Bilello_Gayed_02_DBS_indicator_quantstrat
Dennis Lee
3 October 2015
==================================================================================================================================
Motivational Buckets
The objective is to duplicate the MTA 2014 Charles Dow Award Winner’s Beta Rotation Strategy (BRS) by Pension Partners (Bilello C.V. & Gayed M.A.). When the price ratio (or relative strength) of the Utilities sector to the broad market is positive over the four(4)-week period, position into Utilities for the following week. When the ratio (or the relative strength) is negative, position into the broad market for the following week. This strategy uses TWO (2) ETFS with the best approximation of the Fama-French data set for the Utilities sector and broad market, namely Utilities Select Sector SPDR Fund (“XLU”) and the Vanguard Total Stock Market ETF (“VTI”), respectively. URL: www.ssrn.com/abstract=2417974
set user.strategyName - strategy name, e.g. “BetaRotation”
set user.initDate - set initDate prior to the first close price given using the format “YYYY-MM-DD”, e.g. “2005-01-01”
set user.endDate - set endDate using the format “YYYY-MM-DD”, e.g. as.character(Sys.Date())
set user.tradeSize - set trade size in units, e.g. 100
set user.initEq - set initial equity or starting capital (default=100000)
set user.sym - set character vector of stock symbols
Example
> install.packages(“FinancialInstrument”, dependencies = TRUE)
> install.packages(“quantstrat”, repos=“http://R-Forge.R-project.org”, dependencies = TRUE)
> install.packages(“devtools”, dependencies = TRUE)
> devtools::install_github(repo=“IKTrading”, username=“IlyaKipnis”)
1.1) (size: 44) Set user inputs and fetch prices of user symbols. Create new symbol RTO = XLU / VTI.
1.2) (size: 7) Create four objects: (i) portfolio, (ii) account, (iii) orders, and (iv) strategy.
1.3) (size: 9) Create an indicator function dbsXts() that takes in some variant of OHLC and returns an Xts object.
1.4) (size: 5) Use the function applyIndicators() to see how the indicators will be appended to the mktdata object.
1.5) (size: 2) Plot a chart of S&P 500 with the DBS indicator.
1.6) (size: 14) Use the function applySignals() to see how the signals will be appended to the mktdata object.
1.7) (size: 23) Use the function applyRules(), instead of applyStrategy(), to backtest a different symbol that is not associated with portfolio.name. Call the three update functions as the final step to creating a backtesting result.
1.8) (size: 4) Displays a chart of SPY, the holding positions, the cumulative P&L and drawdown during backtesting.
==================================================================================================================================
1.1 (size: 44) The initial equity or starting capital is 100000 unless otherwise given. In this case, the initial equity is the number of symbols times 100000. The function stock() from package FinancialInstrument defines each symbol as a stock currency instrument. Create a new symbol by XLU divide by VTI.
if( Sys.info()["sysname"] == "Linux" )
source("~/100 FxOption/103 FxOptionVerBack/081 Fx Bit/R-source/PlusReg.R", echo=FALSE)
if( Sys.info()["sysname"] == "Windows" )
{
if( Sys.info()["nodename"] == "GARYLEE-PC" )
{
source("D:/denbrige/100 FxOption/103 FxOptionVerBack/081 Fx Bit/R-source/PlusReg.R", echo=FALSE)
RegHomeDir <- function()
{
retDir <- NULL
if( RegIsLinuxBln() )
retDir <- paste0("/home/",Sys.info()["user"],"/")
if( RegIsWindowsBln() )
retDir <- paste0("D:/",Sys.info()["user"],"/")
retDir
}
}
else
{
source("C:/Users/denbrige/100 FxOption/103 FxOptionVerBack/081 Fx Bit/R-source/PlusReg.R", echo=FALSE)
}
}
suppressPackageStartupMessages(source(paste0(RegRSourceDir(),"PlusAddThat.R"), echo=FALSE))
suppressPackageStartupMessages(source(paste0(RegRSourceDir(),"PlusDbs.R"), echo=FALSE))
suppressPackageStartupMessages(require( quantmod ))
suppressPackageStartupMessages(require( quantstrat ))
suppressPackageStartupMessages(require( PerformanceAnalytics ))
Sys.setenv(TZ="UTC")
invisible( currency('USD') )
options(width=120)
user.strategyName <- "BetaRotation"
user.initDate <- "2005-01-01"
user.endDate <- as.character(Sys.Date())
user.tradeSize <- 100
user.sym <- c("SPY") #S&P500
user.initEq <- user.tradeSize * length(user.sym)
dbs.sym <- c("XLU", #Utilities Select Sector SPDR Fund
"VTI", #Vanguard Total Stock Market ETF
user.sym)
suppressWarnings(suppressMessages(getSymbols(dbs.sym, from=user.initDate, to=user.endDate, src="yahoo", adjust=TRUE)))## [1] "XLU" "VTI" "SPY"RTO <- XLU / VTI
names(RTO) <- c("RTO.Open", "RTO.High", "RTO.Low", "RTO.Close", "RTO.Volume", "RTO.Adjusted")
portfolio.sym <- c("RTO", user.sym)
stock(portfolio.sym, currency="USD", multiplier=1)## [1] "RTO" "SPY"head(RTO)## RTO.Open RTO.High RTO.Low RTO.Close RTO.Volume RTO.Adjusted
## 2005-01-03 0.3886388 0.3919688 0.3904092 0.3901185 3.859810 0.3901185
## 2005-01-04 0.3886984 0.3908185 0.3933024 0.3926524 5.272727 0.3926524
## 2005-01-05 0.3931402 0.3924621 0.3891692 0.3891692 3.760242 0.3891692
## 2005-01-06 0.3881458 0.3891317 0.3889568 0.3893292 3.826561 0.3893292
## 2005-01-07 0.3914953 0.3926185 0.3910580 0.3905816 1.356195 0.3905816
## 2005-01-10 0.3891558 0.3902384 0.3892234 0.3910392 3.535091 0.3910392==================================================================================================================================
1.2 (size: 7) Create four objects: (i) portfolio, (ii) account, (iii) orders, and (iv) strategy. The portfolio consists of user-defined symbols to which the strategy will be applied. Set store=TRUE to store the strategy in the .strategy environment, instead of return it. Set indicator parameters.
strategy.name <- portfolio.name <- account.name <- user.strategyName
invisible( initPortf(portfolio.name, symbols=portfolio.sym, initDate=user.initDate, currency='USD') )
invisible( initAcct(account.name, portfolios=portfolio.name, initDate=user.initDate, currency='USD', initEq=user.initEq) )
initOrders(portfolio.name, initDate=user.initDate)
strategy(strategy.name, store=TRUE)
user.dbs.periodNum <- 20
user.dbs.threshold <- 0.0==================================================================================================================================
1.3 (size: 9) Create an indicator function dbsXts() that takes in some variant of an OHLC object (whether one column-most likely Close, HLC, or whatever else) and a period (default=20). In theory, you could replace the function dbsXts() with TTR::ROC() as both functions perform the same transformation. In the latter case, add.indicator( strategy.name, name=“ROC”, arguments=list( OHLC=quote(OHLC(mktdata)), n=user.dbs.periodNum, type=“discrete”), label=“roc” ).
dbsXts <- function(OHLC, periodNum=20)
{
retXts <- merge.xts( ROC(Op(OHLC), periodNum, type="discrete"),
ROC(Hi(OHLC), periodNum, type="discrete"),
ROC(Lo(OHLC), periodNum, type="discrete"),
ROC(Cl(OHLC), periodNum, type="discrete") )
names(retXts) <- c("Open", "High", "Low", "Close")
retXts
}==================================================================================================================================
1.4 (size: 5) The function applyIndicators() allows a user to see how the indicators will be appended to the mktdata object in the backtest. If the function call fails, it means that there most likely is an issue with column naming.
invisible( add.indicator( strategy.name, name="dbsXts",
arguments=list(OHLC=quote(OHLC(mktdata)), n=user.dbs.periodNum),
label="dbs" ) )
test <- applyIndicators( strategy.name, mktdata=OHLC(RTO) )
head(test, 25)## RTO.Open RTO.High RTO.Low RTO.Close Open.dbs High.dbs Low.dbs Close.dbs
## 2005-01-03 0.3886388 0.3919688 0.3904092 0.3901185 NA NA NA NA
## 2005-01-04 0.3886984 0.3908185 0.3933024 0.3926524 NA NA NA NA
## 2005-01-05 0.3931402 0.3924621 0.3891692 0.3891692 NA NA NA NA
## 2005-01-06 0.3881458 0.3891317 0.3889568 0.3893292 NA NA NA NA
## 2005-01-07 0.3914953 0.3926185 0.3910580 0.3905816 NA NA NA NA
## 2005-01-10 0.3891558 0.3902384 0.3892234 0.3910392 NA NA NA NA
## 2005-01-11 0.3903533 0.3919377 0.3906917 0.3922665 NA NA NA NA
## 2005-01-12 0.3914231 0.3924039 0.3924738 0.3916837 NA NA NA NA
## 2005-01-13 0.3938684 0.3968288 0.3957823 0.3966528 NA NA NA NA
## 2005-01-14 0.3984767 0.3981543 0.3959336 0.3982167 NA NA NA NA
## 2005-01-18 0.3973556 0.3972042 0.3963261 0.3973628 NA NA NA NA
## 2005-01-19 0.3991075 0.3985925 0.3992216 0.3992216 NA NA NA NA
## 2005-01-20 0.3975794 0.4011823 0.3997392 0.4003577 NA NA NA NA
## 2005-01-21 0.3984474 0.4003745 0.4011049 0.4008279 NA NA NA NA
## 2005-01-24 0.4025037 0.4049460 0.4034509 0.4065085 NA NA NA NA
## 2005-01-25 0.4054375 0.4054533 0.4035745 0.4031286 NA NA NA NA
## 2005-01-26 0.4024419 0.4064128 0.4024549 0.4064748 NA NA NA NA
## 2005-01-27 0.4085707 0.4076290 0.4052445 0.4079392 NA NA NA NA
## 2005-01-28 0.4053252 0.4081251 0.4076980 0.4098802 NA NA NA NA
## 2005-01-31 0.4087238 0.4102954 0.4080662 0.4103668 NA NA NA NA
## 2005-02-01 0.4112325 0.4111180 0.4096789 0.4107884 0.05813533 0.04885375 0.04935765 0.05298384
## 2005-02-02 0.4133393 0.4115653 0.4105047 0.4104872 0.06339345 0.05308543 0.04373826 0.04542123
## 2005-02-03 0.4133393 0.4130189 0.4099221 0.4136933 0.05137887 0.05237902 0.05332633 0.06301673
## 2005-02-04 0.4153796 0.4133237 0.4139472 0.4125424 0.07016389 0.06216923 0.06424995 0.05962356
## 2005-02-07 0.4148465 0.4147766 0.4120039 0.4118676 0.05964624 0.05643667 0.05356218 0.05449819==================================================================================================================================
1.5) (size: 2) Plot a chart of S&P 500 with the DBS indicator.
barChart(OHLC(SPY), theme="white", name="S&P 500")
addTA(Cl(test)[,2], legend="Price Ratio Utilities/Market (Neg=Long Market; Pos=Long Utilities)")
==================================================================================================================================
1.6) (size: 14) The function add.signal() allows a user to define the relationship between two columns (or between the first column and a threshold value). The cross argument determines whether the signal should return TRUE for the entire duration of the relationship being true (cross=FALSE), or only on the first day (cross=TRUE). In other words, if you set cross=TRUE, the signal should return true only on the first day, otherwise if you set cross=FALSE, the signal should return TRUE for the entire period in which the condition is met. Relationships are specified with a two or three character identifier, such as ‘gt’, ‘lt’, ‘gte’, ‘lte’, ‘eq’. The two signals ‘sigDbsAboveThreshold’ and ‘sigDbsBelowThreshold’ are best used for setting up filters, i.e. we will look for a long entry only when there is a bullish signal and vice-versa. These signals should return TRUE for the entire duration in which the condition is met, and FALSE otherwise. The other two signals ‘sigDbsCrossUp’ and ‘sigDbsCrossDn’ are useful for setting up buy or sell orders. We want to integrate AsiaChart’s RSI Pop strategy here. For example, we want to enter a position only on the second cross signal. The first cross signal setups the entry criteria whether to look for a long/short entry in the future.
invisible( add.signal( strategy.name, name="sigThreshold",
arguments=list(column="Close.dbs", threshold=user.dbs.threshold, relationship="gt", cross=FALSE),
label="bearish_or_exit_long") )
invisible( add.signal( strategy.name, name="sigThreshold",
arguments=list(column="Close.dbs", threshold=-user.dbs.threshold, relationship="lt", cross=FALSE),
label="bullish_or_exit_short") )
invisible( add.signal( strategy.name, name="sigThreshold",
arguments=list(column="Close.dbs", threshold=user.dbs.threshold, relationship="gt", cross=TRUE),
label="short") )
invisible( add.signal( strategy.name, name="sigThreshold",
arguments=list(column="Close.dbs", threshold=-user.dbs.threshold, relationship="lt", cross=TRUE),
label="long") )
tst2 <- applySignals( strategy.name, mktdata=test )
spyXts <- merge.xts(SPY[,1:4], tst2[,5:12])==================================================================================================================================
1.7 (size: 21) The function applyRules(), unlike applySignals() and applyIndicators(), does not return the mktdata object. The reason we use the function applyRules(), instead of applyStrategy(), is because we want to use a different symbol that is not associated with portfolio.name. The three update functions are the final step to creating a backtesting result.
invisible( add.rule(strategy = strategy.name,
name='ruleSignal',
arguments=list(sigcol='long',
sigval=TRUE,
orderqty=user.tradeSize,
ordertype='market',
orderside='long'),
type='enter') )
invisible( add.rule(strategy = strategy.name,
name='ruleSignal',
arguments=list(sigcol='short',
sigval=TRUE,
orderqty='all',
ordertype='market',
orderside='long'),
type='exit') )
invisible( applyRules( portfolio.name, symbol=user.sym, strategy.name, mktdata=spyXts, verbose=FALSE ) )
ob <- getOrderBook(portfolio.name)
invisible( updatePortf(portfolio.name) )
invisible( updateAcct(portfolio.name) )
invisible( updateEndEq(portfolio.name) )==================================================================================================================================
1.8 (size: 6) The function chart.Posn() displays a chart of the user symbol, the holding positions, the cumulative P&L and drawdown. We obtain the equity curve by calling getAccount(), which returns an account object. We plot the equity curve by passing a\(summary\)End.Eq as input to the function plot(). The Sharpe ratio and other performance metrics can be obtained by passing the strategy name as input to the function tradeStats().
chart.Posn(strategy.name, Symbol=user.sym)
a <- getAccount(portfolio.name)
equity <- a$summary$End.Eq
plot(equity, main="Beta Rotation Strategy Equity Curve")
tstats <- t(tradeStats(strategy.name))
tstats## SPY
## Portfolio "BetaRotation"
## Symbol "SPY"
## Num.Txns "236"
## Num.Trades "117"
## Net.Trading.PL "4015.11"
## Avg.Trade.PL "34.31718"
## Med.Trade.PL "-11.89852"
## Largest.Winner "1888.305"
## Largest.Loser "-848.6655"
## Gross.Profits "16674.01"
## Gross.Losses "-12658.9"
## Std.Dev.Trade.PL "366.3552"
## Percent.Positive "46.15385"
## Percent.Negative "53.84615"
## Profit.Factor "1.317177"
## Avg.Win.Trade "308.7779"
## Med.Win.Trade "209.6551"
## Avg.Losing.Trade "-200.9348"
## Med.Losing.Trade "-151.9314"
## Avg.Daily.PL "34.31718"
## Med.Daily.PL "-11.89852"
## Std.Dev.Daily.PL "366.3552"
## Ann.Sharpe "1.486995"
## Max.Drawdown "-3911.967"
## Profit.To.Max.Draw "1.026366"
## Avg.WinLoss.Ratio "1.536707"
## Med.WinLoss.Ratio "1.379933"
## Max.Equity "5415.922"
## Min.Equity "-1584.868"
## End.Equity "4015.11"==================================================================================================================================