express or implied
# warranty. In no event will the authors be held liable for any damages
# arising from the use of this software.
#
# Permission is granted to anyone to use this software for any purpose,
# including commercial applications, and to alter it and redistribute it
# freely, subject to the following restrictions:
#
# 1. The origin of this software must not be misrepresented; you must not
# claim that you wrote the original software. If you use this software
# in a product, an acknowledgment in the product documentation would be
# appreciated but is not required.
# 2. Altered source versions must be plainly marked as such, and must not be
# misrepresented as being the original software.
# 3. This notice may not be removed or altered from any source distribution.
###############################################################################
# Backtest Functions
#
# For more information please email at TheSystematicInvestor at gmail
###############################################################################
###############################################################################
# Align dates, faster version of merge function
#' @export
###############################################################################
bt.merge <- function
(
b, # environment with symbols time series
align = c('keep.all', 'remove.na'), # alignment type
dates = NULL # subset of dates
)
{
align = align[1]
symbolnames = b$symbolnames
nsymbols = len(symbolnames)
# count all series
ncount = sapply(symbolnames, function(i) nrow(b[[i]]))
all.dates = double(sum(ncount))
# put all dates into one large vector
itemp = 1
for( i in 1:nsymbols ) {
all.dates[itemp : (itemp + ncount[i] -1)] = xts::.index(b[[ symbolnames[i] ]])
itemp = itemp + ncount[i]
}
# find unique
temp = sort(all.dates)
unique.dates = c(temp[1], temp[-1][diff(temp)!=0])
# trim if date is supplied
if(!is.null(dates)) {
class(unique.dates) = c('POSIXct', 'POSIXt')
temp = make.xts(integer(len(unique.dates)), unique.dates)
unique.dates = xts::.index(temp[dates])
}
# date map
date.map = matrix(NA, nr = len(unique.dates), nsymbols)
itemp = 1
for( i in 1:nsymbols ) {
index = match(all.dates[itemp : (itemp + ncount[i] -1)], unique.dates)
sub.index = which(!is.na(index))
date.map[ index[sub.index], i] = sub.index
itemp = itemp + ncount[i]
}
# trim logic
index = c()
if( align == 'remove.na' ) {
index = which(count(date.map, side=1) < nsymbols )
}
# keep all
# else {
# index = which(count(date.map, side=1) < max(1, 0.1 * nsymbols) )
# }
if(len(index) > 0) {
date.map = date.map[-index,, drop = FALSE]
unique.dates = unique.dates[-index]
}
class(unique.dates) = c('POSIXct', 'POSIXt')
return( list(all.dates = unique.dates, date.map = date.map))
}
###############################################################################
# Prepare backtest data environment
#
# it usually contains:
# * b$symbolnames
# * b$universe
# * b$prices
# * b - asset hist data
#
#' @export
###############################################################################
bt.prep <- function
(
b, # environment with symbols time series
align = c('keep.all', 'remove.na'), # alignment type
dates = NULL, # subset of dates
fill.gaps = F, # fill gaps introduced by merging
basic = F # control if xts object are created
)
{
# setup
if( !exists('symbolnames', b, inherits = F) ) b$symbolnames = ls(b)
symbolnames = b$symbolnames
nsymbols = len(symbolnames)
if( nsymbols > 1 ) {
# merge
out = bt.merge(b, align, dates)
for( i in 1:nsymbols ) {
temp = coredata( b[[ symbolnames[i] ]] )[ out$date.map[,i],, drop = FALSE]
b[[ symbolnames[i] ]] = iif(basic, temp, make.xts( temp, out$all.dates))
# fill gaps logic
map.col = find.names('Close,Volume,Open,High,Low,Adjusted', b[[ symbolnames[i] ]])
if(fill.gaps & !is.na(map.col$Close)) {
close = coredata(b[[ symbolnames[i] ]][,map.col$Close])
n = len(close)
last.n = max(which(!is.na(close)))
close = ifna.prev(close)
if(last.n + 5 < n) close[last.n : n] = NA
b[[ symbolnames[i] ]][, map.col$Close] = close
index = !is.na(close)
if(!is.na(map.col$Volume)) {
index1 = is.na(b[[ symbolnames[i] ]][, map.col$Volume]) & index
b[[ symbolnames[i] ]][index1, map.col$Volume] = 0
}
#for(j in colnames(b[[ symbolnames[i] ]])) {
for(field in spl('Open,High,Low')) {
j = map.col[[field]]
if(!is.null(j)) {
index1 = is.na(b[[ symbolnames[i] ]][,j]) & index
b[[ symbolnames[i] ]][index1, j] = close[index1]
}}
j = map.col$Adjusted
if(!is.null(j)) {
b[[ symbolnames[i] ]][index, j] = ifna.prev(b[[ symbolnames[i] ]][index, j])
}
#for(j in setdiff(1:ncol( b[[ symbolnames[i] ]] ), unlist(map.col))) {
# b[[ symbolnames[i] ]][index, j] = ifna.prev(b[[ symbolnames[i] ]][index, j])
#}
}
}
} else {
if(!is.null(dates)) b[[ symbolnames[1] ]] = b[[ symbolnames[1] ]][dates,]
out = list(all.dates = index.xts(b[[ symbolnames[1] ]]) )
if(basic) b[[ symbolnames[1] ]] = coredata( b[[ symbolnames[1] ]] )
}
# dates
b$dates = out$all.dates
# empty matrix
dummy.mat = matrix(double(), len(out$all.dates), nsymbols)
colnames(dummy.mat) = symbolnames
if(!basic) dummy.mat = make.xts(dummy.mat, out$all.dates)
# weight matrix holds signal and weight information
b$weight = dummy.mat
# execution price, if null use Close
b$execution.price = dummy.mat
# populate prices matrix
for( i in 1:nsymbols ) {
if( has.Cl( b[[ symbolnames[i] ]] ) ) {
dummy.mat[,i] = Cl( b[[ symbolnames[i] ]] );
}
}
b$prices = dummy.mat
}
# matrix form
#' @export
bt.prep.matrix <- function
(
b, # environment with symbols time series
align = c('keep.all', 'remove.na'), # alignment type
dates = NULL, # subset of dates
basic = F # control if xts object are created
)
{
align = align[1]
nsymbols = len(b$symbolnames)
# merge
if(!is.null(dates)) {
temp = make.xts(1:len(b$dates), b$dates)
temp = temp[dates]
index = as.vector(temp)
for(i in b$fields) b[[ i ]] = b[[ i ]][index,, drop = FALSE]
b$dates = b$dates[index]
}
if( align == 'remove.na' ) {
index = which(count(b$Cl, side=1) < nsymbols )
} else {
index = which(count(b$Cl,side=1) < max(1,0.1 * nsymbols) )
}
if(len(index) > 0) {
for(i in b$fields) b[[ i ]] = b[[ i ]][-index,, drop = FALSE]
b$dates = b$dates[-index]
}
# empty matrix
dummy.mat = iif(basic, b$Cl, make.xts(b$Cl, b$dates))
# weight matrix holds signal and weight information
b$weight = NA * dummy.mat
b$execution.price = NA * dummy.mat
b$prices = dummy.mat
}
bt.prep.matrix.test <- function() {
#*****************************************************************
# Load historical data
#******************************************************************
load.packages('quantmod')
# example csv file holds returns
# Date ,A,B
# Jan-70,0.01,0.02
returns = read.xts('Example.csv', date.fn=function(x) paste('1',x), format='%d %b-%y')
prices = bt.apply.matrix(1 + returns, cumprod)
data <- new.env()
data$symbolnames = colnames(prices)
data$dates = index(prices)
data$fields = 'Cl'
data$Cl = prices
bt.prep.matrix(data)
#*****************************************************************
# Code Strategies
#******************************************************************
# Buy & Hold
data$weight[] = NA
data$weight[] = 1
buy.hold = bt.run.share(data)
#*****************************************************************
# Create Report
#******************************************************************
plotbt(buy.hold, plotX = T, log = 'y', LeftMargin = 3)
mtext('Cumulative Performance', side = 2, line = 1)
}
###############################################################################
# Remove symbols from environment
#' @export
###############################################################################
bt.prep.remove.symbols.min.history <- function
(
b, # environment with symbols time series
min.history = 1000 # minmum number of observations
)
{
bt.prep.remove.symbols(b, which( count(b$prices, side=2) < min.history ))
}
#' @export
bt.prep.remove.symbols <- function
(
b, # environment with symbols time series
index # index of symbols to remove
)
{
if( len(index) > 0 ) {
if( is.character(index) ) index = match(index, b$symbolnames)
b$prices = b$prices[, -index]
b$weight = b$weight[, -index]
b$execution.price = b$execution.price[, -index]
env.rm(b$symbolnames[index], b)
b$symbolnames = b$symbolnames[ -index]
}
}
###############################################################################
#' Trim data environment
#'
#' This function will remove weights that are smaller than given threshold
#'
#' @param b original environment with symbols time series
#' @param dates dates to keep from original environment
#'
#' @return updated environment with symbols time series
#'
#' @examples
#' \dontrun{
#' bt.prep.trim(data, endpoints(data$prices, 'months'))
#' bt.prep.trim(data, '2006::')
#' }
#' @export
bt.prep.trim <- function
(
b, # environment with symbols time series
dates = NULL # subset of dates
)
{
if(is.null(dates)) return(b)
# convert dates to dates.index
dates.index = dates2index(b$prices, dates)
data.copy <- new.env()
for(s in b$symbolnames) data.copy[[s]] = b[[s]][dates.index,,drop=F]
data.copy$symbolnames = b$symbolnames
data.copy$dates = b$dates[dates.index]
data.copy$prices = b$prices[dates.index,,drop=F]
data.copy$weight = b$weight[dates.index,,drop=F]
data.copy$execution.price = b$execution.price[dates.index,,drop=F]
return(data.copy)
}
###############################################################################
# Helper function to backtest for type='share'
#' @export
###############################################################################
bt.run.share <- function
(
b, # environment with symbols time series
prices = b$prices, # prices
clean.signal = T, # flag to remove excessive signal
trade.summary = F, # flag to create trade summary
do.lag = 1, # lag signal
do.CarryLastObservationForwardIfNA = TRUE,
silent = F,
capital = 100000,
commission = 0,
weight = b$weight,
dates = 1:nrow(b$prices)
)
{
# make sure that prices are available, assume that
# weights account for missing prices i.e. no price means no allocation
prices[] = bt.apply.matrix(coredata(prices), ifna.prev)
weight = mlag(weight, do.lag - 1)
do.lag = 1
if(clean.signal)
weight[] = bt.exrem(weight)
weight = (capital / prices) * weight
bt.run(b,
trade.summary = trade.summary,
do.lag = do.lag,
do.CarryLastObservationForwardIfNA = do.CarryLastObservationForwardIfNA,
type='share',
silent = silent,
capital = capital,
commission = commission,
weight = weight,
dates = dates)
}
###############################################################################
# Run backtest
#
# Inputs are assumed as if they were computed at point in time (i.e. no lags)
#
# For 'weight' back-test, the default action is to lage weights by one day,
# because weights are derived using all the information avalaible today,
# so we can only implement these weights tomorrow:
# portfolio.returns = lag(weights,1) * returns = weights * ( p / lag(p,1) - 1 )
# user can specify a different lag for weights, by changing the do.lag parameter.
#
# For example, for the end of the month strategy: if we open position at the close
# on the 30th, hold position on the 31st and sell it at the close on the 1st. If our
# weights have 0 on the 30th, 1 on the 31st, 1 on the 1st, and 0 on the 2nd, we
# can specify do.lag = 0 to get correct portfolio.returns
#
# Alternatively, if our weights have 0 on the 29th, 1 on the 30st, 1 on the 31st, and 0 on the 1nd, we
# can leave do.lag = 1 to get correct portfolio.returns
#
# For 'share' back-test, the portfolio returns:
# portfolio.returns = lag(shares,1) * ( p - lag(p,1) ) / ( lag(shares,1) * lag(p,1) )
#
###############################################################################
# some operators do not work well on xts
# weight[] = apply(coredata(weight), 2, ifna_prev)
#' @export
###############################################################################
bt.run <- function
(
b, # environment with symbols time series
trade.summary = F, # flag to create trade summary
do.lag = 1, # lag signal
do.CarryLastObservationForwardIfNA = TRUE,
type = c('weight', 'share'),
silent = F,
capital = 100000,
commission = 0,
weight = b$weight,
dates = 1:nrow(b$prices)
)
{
# convert dates to dates.index
dates.index = dates2index(b$prices, dates)
# setup
type = type[1]
# create signal
weight[] = ifna(weight, NA)
# lag
if(do.lag > 0)
weight = mlag(weight, do.lag) # Note k=1 implies a move *forward*
# backfill
if(do.CarryLastObservationForwardIfNA)
weight[] = apply(coredata(weight), 2, ifna.prev)
weight[is.na(weight)] = 0
# find trades
weight1 = mlag(weight, -1)
tstart = weight != weight1 & weight1 != 0
tend = weight != 0 & weight != weight1
trade = ifna(tstart | tend, FALSE)
# prices
prices = b$prices
# execution.price logic
if( sum(trade) > 0 ) {
execution.price = coredata(b$execution.price)
prices1 = coredata(b$prices)
prices1[trade] = iif( is.na(execution.price[trade]), prices1[trade], execution.price[trade] )
prices[] = prices1
}
# type of backtest
if( type == 'weight') {
ret = prices / mlag(prices) - 1
ret[] = ifna(ret, NA)
ret[is.na(ret)] = 0
} else { # shares, hence provide prices
ret = prices
}
#weight = make.xts(weight, b$dates)
temp = b$weight
temp[] = weight
weight = temp
# prepare output
bt = bt.summary(weight, ret, type, b$prices, capital, commission)
bt$dates.index = dates.index
bt = bt.run.trim.helper(bt, dates.index)
if( trade.summary ) bt$trade.summary = bt.trade.summary(b, bt)
if( !silent ) {
# print last signal / weight observation
cat('Latest weights :\n')
print(round(100*last(bt$weight),2))
cat('\n')
cat('Performance summary :\n')
cat('', spl('CAGR,Best,Worst'), '\n', sep = '\t')
cat('', sapply(cbind(bt$cagr, bt$best, bt$worst), function(x) round(100*x,1)), '\n', sep = '\t')
cat('\n')
}
return(bt)
}
# trim bt object, used internally
#' @export
bt.run.trim.helper = function(bt, dates.index) {
n.dates = len(dates.index)
for(n in ls(bt)) {
if( !is.null(dim(bt[[n]])) ) {
if( nrow(bt[[n]]) > n.dates )
bt[[n]] = bt[[n]][dates.index,,drop=F]
} else if( len(bt[[n]]) > n.dates )
bt[[n]] = bt[[n]][dates.index]
}
bt$equity = bt$equity / as.double(bt$equity[1])
bt$best = max(bt$ret)
bt$worst = min(bt$ret)
bt$cagr = compute.cagr(bt$equity)
bt
}
###############################################################################
#tic(11)
#for(j in 1:10)
# a = as.vector(prices)
#toc(11)
#
#tic(11)
#for(j in 1:10)
# a = coredata(prices)
#toc(11)
# Interestingly coredata is a lot faster
#
###############################################################################
# Backtest summary
#' @export
###############################################################################
bt.summary <- function
(
weight, # signal / weights matrix
ret, # returns for type='weight' and prices for type='share'
type = c('weight', 'share'),
close.prices,
capital = 100000,
commission = 0 # cents / share commission
)
{
# cents / share commission
# trade cost = abs(share - mlag(share)) * commission$cps
# fixed commission per trade to more effectively to penalize for turnover
# trade cost = sign(abs(share - mlag(share))) * commission$fixed
# percentage commission
# trade cost = price * abs(share - mlag(share)) * commission$percentage
# Todo
# - ability to set different commissions at start/end of the trade
# - add percentage.fixed, same logic as fixed, but applied to capital
# trade cost = sign(abs(share - mlag(share))) *
# price * abs(pmax(share, mlag(share))) * commission$percentage.fixed
# - ability to set commissions for each asset; hence modeling different tax
# rates for stock and equities; these special commissions/taxes should
# only be applied at the year end and depend on asset and holding period
# only for trades that were completed in the given year => probably a separate function
if( !is.list(commission) ) {
if( type == 'weight')
commission = list(cps = 0.0, fixed = 0.0, percentage = commission)
else
commission = list(cps = commission, fixed = 0.0, percentage = 0.0)
}
type = type[1]
n = nrow(ret)
bt = list()
bt$weight = weight
bt$type = type
# for commission calculations, un lag the signal
com.weight = mlag(weight,-1)
if( type == 'weight') {
temp = ret[,1]
temp[] = rowSums(ret * weight) -
rowSums(abs(com.weight - mlag(com.weight)) * commission$percentage, na.rm=T)
- rowSums(sign(abs(com.weight - mlag(com.weight))) * commission$fixed, na.rm=T)
bt$ret = temp
#bt$ret = make.xts(rowSums(ret * weight) - rowSums(abs(weight - mlag(weight))*commission, na.rm=T), index.xts(ret))
#bt$ret = make.xts(rowSums(ret * weight), index.xts(ret))
} else {
bt$share = weight
bt$capital = capital
prices = ret
# backfill prices
#prices1 = coredata(prices)
#prices1[is.na(prices1)] = ifna(mlag(prices1), NA)[is.na(prices1)]
#prices[] = prices1
prices[] = bt.apply.matrix(coredata(prices), ifna.prev)
close.prices[] = bt.apply.matrix(coredata(close.prices), ifna.prev)
# new logic
#cash = capital - rowSums(bt$share * mlag(prices), na.rm=T)
cash = capital - rowSums(bt$share * mlag(close.prices), na.rm=T)
# find trade dates
share.nextday = mlag(bt$share, -1)
tstart = bt$share != share.nextday & share.nextday != 0
tend = bt$share != 0 & bt$share != share.nextday
trade = ifna(tstart | tend, FALSE)
tstart = trade
index = mlag(apply(tstart, 1, any))
index = ifna(index, FALSE)
index[1] = T
totalcash = NA * cash
totalcash[index] = cash[index]
totalcash = ifna.prev(totalcash)
totalcash = ifna(totalcash,0) # check this
# We can introduce transaction cost to portfolio returns as
# abs(bt$share - mlag(bt$share)) * 0.01
portfolio.ret = (totalcash + rowSums(bt$share * prices, na.rm=T)
- rowSums(abs(com.weight - mlag(com.weight)) * commission$cps, na.rm=T)
- rowSums(sign(abs(com.weight - mlag(com.weight))) * commission$fixed, na.rm=T)
- rowSums(prices * abs(com.weight - mlag(com.weight)) * commission$percentage, na.rm=T)
) / (totalcash + rowSums(bt$share * mlag(prices), na.rm=T) ) - 1
#portfolio.ret = (totalcash + rowSums(bt$share * prices, na.rm=T) ) / (totalcash + rowSums(bt$share * mlag(prices), na.rm=T) ) - 1
bt$weight = bt$share * mlag(prices) / (totalcash + rowSums(bt$share * mlag(prices), na.rm=T) )
bt$weight[is.na(bt$weight)] = 0
#bt$ret = make.xts(ifna(portfolio.ret,0), index.xts(ret))
temp = ret[,1]
temp[] = ifna(portfolio.ret,0)
temp[1] = 0
bt$ret = temp
}
bt$best = max(bt$ret)
bt$worst = min(bt$ret)
bankrupt = which(bt$ret <= -1)
if(len(bankrupt) > 0) bt$ret[bankrupt[1]:n] = -1
bt$equity = cumprod(1 + bt$ret)
bt$cagr = compute.cagr(bt$equity)
return(bt)
}
bt.summary.test <- function() {
#*****************************************************************
# Load historical data
#******************************************************************
load.packages('quantmod')
data <- new.env()
getSymbols('EEM', src = 'yahoo', from = '1980-01-01', env = data, auto.assign = T)
bt.prep(data, align='keep.all', dates='2013:08::2013:08:10')
buy.date = '2013:08:05'
sell.date = '2013:08:06'
#*****************************************************************
# Code Strategies
#******************************************************************
# set dummy prices
coredata(data$prices) <-c(10,20,40,60,20,160,60)
prices = data$prices
# weight back-test
data$weight[] = NA
data$weight[buy.date] = 1
data$weight[sell.date] = 0
commission = list(cps = 0.0, fixed = 0.0, percentage = 1/100)
model3 = bt.run(data, commission = commission, silent = T)
model3$ret
#There is 1% drop on 5th due to buying stock, and on the 6th return is 0.49 = 0.5 - 0.01 (commission)
# share back-test
data$weight[] = NA
data$weight[buy.date] = 1
data$weight[sell.date] = 0
commission = list(cps = 0.0, fixed = 0.0, percentage = 1/100)
model3 = bt.run.share(data, commission = commission, capital = 100000, silent = T)
model3$ret
#There is 1% drop on 5th due to buying stock, and on the 6th return is
#0.485 = (2500 * 60 - 2500 * 60 * 0.01) / (2500 * 40) - 1
#i.e. return = (share * price + cash - total.commission) / (share * mlag(price) + cash) - 1
}
###############################################################################
# Remove all leading NAs in model equity
#' @export
###############################################################################
bt.trim <- function
(
...,
dates = '::'
)
{
models = variable.number.arguments( ... )
for( i in 1:len(models) ) {
bt = models[[i]]
n = len(bt$equity)
first = which.max(!is.na(bt$equity) & bt$equity != 1)
if(first > 1 && !is.na(bt$equity[(first-1)]))
first = first - 1
if (first < n) {
index = first:n
dates.range = range(dates2index(bt$equity[index],dates))
index = index[dates.range[1]] : index[dates.range[2]]
bt$dates.index = bt$dates.index[index]
bt$equity = bt$equity[index]
bt$equity = bt$equity / as.double(bt$equity[1])
bt$ret = bt$ret[index]
if (!is.null(bt$weight)) bt$weight = bt$weight[index,,drop=F]
if (!is.null(bt$share)) bt$share = bt$share[index,,drop=F]
bt$best = max(bt$ret)
bt$worst = min(bt$ret)
bt$cagr = compute.cagr(bt$equity)
}
models[[i]] = bt
}
return (models)
}
bt.trim.test <- function() {
#*****************************************************************
# Load historical data
#******************************************************************
load.packages('quantmod')
data <- new.env()
getSymbols(spl('SPY,GLD'), src = 'yahoo', from = '1980-01-01', env = data, auto.assign = T)
for(i in ls(data)) data[[i]] = adjustOHLC(data[[i]], use.Adjusted=T)
bt.prep(data, align='keep.all')
#*****************************************************************
# Code Strategies
#******************************************************************
models = list()
data$weight[] = NA
data$weight$SPY[] = 1
models$SPY = bt.run.share(data, clean.signal=F)
data$weight[] = NA
data$weight$GLD[] = 1
models$GLD = bt.run.share(data, clean.signal=F)
strategy.performance.snapshoot(bt.trim(models), T)
}
# bt.run - really fast with no bells or whisles
# working directly with xts is alot slower, so use coredata
#' @export
bt.run.weight.fast <- function
(
b, # environment with symbols time series
do.lag = 1, # lag signal
do.CarryLastObservationForwardIfNA = TRUE
)
{
# Signal => weight
weight = ifna(coredata(b$weight), NA)
# lag
if(do.lag > 0) weight = mlag(weight, do.lag) # Note k=1 implies a move *forward*
# backfill
if(do.CarryLastObservationForwardIfNA) weight[] = apply(coredata(weight), 2, ifna.prev)
weight[is.na(weight)] = 0
# returns
prices = coredata(b$prices)
ret = prices / mlag(prices) - 1
ret[] = ifna(ret, 0)
ret = rowSums(ret * weight)
# prepare output
list(weight = weight, ret = ret, equity = cumprod(1 + ret))
}
###############################################################################
# Portfolio turnover
# http://wiki.fool.com/Portfolio_turnover
# sales or purchases and dividing it by the average monthly value of the fund's assets
#' @export
###############################################################################
compute.turnover <- function
(
bt, # backtest object
b, # environment with symbols time series
exclude.first.trade = T # first trade is the reason for 100% turnover
# i.e. going from cash to fully invested
)
{
year.ends = unique(c(endpoints(bt$weight, 'years'), nrow(bt$weight)))
year.ends = year.ends[year.ends>0]
nr = len(year.ends)
period.index = c(1, year.ends)
# find first investment date
first = which.max(!is.na(bt$equity) & bt$equity != 1)
if(first > 1 && !is.na(bt$equity[(first-1)]))
first = first - 1
if( bt$type == 'weight') {
portfolio.value = rowSums(abs(bt$weight), na.rm=T)
portfolio.turnover = rowSums( abs(bt$weight - mlag(bt$weight)), na.rm=T)
portfolio.turnover[ rowSums( !is.na(bt$weight) & !is.na(mlag(bt$weight)) ) == 0 ] = NA
} else {
prices = mlag(b$prices[bt$dates.index,,drop=F])
if( is.null(bt$cash) ) {
# logic from bt.summary function
cash = bt$capital - rowSums(bt$share * prices, na.rm=T)
# find trade dates
share.nextday = mlag(bt$share, -1)
tstart = bt$share != share.nextday & share.nextday != 0
index = mlag(apply(tstart, 1, any))
index = ifna(index, FALSE)
totalcash = NA * cash
totalcash[index] = cash[index]
totalcash = ifna.prev(totalcash)
} else
totalcash = bt$cash
portfolio.value = totalcash + rowSums(bt$share * prices, na.rm=T)
portfolio.turnover = rowSums( prices * abs(bt$share - mlag(bt$share)), na.rm=T)
portfolio.turnover[ rowSums( !is.na(bt$share) & !is.na(mlag(bt$share)) & !is.na(prices) ) == 0 ] = NA
}
if(exclude.first.trade) portfolio.turnover[first] = 0
portfolio.turnover[1:2] = 0
temp = NA * period.index
for(iyear in 2:len(period.index)) {
temp[iyear] = sum( portfolio.turnover[ (1+period.index[iyear-1]) : period.index[iyear] ], na.rm=T) /
mean( portfolio.value[ (1+period.index[iyear-1]) : period.index[iyear] ], na.rm=T)
}
if(exclude.first.trade)
turnover = mean(temp[period.index > first], na.rm=T)
else
turnover = mean(temp[period.index >= first], na.rm=T)
ifna(turnover,0)
}
# debug
# write.xts(make.xts(bt$cash, index(bt$weight)), 'cash.csv')
# write.xts(make.xts(bt$share, index(bt$weight)), 'share.csv')
# write.xts(prices, 'price.csv')
#
# portfolio.value = make.xts(portfolio.value,index(prices))
# portfolio.turnover = make.xts(portfolio.turnover,index(prices))
# iyear='1998'
# mean(portfolio.value[iyear])
# sum(portfolio.turnover[iyear])
# sum(portfolio.turnover[iyear]) / mean(portfolio.value[iyear])
###############################################################################
# Compute Portfolio Maximum Deviation
#' @export
###############################################################################
compute.max.deviation <- function
(
bt,
target.allocation
)
{
weight = bt$weight[-1,]
max(abs(weight - repmat(target.allocation, nrow(weight), 1)))
}
###############################################################################
# Backtest Trade summary
#' @export
###############################################################################
bt.trade.summary <- function
(
b, # environment with symbols time series
bt # backtest object
)
{
if( bt$type == 'weight') weight = bt$weight else weight = bt$share
out = NULL
# find trades
weight1 = mlag(weight, -1)
tstart = weight != weight1 & weight1 != 0
tend = weight != 0 & weight != weight1
tstart[1, weight[1,] != 0] = T
n = nrow(weight)
tend[n, weight[n,] != 0] = T
#tend[1, ] = F
trade = ifna(tstart | tend, FALSE)
# prices
prices = b$prices[bt$dates.index,,drop=F]
# execution price logic
if( sum(trade) > 0 ) {
execution.price = coredata(b$execution.price[bt$dates.index,,drop=F])
prices1 = coredata(b$prices[bt$dates.index,,drop=F])
prices1[trade] = iif( is.na(execution.price[trade]), prices1[trade], execution.price[trade] )
# backfill pricess
prices1[is.na(prices1)] = ifna(mlag(prices1), NA)[is.na(prices1)]
prices[] = prices1
# get actual weights
weight = bt$weight
# extract trades
symbolnames = b$symbolnames
nsymbols = len(symbolnames)
ntrades = max(sum(tstart,na.rm=T), sum(tend,na.rm=T))
trades = matrix(NA,nr=ntrades,nc=7)
colnames(trades) = spl('date,symbol,weight,entry.date,exit.date,entry.price,exit.price')
itrade = 1
for( i in 1:nsymbols ) {
tstarti = which(tstart[,i])
tendi = which(tend[,i])
#cat(colnames(data$prices)[i], len(tstarti), len(tendi), '\n')
if( len(tstarti) > 0 ) {
#if( len(tendi) < len(tstarti) ) tendi = c(tendi, nrow(weight))
if( len(tendi) > len(tstarti) ) tstarti = c(1, tstarti)
ntrade = len(tstarti)
ntrade.index = itrade:(itrade+ntrade-1)
trades[ntrade.index,] =
cbind((tstarti+1), i, coredata(weight[(tstarti+1), i]),
tstarti, tendi,
coredata(prices[tstarti, i]), coredata(prices[tendi,i])
)
itrade = itrade + ntrade
}
}
# prepare output
out = list()
out$stats = cbind(
bt.trade.summary.helper(trades),
bt.trade.summary.helper(trades[trades[, 'weight'] >= 0, , drop=F]),
bt.trade.summary.helper(trades[trades[, 'weight'] <0, , drop=F])
)
colnames(out$stats) = spl('All,Long,Short')
dates = index(weight)
dates0 = format(dates, '%Y-%m-%d')
index = order(dates[trades[,'entry.date']])