Trading With The Trend

You can only enter in the direction of the red Slow.Diff indicator, if its above zero you can take long signals, if its below zero, you can take short signals. The Fast.Diff indicator dictates the entries. When the blue line goes from negative to positive, its a long trade (and the slower red Slow.Diff indicator is above zero). Same thing for shorts. This is also referred to as a moving average crossover trading system.

To run this system, we need to build rules to hunt them down. The rules are:

if no position: red > 0 and blue-1 < 0 and blue > 0 go long
if long: blue < 0 exit long

if no position: red < 0 and blue-1 > 0 and blue < 0 go short
if short: blue > 0 exit short

This is the base of a trend following system with plenty of room for improvement.

There are different ways to code this, but a simple way is to use a lag on the fast indicator.

Let’s first remember a basic report we built with some EMA’s and comparisons of fast and slow trends.

library('quantmod')
## Loading required package: xts
## Loading required package: zoo
## 
## Attaching package: 'zoo'
## The following objects are masked from 'package:base':
## 
##     as.Date, as.Date.numeric
## Loading required package: TTR
## Version 0.4-0 included new data defaults. See ?getSymbols.
library('TTR')
getSymbols(c('SPY', 'AMZN', '^GSPC'), src='yahoo')
## 'getSymbols' currently uses auto.assign=TRUE by default, but will
## use auto.assign=FALSE in 0.5-0. You will still be able to use
## 'loadSymbols' to automatically load data. getOption("getSymbols.env")
## and getOption("getSymbols.auto.assign") will still be checked for
## alternate defaults.
## 
## This message is shown once per session and may be disabled by setting 
## options("getSymbols.warning4.0"=FALSE). See ?getSymbols for details.
## [1] "SPY"   "AMZN"  "^GSPC"

And the graph for SPY

chartSeries(SPY, theme=chartTheme('white'),
            type = c("auto", "matchsticks"), 
            subset = '2016-01::',
            show.grid = TRUE,
            major.ticks='auto', minor.ticks=TRUE,
            multi.col = FALSE,
            TA="addEMA(50, col='black');addEMA(200, col='blue')")

SPY.EMA.10 <- EMA(SPY$SPY.Close, n=10 ) 
SPY.EMA.50 <- EMA(SPY$SPY.Close, n=50 ) 
SPY.EMA.200 <- EMA(SPY$SPY.Close, n=200 ) 
SPY.Fast.Diff <- SPY.EMA.10 - SPY.EMA.50
SPY.Slow.Diff <- SPY.EMA.50 - SPY.EMA.200
addTA(SPY.Fast.Diff, col='blue', type='h',legend="10-50 MA used for in-out of market")

addTA(SPY.Slow.Diff, col='red', type='h',legend="50-200 MA give trending sense")

addTA=c(addMACD(),addVo(),addADX(n = 14, maType = "EMA"))

Here’s another example of S&P 500 SAME TIME FRAME AS CLASS

getSymbols(c('^GSPC'), src='yahoo')
## [1] "^GSPC"
chartSeries(GSPC, theme=chartTheme('white'),
            type = c("auto", "matchsticks"), 
            subset = '2007-01::2015',
            show.grid = TRUE,
            major.ticks='auto', minor.ticks=TRUE,
            multi.col = FALSE,
            TA="addEMA(50, col='black');addEMA(200, col='blue')")

GSPC.EMA.10 <- EMA(GSPC$GSPC.Close, n=10 ) 
GSPC.EMA.50 <- EMA(GSPC$GSPC.Close, n=50 ) 
GSPC.EMA.200 <- EMA(GSPC$GSPC.Close, n=200 ) 
GSPC.Fast.Diff <- GSPC.EMA.10 - GSPC.EMA.50
GSPC.Slow.Diff <- GSPC.EMA.50 - GSPC.EMA.200
addTA(GSPC.Fast.Diff, col='blue', type='h',legend="10-50 MA used for in-out of market")

addTA(GSPC.Slow.Diff, col='red', type='h',legend="50-200 MA give trending sense")

addTA=c(addMACD(),addVo(),addADX(n = 14, maType = "EMA"))

And AMZN

chartSeries(AMZN, theme=chartTheme('white'),
            type = c("auto", "matchsticks"), 
            subset = '2016-01::',
            show.grid = TRUE,
            major.ticks='auto', minor.ticks=TRUE,
            multi.col = FALSE,
            TA="addEMA(50, col='black');addEMA(200, col='blue')")

AMZN.EMA.10 <- EMA(AMZN$AMZN.Close, n=10 ) 
AMZN.EMA.50 <- EMA(AMZN$AMZN.Close, n=50 ) 
AMZN.EMA.200 <- EMA(AMZN$AMZN.Close, n=200 ) 
AMZN.Fast.Diff <- AMZN.EMA.10 - AMZN.EMA.50
AMZN.Slow.Diff <- AMZN.EMA.50 - AMZN.EMA.200
addTA(AMZN.Fast.Diff, col='blue', type='h',legend="10-50 MA used for in-out of market")

addTA(AMZN.Slow.Diff, col='red', type='h',legend="50-200 MA give trending sense")

addTA=c(addMACD(),addVo(),addADX(n = 14, maType = "EMA"))

Another key aspect in the hunt for trades is the shift function from the binhf package (https://cran.r-project.org/web/packages/binhf/index.html). This function will allow us to easily compare a vector of data with the value preceding it (or x values preceding it). Here we’ll use the shift function to find when our differenced indicators change from below to above the zero line, or from above to below (basically when the value changes sign).

Let’s install binhf package. NOTE: SMA not EMA as above.

# install.packages('binhf')
library(binhf)
## Loading required package: wavethresh
## Loading required package: MASS
## WaveThresh: R wavelet software, release 4.6.8, installed
## Copyright Guy Nason and others 1993-2016
## Note: nlevels has been renamed to nlevelsWT
## Loading required package: adlift
## Loading required package: EbayesThresh
## 
##  **********************************************
##  adlift: a package to perform wavelet lifting schemes
## 
##  --- Written by Matt Nunes and Marina Knight ---
##    Current package version:  1.4-1  ( 2018-07-09 ) 
## 
##              -+ packaged by MAN +-           
##  **********************************************
##  
##  adlift 1.4-1 loaded
## 
## Attaching package: 'adlift'
## The following object is masked from 'package:EbayesThresh':
## 
##     postmean.cauchy
## 
##  **********************************************
##  binhf: Haar-Fisz functions for binomial data
## 
##  --- Written by Matt Nunes ---
##    Current package version:  1.0-3  ( 2018-07-18 ) 
## 
##  
##  **********************************************
##  
##  binhf 1.0-3 loaded
## 
## Attaching package: 'binhf'
## The following object is masked from 'package:EbayesThresh':
## 
##     negloglik.laplace
## The following object is masked from 'package:base':
## 
##     norm
# in class he switched from EMA to SMA
GSPC.SMA.10 <- SMA(GSPC$GSPC.Close, n=10 ) 
GSPC.SMA.50 <- SMA(GSPC$GSPC.Close, n=50 ) 
GSPC.SMA.200 <- SMA(GSPC$GSPC.Close, n=200 ) 
GSPC.FastS.Diff <- GSPC.SMA.10 - GSPC.SMA.50
GSPC.SlowS.Diff <- GSPC.SMA.50 - GSPC.SMA.200
# let's look at a shift in the tail
tail(as.numeric(GSPC.FastS.Diff))
## [1] 78.52599 71.80420 65.31080 58.19900 50.41280 37.13321
tail(shift(v=as.numeric(GSPC.FastS.Diff), places=1, dir="right")) 
## [1] 78.87199 78.52599 71.80420 65.31080 58.19900 50.41280

Notice above how the data in the two tails is actually just shifted, 1 place to the left. This will allow us to compare the two tails in a column for to see a trend.

Now, we’ll code the rules:

if no position: red > 0 and blue-1 < 0 and blue > 0 go long
if long: blue < 0 exit long

if no position: red < 0 and blue-1 > 0 and blue < 0 go short
if short: blue > 0 exit short
GSPC.Long_Trades <- ifelse(
      GSPC.SlowS.Diff > 0 &
      GSPC.FastS.Diff > 0 & 
      shift(v=as.numeric(GSPC.FastS.Diff), places=1,dir = 'right') < 0,
      GSPC$GSPC.Close, NA)

GSPC.Short_Trades <- ifelse(
      GSPC.SlowS.Diff < 0 &
      GSPC.FastS.Diff < 0 & 
      shift(v=as.numeric(GSPC.FastS.Diff), places=1,dir = 'right') > 0,
      GSPC$GSPC.Close, NA)

plot(GSPC$GSPC.Close)

points(GSPC.Long_Trades, col='blue', cex=1.5, pch=18)

points(GSPC.Short_Trades, col='red', cex=1.5, pch=18)

#addTA(GSPC.Fast.Diff, col='blue', type='h',legend="10-50 MA used for in-out of market")
#addTA(GSPC.Slow.Diff, col='red', type='h',legend="50-200 MA give trending sense")
#addTA=c(addMACD(),addADX(n = 14, maType = "EMA"))

Now with SPY and EMA’s.

SPY.Long_Trades <- ifelse(
      SPY.Slow.Diff > 0 &
      SPY.Fast.Diff > 0 & 
      shift(v=as.numeric(SPY.Fast.Diff), places=1,dir = 'right') < 0,
      SPY$SPY.Close, NA)

SPY.Short_Trades <- ifelse(
      SPY.Slow.Diff < 0 &
      SPY.Fast.Diff < 0 & 
      shift(v=as.numeric(SPY.Fast.Diff), places=1,dir = 'right') > 0,
      SPY$SPY.Close, NA)

plot(SPY$SPY.Close)

points(SPY.Long_Trades, col='blue', cex=1.5, pch=18)

points(SPY.Short_Trades, col='red', cex=1.5, pch=18)

This allows us to compare the values of two different rows on the same row. We still have our indicator value of today, but we now can compare it with yesterday’s value on the same row. Sure, we could have just easily created a loop and run through each value but by doing it this way we stick to vector comparison in its simplest form.

Now, let’s translate our trend trading system pseudo code into R code:

look for long entries

Long_Trades <- ifelse( Slow.Diff > 0 & Fast.Diff > 0 & shift(v=as.numeric(Fast.Diff), places=1, dir=“right”) < 0, EWP$EWP.Close, NA)

look for long exits (same thing but inverse signts)

Short_Trades <- ifelse( Slow.Diff < 0 & Fast.Diff < 0 & shift(v=as.numeric(Fast.Diff), places=1, dir=“right”) > 0, EWP$EWP.Close, NA) plot(EWP) ## Warning in plot.xts(EWP): only the univariate series will be plotted points(Long_Trades, col=‘blue’, cex=1.5, pch=18) points(Short_Trades, col=‘red’, cex=1.5, pch=18)

Mixture of entry points and that is usually how it works on a trading, bouncing trend. Though we aren’t going to design full trending systems here, a stop-loss exit order is key to any directional trading so you don’t lose everything! Let’s see what it does on trending market:

EMA.Fast <- EMA(SPY\(SPY.Close, n=10, ) EMA.Medium <- EMA(SPY\)SPY.Close, n=50, ) EMA.Slow <- EMA(SPY$SPY.Close, n=200, ) Fast.Diff <- EMA.Fast - EMA.Medium Slow.Diff <- EMA.Medium - EMA.Slow

look for long entries

Long_Trades <- ifelse( Slow.Diff > 0 & Fast.Diff > 0 & shift(v=as.numeric(Fast.Diff), places=1, dir=“right”) < 0, SPY$SPY.Close, NA)

look for long exits (same thing but inverse signts)

Short_Trades <- ifelse( Slow.Diff < 0 & Fast.Diff < 0 & shift(v=as.numeric(Fast.Diff), places=1, dir=“right”) > 0, SPY$SPY.Close, NA) plot(SPY) ## Warning in plot.xts(SPY): only the univariate series will be plotted points(Long_Trades, col=‘blue’, cex=1.5, pch=18) points(Short_Trades, col=‘red’, cex=1.5, pch=18)