Summary
An algorithmic trading strategy is created in this notebook for Tesla using the simple moving average strategy.
A “buy” or “sell” signal/decision will be generated and executed/traded by the algorithm.
Simple moving averages are calculated by taking the arithmetic mean of a stock price for the desired past n days. Here:
- Long-term moving average (SMA100) is based on the most recent 100 days
- Short-term moving average (SMA 50) is based on the most recent 50 days
Tesla’s stock price from Yahoo Finance from 2017 Jan to 2022 May are used.
The outcome of this algo strategy yields the following results:
- Higher expected annualized return(82%) than actual returns (70%)
- Same volatility of return (62.4%)
- Higher Sharpe ratio (1.28 vs 1.09)
Load Packages & Extract Data from Yahoo Finance
library(quantmod)
library(tidyverse)
library(TTR)
library(kableExtra)
library(PerformanceAnalytics)
library(reshape)
#Obtaining stock price data ONLY column 6. xts = extensible time series
TSLA <- getSymbols("TSLA", source="yahoo", auto.assign=FALSE,
return.class="xts")[,6]
#Calculate actual log returns
Actual_Ret <- diff(log(TSLA))
colnames(Actual_Ret) <- "TSLA"
#Choose start date & view price trend
TSLA <- round(TSLA["2017/"],2)
Actual_Ret <- round(Actual_Ret["2017/"],4)Visualize Actual Stock Price
plot(TSLA, main="TSLA Actual Stock Price (2017Jan - 2022May)")
Visualize Actual Daily Returns
The actual daily returns (difference in stock price from previous close) plotted below shows an average of 0%. This is a common expectation for the daily returns of a stock which are different from the annualized returns due to the compounding effect of the daily returns.
#Plot the returns
plot(Actual_Ret)
Set up
#Generate Simple Moving Averages (long-term & short-term)
SMA100 <- round(SMA(TSLA, 100),2) #Long-term moving average
SMA50 <- round(SMA(TSLA, 50),2) #Short-term moving average
#Put desired columns into a dataframe. "na.omit" to drop rows with n/a. "cbind" to combine desired columns.
Data <- na.omit(as.data.frame(cbind(TSLA, Actual_Ret, SMA50, SMA100)))
colnames(Data) <- c("TSLA_Price","Actual_Ret","SMA50","SMA100" )Create a Trading Strategy & Backtest
#Condition for trend-following strategy (SMA)."UD" = price moving up or down
Data$UD <- ifelse(Data$SMA50 >= Data$SMA100, 1, 0)
#Generate "buy" (1) signal if short-term moving average > long-term moving average, else signal "Sell" (-1)
Data$Trade <- ifelse(Data$UD == 1, "BUY", "SELL")
#If "Position" is "Buy" show 1, else -1 (Sell)
Data$Position <- ifelse(Data$Trade == "BUY", 1, -1)
#Calculate returns based on this algorithm
Data$AlgoReturns_Daily <- round((Data$Actual_Ret * Data$Position),4)
#Annualized the Algo Returns
AnnualizedReturn <- ((mean(Data$AlgoReturns_Daily)+1)^252 - 1)
#Calculate the standard deviation for Sharpe ratio. Assume a risk-free rate (rf)
Standev <-StdDev.annualized(Data$AlgoReturns_Daily, scale = 252)
rf <- 0.02
SharpeRatio <- (AnnualizedReturn - rf)/Standev
#Actual Annualized Returns
AnnualizedReturn_Actual <- ((mean(Data$Actual_Ret)+1)^252 - 1)
#For the Actual Annualized Returns, calculate the standard deviation for Sharpe ratio. Assume a risk-free rate (rf)
Standev_Actual <- StdDev.annualized(Data$Actual_Ret, scale = 252)
rf <- 0.02
SharpeRatio_Actual <- (AnnualizedReturn_Actual - rf)/Standev_ActualVisualize Short- & Long-term Moving Averages against Actual Stock Price
#Visualize Price vs Moving Averages
plot(Data$TSLA_Price, type = "l", col = "black", ylab = "", xlab = "", xaxt='n', cex.axis=1)
title(ylab = "Price", line=2.5, cex.lab=1.2)
title(xlab = 'Time', line=1.0, cex.lab=1.2)
title(xlab = '(2017Jan to 2022May)', line=2.3, cex.lab=0.8)
par(new = TRUE)
plot(Data$SMA50, type = "l" , col = "cyan", ylab = "", xlab = "", yaxt="n",xaxt='n')
par(new = TRUE)
plot(Data$SMA100, type = "l" , col = "blue", ylab = "", xlab = "",yaxt="n",xaxt='n')
legend("topleft",legend=c("Actual Price","SMA 50","SMA100"),col=c("black", "cyan","blue"),lty = 1, cex=0.8)
Obtain Algo Output
The table below displays a subset of the complete table of the algo’s output.
Essentially, based on the conditions set up above, either a “buy” or “sell” signal to trade/execute is generated (under “Trade” column):
- Buy: If SMA50 > SMA100 when SMA50 crosses SMA100 from below (stock price is anticipated by Algo to go up)
- Sell: If SMA50 < SMA100 when SMA50 crosses SMA100 from above (stock price is anticipated by Algo to go down)
If the “Actual_Ret” column is negative, and Algo signals a “Sell” (in the “Trade” column), then the “AlgoReturns_Daily” column registers a positive gain since the Algo correctly generated a decision on a falling stock price.
Conversely, If “Actual_Ret” is positive, but a “Sell” signal is generated, that means the Algo incorrectly predicted the direction of stock movement and the corresponding AlgoReturns_Daily for that trading day would then show a negative return.
#Sort results by latest date first
Data_latest_1st <-apply(Data, 2, rev) #1st calculate reverse (rows =2)
Data_latest_1st <-as.data.frame(Data_latest_1st) #Then, convert back into dataframe
# Show results for every 30 trading days
Data_sample = Data_latest_1st %>% slice(which(row_number() %% 30 == 1))
Data_sample %>%
kbl() %>%
kable_paper("hover", full_width = F)| TSLA_Price | Actual_Ret | SMA50 | SMA100 | UD | Trade | Position | AlgoReturns_Daily | |
|---|---|---|---|---|---|---|---|---|
| 2022-05-06 | 865.65 | -0.0088 | 943.25 | 954.88 | 0 | SELL | -1 | 0.0088 |
| 2022-03-24 | 1013.92 | 0.0147 | 898.15 | 985.09 | 0 | SELL | -1 | -0.0147 |
| 2022-02-09 | 932.00 | 0.0108 | 1003.53 | 978.86 | 1 | BUY | 1 | 0.0108 |
| 2021-12-28 | 1088.47 | -0.0050 | 1048.24 | 896.85 | 1 | BUY | 1 | -0.0050 |
| 2021-11-12 | 1033.42 | -0.0287 | 884.19 | 784.34 | 1 | BUY | 1 | -0.0287 |
| 2021-10-01 | 775.22 | -0.0003 | 721.62 | 674.81 | 1 | BUY | 1 | -0.0003 |
| 2021-08-19 | 673.47 | -0.0228 | 665.10 | 660.04 | 1 | BUY | 1 | -0.0228 |
| 2021-07-08 | 652.81 | 0.0126 | 630.44 | 661.26 | 0 | SELL | -1 | -0.0126 |
| 2021-05-25 | 604.69 | -0.0029 | 663.70 | 720.12 | 0 | SELL | -1 | 0.0029 |
| 2021-04-13 | 762.32 | 0.0825 | 714.76 | 702.61 | 1 | BUY | 1 | 0.0825 |
| 2021-03-01 | 718.43 | 0.0616 | 777.16 | 630.32 | 1 | BUY | 1 | 0.0616 |
| 2021-01-14 | 845.00 | -0.0111 | 604.85 | 513.52 | 1 | BUY | 1 | -0.0111 |
| 2020-12-01 | 584.76 | 0.0298 | 443.29 | 400.86 | 1 | BUY | 1 | 0.0298 |
| 2020-10-19 | 430.83 | -0.0203 | 408.45 | 327.34 | 1 | BUY | 1 | -0.0203 |
| 2020-09-04 | 418.32 | 0.0274 | 326.87 | 248.20 | 1 | BUY | 1 | 0.0274 |
| 2020-07-24 | 283.40 | -0.0656 | 219.41 | 173.25 | 1 | BUY | 1 | -0.0656 |
| 2020-06-11 | 194.57 | -0.0523 | 152.90 | 141.86 | 1 | BUY | 1 | -0.0523 |
| 2020-04-29 | 160.10 | 0.0400 | 127.72 | 117.27 | 1 | BUY | 1 | 0.0400 |
| 2020-03-17 | 86.04 | -0.0340 | 131.67 | 101.29 | 1 | BUY | 1 | -0.0340 |
| 2020-02-03 | 156.00 | 0.1814 | 88.72 | 72.16 | 1 | BUY | 1 | 0.1814 |
| 2019-12-18 | 78.63 | 0.0367 | 64.18 | 55.31 | 1 | BUY | 1 | 0.0367 |
| 2019-11-05 | 63.44 | -0.0008 | 50.79 | 48.66 | 1 | BUY | 1 | -0.0008 |
| 2019-09-24 | 44.64 | -0.0776 | 46.88 | 45.41 | 1 | BUY | 1 | -0.0776 |
| 2019-08-12 | 45.80 | -0.0259 | 45.88 | 47.61 | 0 | SELL | -1 | 0.0259 |
| 2019-06-28 | 44.69 | 0.0028 | 44.52 | 50.88 | 0 | SELL | -1 | -0.0028 |
| 2019-05-16 | 45.67 | -0.0157 | 52.72 | 57.55 | 0 | SELL | -1 | 0.0157 |
| 2019-04-03 | 58.36 | 0.0205 | 58.53 | 63.27 | 0 | SELL | -1 | -0.0205 |
| 2019-02-20 | 60.51 | -0.0101 | 64.70 | 63.85 | 1 | BUY | 1 | -0.0101 |
| 2019-01-07 | 66.99 | 0.0529 | 67.89 | 63.13 | 1 | BUY | 1 | 0.0529 |
| 2018-11-20 | 69.50 | -0.0171 | 60.92 | 62.16 | 0 | SELL | -1 | 0.0171 |
| 2018-10-09 | 52.56 | 0.0477 | 61.57 | 62.45 | 0 | SELL | -1 | -0.0477 |
| 2018-08-27 | 63.85 | -0.0111 | 65.71 | 62.69 | 1 | BUY | 1 | -0.0111 |
| 2018-07-16 | 62.02 | -0.0279 | 62.83 | 62.13 | 1 | BUY | 1 | -0.0279 |
| 2018-06-01 | 58.36 | 0.0246 | 57.89 | 62.55 | 0 | SELL | -1 | -0.0246 |
| 2018-04-19 | 60.02 | 0.0227 | 62.92 | 64.45 | 0 | SELL | -1 | -0.0227 |
| 2018-03-07 | 66.46 | 0.0124 | 66.92 | 65.79 | 1 | BUY | 1 | 0.0124 |
| 2018-01-23 | 70.56 | 0.0035 | 64.66 | 66.91 | 0 | SELL | -1 | -0.0035 |
| 2017-12-07 | 62.25 | -0.0065 | 65.27 | 67.73 | 0 | SELL | -1 | 0.0065 |
| 2017-10-25 | 65.17 | -0.0347 | 70.56 | 70.22 | 1 | BUY | 1 | -0.0347 |
| 2017-09-13 | 73.25 | 0.0095 | 68.52 | 68.10 | 1 | BUY | 1 | 0.0095 |
| 2017-08-01 | 63.91 | -0.0121 | 68.74 | 63.78 | 1 | BUY | 1 | -0.0121 |
| 2017-06-19 | 73.96 | -0.0043 | 65.03 | 58.82 | 1 | BUY | 1 | -0.0043 |
Compare Returns & Sharpe Ratio by Algo against Actual
#Print the results
cat(paste("Expected Annualized Return by this Algo: ", round(AnnualizedReturn*100, digits=2),"%\n",
" Standard deviation of Algo Return: ", round(Standev*100, digits=2),"%\n",
" Sharpe ratio of this Algo: ", round(SharpeRatio, digits=2),"(for a risk-free rate of", round((rf)*100,digits=2),"%)\n\n\n",
" Actual Annualized Return: ", round(AnnualizedReturn_Actual*100, digits=2),"%\n"," Standard deviation of Actual Return: ", round(Standev_Actual*100, digits=2),"%\n"," Sharpe ratio of Actual Return: ", round(SharpeRatio_Actual, digits=2),"(for a risk-free rate of", round((rf)*100,digits=2),"%)"))## Expected Annualized Return by this Algo: 82.1 %
## Standard deviation of Algo Return: 62.38 %
## Sharpe ratio of this Algo: 1.28 (for a risk-free rate of 2 %)
##
##
## Actual Annualized Return: 70.26 %
## Standard deviation of Actual Return: 62.4 %
## Sharpe ratio of Actual Return: 1.09 (for a risk-free rate of 2 %)