# Load packages
# Core
library(tidyverse)
library(tidyquant)
Visualize and compare skewness of your portfolio and its assets.
Choose your stocks.
from 2012-12-31 to 2017-12-31
symbols <- c("SPY", "WMT", "COST", "AMZN")
prices <- tq_get(x = symbols,
get = "stock.prices",
from = "2012-12-31",
to = "2017-12-31")
asset_returns_tbl <- prices %>%
# I added asset_returns_tbl here (wasn't in video) otherwise the code would not run, and mistake code wouldn't be fixable
# In video
group_by(symbol) %>%
tq_transmute(select = adjusted,
mutate_fun = periodReturn,
period = "monthly",
type = "log") %>%
slice(-1) %>%
#Remove the first row, but since data is group, it will remove the first line of each group
ungroup() %>%
set_names(c("asset", "date", "returns"))
# symbols
symbols <- asset_returns_tbl %>% distinct(asset) %>% pull()
symbols
## [1] "AMZN" "COST" "SPY" "WMT"
# weights
weights <- c(0.30, 0.30, 0.20, 0.20)
weights
## [1] 0.3 0.3 0.2 0.2
w_tbl <- tibble(symbols, weights)
w_tbl
## # A tibble: 4 × 2
## symbols weights
## <chr> <dbl>
## 1 AMZN 0.3
## 2 COST 0.3
## 3 SPY 0.2
## 4 WMT 0.2
# ?tq_portfolio
portfolio_returns_tbl <- asset_returns_tbl %>%
tq_portfolio(assets_col = asset,
returns_col = returns,
weights = w_tbl,
rebalance_on = "months",
col_rename = "returns")
portfolio_returns_tbl
## # A tibble: 60 × 2
## date returns
## <date> <dbl>
## 1 2013-01-31 0.0427
## 2 2013-02-28 0.00121
## 3 2013-03-28 0.0363
## 4 2013-04-30 0.00325
## 5 2013-05-31 0.0201
## 6 2013-06-28 0.00825
## 7 2013-07-31 0.0616
## 8 2013-08-30 -0.0526
## 9 2013-09-30 0.0497
## 10 2013-10-31 0.0694
## # ℹ 50 more rows
portfolio_skew_tidyquant_builtin_percent <- portfolio_returns_tbl %>%
tq_performance(Ra = returns,
performance_fun = table.Stats) %>%
select(Skewness)
portfolio_skew_tidyquant_builtin_percent
## # A tibble: 1 × 1
## Skewness
## <dbl>
## 1 -0.0778
#Calculate sd of portfolio returns
sd_portfolio <- sd(portfolio_returns_tbl$returns)
mean_portfolio <- mean(portfolio_returns_tbl$returns)
portfolio_returns_tbl %>%
# Add a new variable
mutate(extreme_neg = ifelse(returns < mean_portfolio - 2 * sd_portfolio,
"ext_neg",
"not_ext_neg")) %>%
ggplot(aes(x = returns, fill = extreme_neg)) +
geom_histogram(binwidth = 0.003) +
scale_x_continuous(breaks = seq(-0.06, 0.06, 0.02)) +
#order of data: far left value, far right value, intervals
labs(x = "monthly returns")
# Data transformation: calculate skewness
asset_skewness_tbl <- asset_returns_tbl %>%
group_by(asset) %>%
summarise(skew = skewness(returns)) %>%
ungroup() %>%
# Add portfolio skewness
add_row(tibble(asset = "Portfolio",
skew = skewness(portfolio_returns_tbl$returns)))
asset_skewness_tbl
## # A tibble: 5 × 2
## asset skew
## <chr> <dbl>
## 1 AMZN 0.187
## 2 COST -0.244
## 3 SPY -0.264
## 4 WMT 0.0723
## 5 Portfolio -0.0778
# Plot skewness
asset_skewness_tbl %>%
ggplot(aes(x = asset, y = skew, color = asset)) +
geom_point() +
ggrepel::geom_text_repel(aes(label = asset),
data = asset_skewness_tbl %>%
filter(asset == "Portfolio")) +
labs(y = "skewness")
# Transform data: calculate rolling skewness
rolling_skew_tbl <- portfolio_returns_tbl %>%
tq_mutate(Select = returns,
mutate_fun = rollapply,
width = 24,
FUN = skewness,
col_rename = "Skew") %>%
select(-returns) %>%
na.omit()
# Plot
rolling_skew_tbl %>%
ggplot(aes(x = date, y = Skew)) +
geom_line(color = "cornflowerblue") +
geom_hline(yintercept = 0, linetype = "dotted", size = 2) +
# Formatting
scale_y_continuous(limits = c(-1,1), breaks = seq(-1,1,0.2)) +
theme(plot.title = element_text(hjust = 0.5)) +
# Labelling
labs(y = "Skewness",
x = NULL,
title = "Rolling 24-Month Skewness") +
annotate(geom = "text",
x = as.Date("2016-07-01"), y = 0.8,
color = "red", size = 5,
label = str_glue("The 24-month skewness is positive for about half of the lifetime,
even though the overall skewness is negative"))
Is any asset in your portfolio more likely to return extreme positive returns than your portfolio collectively? Discuss in terms of skewness. You may also refer to the distribution of returns you plotted in Code along 4.
Yes, AMZN is more likely to return extreme positive returns than my portfolio collectively. As seen in this graph,it has however more risks too. In term of skewness, AMNZ has a positive value of ~0.17, compared to ~0.7 for WMT, and ~ -0.09 for my portfolio, thus making it the one with the highest positive skewness by far.
As I did not have the same stocks in Apply 4, I’m putting a picture of what the graph would have been below:
out.width = "20px"
echo = FALSE
fig.cap = "Apply 4 img.png"
knitr::include_graphics("Apply 4 img.png")
Based on this distribution of returns, AMZN tends to have more frequent large gains, as well as large losses, while WMT seems to have more often moderate gains. The larger right tail in the graph for AMZN also indicates larger and more frequent gains, thus more frequently positive outcomes. This is consistent with my earlier response, confirming that AMZN is the most likely to return extreme positive returns than the portfolio or any other assets in the portfolio.