---
title: "Social Media Portfolio Performance"
output: 
  flexdashboard::flex_dashboard:
    orientation: columns
    vertical_layout: fill
    source_code: embed
---

```{r setup, include=FALSE}
library(flexdashboard)
library(flexdashboard)
library(shiny)
library(scales)
library(tidyverse)
library(highcharter)
library(htmltools)
library(tidyquant)
library(shinydashboard)
library(data.table)
library(gt)
library(plotly)
library(ggthemes)
```

```{r echo=FALSE}
rm(list=ls())
raw_data <- read_csv("C:/Users/alber/OneDrive/Documents/raw data.csv")%>%
  select(Symbol,`Market Cap`, `Company Name` )
#removig extra string from the Company name
raw_data$`Company Name`<- raw_data$`Company Name`%>%
  gsub(pattern=", Inc.", replacement = "")
```

```{r data}

tickers <- raw_data$Symbol
start_date <- "2019-06-01"
df <- tq_get(x=tickers, get="stock.prices",from=start_date)
monthly_returns <-df%>%
  group_by(symbol)%>%
  tq_transmute(select=close, mutate_fun=periodReturn, col_rename = "monthly.returns", period="monthly")
weights_stocks =raw_data$`Market Cap`/sum(raw_data$`Market Cap`)
#-----------------------------------------------------------
portfolio_monthly_returns <- monthly_returns%>%
  tq_portfolio(
    assets_col=symbol,
    returns_col = monthly.returns,
    weights=weights_stocks,
    col_rename="investment.growth",
    wealth.index=TRUE
  )%>%
    mutate(investment.growth = investment.growth * 1000000)# initial investment is $1000000

# Now import benchmark 
mkt <- tq_get(x="^GSPC", from=start_date)%>%
  group_by(symbol)%>%
  tq_transmute(select=close, mutate_fun=periodReturn, col_rename = "mkt.monthly.returns", period="monthly")%>%
  tq_portfolio(
    assets_col=symbol,
    returns_col = mkt.monthly.returns,
    weights=1,
    col_rename="investment.growth",
    wealth.index=TRUE
  )%>%
    mutate(investment.growth = investment.growth * 1000000)
```

Column {data-width=550 .tabset .tabset-fade}
------------------------------------------------------
### Chart of Market Value - Weighted Portfolio

```{r}
portfolio_monthly_returns$date <- as.Date(portfolio_monthly_returns$date)
portfolio_monthly_returns$investment.growth <-round(
  portfolio_monthly_returns$investment.growth,digits=2
)
#-------------------------------------------------------------------

df <- left_join(portfolio_monthly_returns, mkt, by="date")
names <- c("date","port","mkt")
colnames(df) <- names
#rounding the numbers in the market column
df$mkt <- round(df$mkt, digits=2)
df <- data.table(df)
df <-melt(df, id="date")
df%>%
  data.frame()%>%
  hchart(type="line",mapping=hcaes(x=date,y=value, group=variable, color=variable))%>%
  hc_add_theme(hc_theme_darkunica())%>%
  hc_yAxis(text="Investment Growth")%>%
  hc_tooltip(pointFormat="<b> Portfolio Value:</b> ${point.y} <br>")%>%
  hc_plotOptions(
    series = list(animation = list(duration = 5000))
    )%>%
  hc_title(text="Portfolio Value: Market vs Stock portfolios")%>%
  hc_subtitle(text=paste("Date starts from",start_date))%>%
  hc_yAxis(title = list(text = "Portfolio Growth"),
           opposite = TRUE,
           minorTickInterval = "auto",
           minorGridLineDashStyle = "LongDashDotDot",
           showFirstLabel = TRUE,
           showLastLabel = TRUE)%>%
  hc_rangeSelector(
    enabled = TRUE
  )%>%
  hc_navigator(
    enabled = TRUE
  )%>%
  hc_xAxis(
    plotLines = list(
      list(
        value = ymd(20220224) %>% datetime_to_timestamp(),
        color = "#FF0000",
        label = list(
          text = "Energy Crisis",
          align = "top",
          style = list(color = "#FF0000")
        )
      )
    )
  )%>%
  hc_exporting(enabled = TRUE)
```
### Assets
```{r}
data.frame(
Stocks=raw_data$Symbol, 
"Company Name" =raw_data$`Company Name`)%>%
  gt()%>%
  tab_header(title=c("Portfolio assets"
                        ))
  
```
### Weighting
```{r}
raw_data %>%
  mutate(weights=round(weights_stocks*100, digits=2))%>%
  hchart(type="bar", hcaes(x=`Company Name`,y=weights))%>%
  hc_add_theme(hc_theme_economist())%>%
  hc_colors(colors=viridis::plasma(12))%>%
  hc_tooltip(pointFormat="<b> Weight:</b> {point.y}% <br>")%>%
  hc_title(text="Proportion of stocks within the portfolio")%>%
  hc_subtitle(text="The proportions are based on the respective Market Capitalisation")%>%
  hc_plotOptions(
    series = list(animation = list(duration = 5000))
    )
```
### META
```{r META}
tq_get(x="META", from=start_date)%>%
  hchart(type="line",mapping=hcaes(x=date,y=round(close, digits=2)))%>%
  hc_add_theme(hc_theme_darkunica())%>%
  hc_yAxis(text="Stock Price")%>%
  hc_tooltip(pointFormat="<b> Stock Price:</b> ${point.y} <br>")%>%
  hc_plotOptions(
    series = list(animation = list(duration = 5000))
    )%>%
  hc_title(text="Stock Price: META")%>%
  hc_subtitle(text=paste("Date starts from",start_date))%>%
  hc_yAxis(title = list(text = "Portfolio Growth"),
           opposite = TRUE,
           minorTickInterval = "auto",
           minorGridLineDashStyle = "LongDashDotDot",
           showFirstLabel = TRUE,
           showLastLabel = TRUE)
```

```{r}
# risk free rate
rf <- tq_get("^TYX", from=start_date)%>%
  tq_transmute(select=adjusted,
               mutate_fun=periodReturn, 
               col_rename ="return")
  
rf <-rf%>%
    as.data.frame()%>%
  select(return)%>%
  zoo()%>%
  as.numeric()%>%
  median(na.rm=TRUE)*12
# beta 
port <- filter(df,variable=="port")%>%
  rename(port="variable",
         port_value="value")
mkt <- filter(df, variable=="mkt")%>%
  rename(mkt="variable", 
         mkt_value="value")
data <- data.frame(port, mkt)%>%
  mutate(port_return=log(port_value) - log(lag(port_value)),
         mkt_return=log(mkt_value) - log(lag(mkt_value))
         )
# average return on the portfolio
avg_port_return <- data%>%
  select(port_return)%>%
  zoo()%>%
  as.numeric()%>%
  median(., na.rm=TRUE)*12
# average return on the market 
avg_mkt_return<- data%>%
  select(mkt_return)%>%
  zoo()%>%
  as.numeric()%>%
  median(., na.rm=TRUE)*12
# Regression model
mdl <- lm(data=data, formula=port_return ~ mkt_return)
# beta of portfolio
beta <- mdl$coefficients[2]
# Jensen's alpha 
alpha <- avg_port_return - (rf + beta*(avg_mkt_return - rf))

a <- if (alpha < 0) {
  print("1. The portfolio performs poorly relative to its benchmark (S&P500)")
} else if (alpha == 0 ){
 print("1. The portfolio and the market have equally performed")
} else {
  print("1. Portfolio outperformed the market")
}

#-----------------------------------------------------------------
#Now to calculate the information ratio 
#tracking error 
tracking_error_sd <- data%>%
  transmute( tracking_error= port_return - mkt_return)%>%
  zoo()%>%
  sd(na.rm=TRUE)*sqrt(12)
info_ratio <- (avg_port_return - avg_mkt_return)/ tracking_error_sd

b<- if (info_ratio < 0) {
  print("2. The portfolio returns are more volatile over time and not as consistent in exceeding the benchmark.")
} else if (info_ratio == 0 ){
 print("2. The portfolio are as volatile as the market")
} else {
  print("2. Portfolio is more likely to beat the market in the future")
}
#---------------------------------------------------------------------
#Sharpe Ratio metric 
port_risk <- data%>%
  select(port_return)%>%
  zoo()%>%
  sd(na.rm=TRUE)*sqrt(12)
Sharpe_Ratio <-  (avg_port_return -rf)/port_risk
c<- if (Sharpe_Ratio < 1) {
  print("3. Portfolio offers poor excess returns per unit of risk")
} else if (Sharpe_Ratio == 1 ){
 print("3.The portfolio gains no excess returns")
} else {
  print("3.  The portfolio yields greater excess returns per unit of risk")
}
d <- data.frame("Jensens alpha" = alpha,
           "Information Ratio"=info_ratio, 
           "Sharpe Ratio"=Sharpe_Ratio
                            )%>%
  round(digits=2)
rownames(d)<- ""
```

### Performance metrics
```{r}
#------------------------------------------------------------------
#Insert performance metrics' values in a table
d%>%
  gt()%>%
 tab_header(
   title=md("Portfolio's performance"),
   subtitle= paste("from ", start_date,"to ", Sys.Date())
 )%>%
  tab_source_note(source_note=data.frame(a,b,c))
```


Column {data-width=450 .tabset .tabset-fade}
---------------------------------------------
### Distribution

```{r}
library(ggplot2)
p <- portfolio_monthly_returns%>%
  mutate(investment.returns=log(investment.growth)-log(lag(investment.growth)))%>%
  ggplot(., mapping=aes(x=investment.returns))+ geom_histogram(bins=10,col="#FF7518", fill="#D23B05")+ scale_x_continuous(labels=scales::percent_format())+ labs(x="Investment Returns", title="Distribution of Portfolio Returns")+theme_bw()
ggplotly(p)
```
### Regression
```{r regression}
#regression analysis 
port <- filter(df,variable=="port")%>%
  rename(port="variable",
         port_value="value")
mkt <- filter(df, variable=="mkt")%>%
  rename(mkt="variable", 
         mkt_value="value")
data <- data.frame(port, mkt)%>%
  mutate(port_return=log(port_value) - log(lag(port_value)),
         mkt_return=log(mkt_value) - log(lag(mkt_value))
         )
data%>%
  ggplot(mapping=aes(x=mkt_return,
                     y=port_return)) + geom_smooth(method="lm", se=FALSE, 
                                                   col="#051094") + geom_point(col="#8c4626") + scale_y_continuous(labels = scales::percent_format())+
  scale_x_continuous(labels=scales::percent_format())+
  labs(
    x="Market Returns", 
    y="Portolio Returns",
    title="Capital Asset Pricing Model", 
    subtitle="Regressing portfolio against market"
  )+ theme_bw()
```

### Forecast

```{r}
# Initial price
S0 <- portfolio_monthly_returns$investment.growth %>%
  tail(n=1)
#risk, standard deviation
sigma <- portfolio_monthly_returns%>%
  mutate(investment.returns=log(investment.growth) - log(lag(investment.growth)))%>%
  select(investment.returns)%>%
  na.omit()%>%
  unlist()%>%
  sd()
sigma <- sigma *sqrt(12)
# Average value
r <- portfolio_monthly_returns%>%
  mutate(investment.returns=log(investment.growth) - log(lag(investment.growth)))%>%
  select(investment.returns)%>%
  na.omit()%>%
  unlist()
# median or annualised trend value.
mu <- median(r)*12
# number of sim
nsim <-1000
nrow <-12 # months
#last date 
last_date <- portfolio_monthly_returns%>%
  tail(n=1)%>%
  select(date)%>%
  as.numeric()%>%
  as.Date(format="%Y-%m-%d")

# GBM 
gbm <- matrix(nrow=nrow, ncol=nsim)
# loop
for (simu in 1:nsim) {
    gbm[1, simu] <- S0
    for (day in 2:nrow) {
      epsilon <- rnorm(1)
      dt = 1 / 365
      gbm[day, simu] <- gbm[(day-1), simu] * exp((mu - sigma * sigma / 2) * dt + sigma * epsilon * sqrt(dt))
    }
}
gbm <- gbm %>%
  data.frame()
# add a column of future dates
gbm$dates <- seq(last_date, by="month", length.out=nrow)%>%
  data.frame()
l<- melt(data.table(gbm), id="dates")
p <- l%>%
  ggplot(., mapping=aes(x=dates, y=value, group=variable, color=variable)) +geom_line() + scale_y_continuous(labels=scales::dollar_format()) + labs(x="Dates",y="Investment Value" ,title="Forecasting Portfolio Value", subtitle="Prediting investment growth from today")+ theme_igray()+
  theme(legend.position = 'none')

ggplotly(p)
```