Replicating VIX-Adjusted Momentum using India VIX and the NIFTY 50 index.
library(tidyverse)
library(ggthemes)
library(reshape2)
library(odbc)
library(plutoR)
library(quantmod)
library(lubridate)
library(ggrepel)
library(PerformanceAnalytics)
options("scipen"=999)
options(stringsAsFactors = FALSE)
options(repr.plot.width=16, repr.plot.height=8)
source("config.R")
source("goofy/plot.common.R")
source("goofy/misc.common.R")
indices <- Indices()── �[1mAttaching packages�[22m ─────────────────────────────────────── tidyverse 1.2.1 ──
�[32m✔�[39m �[34mggplot2�[39m 3.2.0 �[32m✔�[39m �[34mpurrr �[39m 0.3.2
�[32m✔�[39m �[34mtibble �[39m 2.1.3 �[32m✔�[39m �[34mdplyr �[39m 0.8.1
�[32m✔�[39m �[34mtidyr �[39m 0.8.3 �[32m✔�[39m �[34mstringr�[39m 1.4.0
�[32m✔�[39m �[34mreadr �[39m 1.3.1 �[32m✔�[39m �[34mforcats�[39m 0.4.0
── �[1mConflicts�[22m ────────────────────────────────────────── tidyverse_conflicts() ──
�[31m✖�[39m �[34mdplyr�[39m::�[32mfilter()�[39m masks �[34mstats�[39m::filter()
�[31m✖�[39m �[34mdplyr�[39m::�[32mlag()�[39m masks �[34mstats�[39m::lag()
Attaching package: ‘reshape2’
The following object is masked from ‘package:tidyr’:
smiths
Loading required package: xts
Loading required package: zoo
Attaching package: ‘zoo’
The following objects are masked from ‘package:base’:
as.Date, as.Date.numeric
Attaching package: ‘xts’
The following objects are masked from ‘package:dplyr’:
first, last
Loading required package: TTR
Version 0.4-0 included new data defaults. See ?getSymbols.
Attaching package: ‘lubridate’
The following object is masked from ‘package:base’:
date
Attaching package: ‘PerformanceAnalytics’
The following object is masked from ‘package:graphics’:
legend
Registering fonts with R
Attaching package: ‘gridExtra’
The following object is masked from ‘package:dplyr’:
combinestartDate <- as.Date('2009-03-03') #the day when India VIX was launched
endDate <- as.Date('2019-06-30')
vixDt <- indices$IndiaVixTimeSeries() %>%
filter(TIME_STAMP >= startDate & TIME_STAMP <= endDate) %>%
select(TIME_STAMP, CLOSE) %>%
collect()
nifty50Dt <- indices$NseTimeSeries() %>%
filter(NAME == 'NIFTY 50' & TIME_STAMP >= startDate & TIME_STAMP <= endDate) %>%
select(TIME_STAMP, CLOSE) %>%
collect()doLookback <- function(lb){
allXts <- merge(xts(nifty50Dt$CLOSE, nifty50Dt$TIME_STAMP), xts(vixDt$CLOSE, vixDt$TIME_STAMP))
names(allXts) <- c('NIFTY', 'VIX')
allXts$SMA <- SMA(allXts$NIFTY, lb)
allXts$D_RET <- dailyReturn(allXts$NIFTY)
allXts$D_RET_LAG_1 <- stats::lag(allXts$D_RET, -1) #signal is generated and traded at the close, so use next day's return
allXts$VIX_ADJ_RET <- allXts$D_RET/allXts$VIX
allXts$VIX_ADJ_RET_SMA <- SMA(allXts$VIX_ADJ_RET, lb)
allXts$ROC <- rollapply(allXts$D_RET, lb, Return.cumulative) #simple compound return or ROC (rate of change)
allXts <- na.omit(allXts)
#go long only if the index is above its 200-day average
allXts$SMA_STRAT <- ifelse(allXts$NIFTY > allXts$SMA, allXts$D_RET_LAG_1, 0)
#go long only if the simple average (VIX-Adjusted Momentum) > 0
allXts$VIX_STRAT <- ifelse(allXts$VIX_ADJ_RET_SMA > 0, allXts$D_RET_LAG_1, 0)
#go long only if the ROC > 0
allXts$ROC_STRAT <- ifelse(allXts$ROC > 0, allXts$D_RET_LAG_1, 0)
#plot returns
toPlot <- merge(allXts$VIX_STRAT, allXts$SMA_STRAT, allXts$ROC_STRAT, allXts$D_RET_LAG_1)
Common.PlotCumReturns(toPlot, sprintf("Back-test for %d look-back", lb), "")
#calculate annual returns
byYearDf <- data.frame(T=0, VIX_MOM=0.0, SMA=0.0, ROC=0.0, BH=0.0)
allYears <- unique(year(index(allXts)))
for(yr in allYears){
byYearDf <- rbind(byYearDf,
c(yr, as.numeric(100.0*Return.cumulative(allXts[toString(yr), 'VIX_STRAT'])),
as.numeric(100.0*Return.cumulative(allXts[toString(yr), 'SMA_STRAT'])),
as.numeric(100.0*Return.cumulative(allXts[toString(yr), 'ROC_STRAT'])),
as.numeric(100.0*Return.cumulative(allXts[toString(yr), 'D_RET_LAG_1']))))
}
byYearDf <- byYearDf[-1,]
print(byYearDf)
#plot annual returns
ggplot(melt(byYearDf, id='T'), aes(x=T, y=value, fill=variable)) +
theme_economist() +
geom_bar(stat="identity", position=position_dodge()) +
scale_x_continuous(labels=byYearDf$T, breaks=byYearDf$T) +
geom_text_repel(aes(label= round(value, 2)), position = position_dodge(0.9)) +
labs(x='', y='(%)', fill='', title=sprintf("Back-test for %d look-back", lb), subtitle="Annual Returns") +
annotate("text", x=max(byYearDf$T), y=min(sapply(byYearDf[,-1], min, na.rm=T)),
label = "@StockViz", hjust=1.1, vjust=-1.1, col="white", cex=6, fontface = "bold", alpha = 0.8)
} doLookback(20)
doLookback(50)
doLookback(100)
doLookback(200) T VIX_MOM SMA ROC BH
2 2009 55.8542447 43.9585559 47.986996234 70.967373
3 2010 14.7814451 12.2552717 7.696151507 17.686633
4 2011 -10.8882121 -13.2163854 -12.244349088 -24.698746
5 2012 18.9921529 14.2197847 15.513872076 28.340972
6 2013 4.9486352 5.6801427 12.472286384 5.894956
7 2014 24.3835287 22.4421699 28.362309580 31.457634
8 2015 -3.2570417 -4.0997036 -5.207613606 -3.872525
9 2016 0.6787297 -0.5741594 0.002138574 2.716245
10 2017 20.1963265 21.0425942 20.432594648 27.581759
11 2018 6.0377957 10.2504119 7.939196687 4.547436
12 2019 -1.6522020 1.2877901 -0.750851842 8.054463
T VIX_MOM SMA ROC BH
2 2009 19.5403008 2.9235720 17.122735 24.253722
3 2010 -0.8853483 16.9658616 -2.258700 17.686633
4 2011 -22.0712387 -6.9516003 -13.994246 -24.698746
5 2012 16.1229491 11.4475629 15.132554 28.340972
6 2013 1.0933239 3.5381681 -4.194506 5.894956
7 2014 24.5973807 20.6776846 24.688502 31.457634
8 2015 -6.2343550 -9.9562368 -9.921087 -3.872525
9 2016 6.9716624 10.1811402 9.549754 2.716245
10 2017 21.9358306 17.8671136 20.404191 27.581759
11 2018 -3.2803592 6.0710868 1.296014 4.547436
12 2019 6.2768668 0.6400177 6.210951 8.054463
T VIX_MOM SMA ROC BH
2 2009 14.453838 10.8847486 14.453838 14.453838
3 2010 3.074059 6.2265444 -2.693106 17.686633
4 2011 -17.124885 -13.1193054 -11.971114 -24.698746
5 2012 20.259606 13.6248986 10.471499 28.340972
6 2013 2.314944 -7.0322024 -6.741117 5.894956
7 2014 31.457634 26.7188666 32.232468 31.457634
8 2015 -12.607978 -7.9263826 -5.779633 -3.872525
9 2016 4.235006 8.7104640 2.805273 2.716245
10 2017 19.418980 19.7503868 16.912948 27.581759
11 2018 -6.710956 -0.6823267 -5.747986 4.547436
12 2019 8.017013 5.6815172 9.088848 8.054463
T VIX_MOM SMA ROC BH
2 2009 1.038931 1.038931 1.0389309 1.038931
3 2010 17.686633 12.514723 17.6866328 17.686633
4 2011 -12.349590 -9.541756 -14.9335901 -24.698746
5 2012 13.931431 7.418435 12.9133781 28.340972
6 2013 3.514348 -6.066451 -5.5739313 5.894956
7 2014 31.457634 31.457634 31.4576341 31.457634
8 2015 -6.347899 -15.667298 -1.7263752 -3.872525
9 2016 3.257257 -4.675845 0.1138283 2.716245
10 2017 27.581759 25.902444 27.5817593 27.581759
11 2018 4.547436 -4.655957 -3.6968017 4.547436
12 2019 8.054463 3.840349 5.5301819 8.054463
Looks like "best" look-back period remains 200 but it doesn't really beat buy-and-hold in terms of returns. However, its ability to dodge deep drawdowns is something that merits further investigation.