high_earnings_yield.Rmd

---
title: "High Earnings Yield Portfolio"
output: html_notebook
---

```{r}
# install.packages("dplyr")
# install.packages("ggplot2")
# install.packages("lubridate")
# install.packages("BatchGetSymbols")
# install.packages("remotes")
# remotes::install_github("Ljupch0/yfinance")
```




```{r}
library(yfinance)
library(quantmod)
library(lubridate)
library(ggplot2)
library(BatchGetSymbols)
library(tidyr)

`%nin%` <- Negate(`%in%`)

coalesce_by_column <- function(df) {
  return(dplyr::coalesce(!!! as.list(df)))
}

get_price <- function(ticker) {
  get_key_stats_proto <- function(ticker) {
    jsonlite::flatten(jsonlite::fromJSON(glue::glue("https://query2.finance.yahoo.com/v10/finance/quoteSummary/{ticker}?modules=price"))[[1]][[1]][[1]]) %>% 
      select(ends_with(".raw")) %>% 
      `names<-`(sub(".raw","", names(.))) %>% 
      mutate(
        ticker = ticker,
        date = Sys.Date()
      ) %>% 
      select(ticker, date, everything())
  }
  purrr::map_df(.x = ticker, ~ get_key_stats_proto(.x) )
}



```


```{r}
## Имиња на акциите во S&P500
sp500 <- BatchGetSymbols::GetSP500Stocks()

## Ги вадиме имиињата на акции со .B бидејќи не враќаат податоци од yfinance
sp500 <- sp500 %>% 
  dplyr::filter(!grepl("*\\.B$", sp500$Tickers))

## Ги симнуваме цените
# sp500_price <- get_price( sp500$Tickers)
# saveRDS(sp500_price, "../data/sp500_price.RDS")
sp500_price <- readRDS("../data/sp500_price.RDS")

## Од цените ни треба Market Capitalizaiton
sp500_market_cap <- sp500_price %>% 
  select(ticker, marketCap)


```

```{r}
## Ги симнуваме финансиските податоци за сите 500 акции. Треба време: 1500 API Повикувања

# sp500_data <- getFinancials(ticker = sp500$Tickers, report_type = "quarterly")
# saveRDS(sp500_data, "../data/sp500_data.RDS")

#sp500_data_yearly <- yfinance::getFinancials(ticker = sp500$Tickers, report_type = "yearly")
#sp500_data_yearly1 <- yfinance::getFinancials(ticker = sp500$Tickers[1:200], report_type = "yearly")
#sp500_data_yearly2 <- yfinance::getFinancials(ticker = sp500$Tickers[201:400], report_type = "yearly")
#sp500_data_yearly3 <- yfinance::getFinancials(ticker = sp500$Tickers[401:503], report_type = "yearly")
#sp500_data_yearly <- bind_rows(sp500_data_yearly1, sp500_data_yearly2, sp500_data_yearly3)
#saveRDS(sp500_data_yearly, "../data/sp500_data_yearly.RDS")

sp500_data_yearly <- readRDS("../data/sp500_data_yearly.RDS")

## Чистење - На пример за ABBV за дата 2020-03-31 има две ставки, каде втората има само податок за minorityInterest кој фали кај првата.
## Ги комбинираме ставките така што секој финансиски извештај е само една ставка
sp500_data <- readRDS("../data/sp500_data.RDS") %>% 
  group_by(ticker, date) %>% 
  summarize_all(coalesce_by_column)
```


```{r}
# Филтер: date ја форматираме како дата, ги редиме по група по дата, и го филтрираме последниот извештај
sp500_latest <- sp500_data %>% 
  dplyr::mutate(date = as.Date(date)) %>%
  group_by(ticker) %>% 
  arrange(date) %>% 
  dplyr::filter(date == last(date) )
```


# Cheap
```{r}
# Пресметки
earnings_yield <- sp500_latest %>% 
  full_join(sp500_market_cap) %>% 
  transmute(
    enterprise_value = marketCap + coalesce(longTermDebt, 0 ) - cash,
    ebitda = totalRevenue - costOfRevenue - sellingGeneralAdministrative - replace_na(researchDevelopment, 0),
    earnings_yield = ebitda / enterprise_value * 100
  ) %>% 
  arrange(desc(earnings_yield)) %>% 
  ungroup() %>% 
  mutate(
    earnings_yield_rank = 1:nrow(.)
  )
```


# Financial Quality: Piotroski F-Score












```{r}

penultimate <- function (vector) {
  nth(vector, -2)
}


piotroski <- sp500_data_yearly %>%
  mutate(
    date = as.Date(date)
  ) %>% 
  arrange(date) %>% 
  group_by(ticker) %>% 
  transmute(
    date = date,
    
    # Profitability
    # Return on Assets (1 point if it is positive in the current year, 0 otherwise);
    roa = netIncome / totalAssets,
    f1_roa_positive = if_else(last(roa) > 0, 1, 0),
    
    # Operating Cash Flow (1 point if it is positive in the current year, 0 otherwise);
    f2_cf_posive = if_else(last(totalCashFromOperatingActivities) > 0, 1, 0 ),
    
    # Change in Return of Assets (ROA) (1 point if ROA is higher in the current year compared to the previous one, 0 otherwise);
    f3_change_roa = if_else(last(roa) > penultimate(roa), 1, 0 ),
    
    # Accruals (1 point if Operating Cash Flow/Total Assets is higher than ROA in the current year, 0 otherwise);
    operating_total_assets = totalCashFromOperatingActivities / totalAssets,
    f4_accruals = ifelse( last(operating_total_assets) > last(roa), 1, 0 ),
    
    # Leverage, Liquidity and Source of Funds
    # Change in Leverage (long-term) ratio (1 point if the ratio is lower this year compared to the previous one, 0 otherwise);
    f5_leverage = if_else(replace_na(last(longTermDebt), 0 ) > replace_na( penultimate(longTermDebt), 0 ) , 1, 0),
    
    # Change in Current ratio (1 point if it is higher in the current year compared to the previous one, 0 otherwise);
    current_ratio = totalCurrentAssets / totalCurrentLiabilities,
    f6_current_change = if_else( last(current_ratio) > penultimate(current_ratio), 1, 0 ),
    
    # Change in the number of shares (1 point if no new shares were issued during the last year);
    f7_change_shares = if_else(is.na(issuanceOfStock), 1, 0),
    
    # Operating Efficiency
    # Change in Gross Margin (1 point if it is higher in the current year compared to the previous one, 0 otherwise);
    gross_margin = grossProfit / totalRevenue,
    f8_gross_margin = ifelse(last(gross_margin) > penultimate(gross_margin), 1, 0 ),
    
    # Change in Asset Turnover ratio (1 point if it is higher in the current year compared to the previous one, 0 otherwise);
    totalAssets = totalAssets,
    average_assets = ( totalAssets + lag(totalAssets) ) / 2,
    asset_turnover = totalRevenue / average_assets,
    f9_asset_turn = if_else(last(asset_turnover) > penultimate(asset_turnover), 1, 0)
    
  ) %>% 
  select(ticker, date, starts_with("f")) %>% 
  dplyr::filter(date == last(date)) %>%  
  mutate(
    f_score = f1_roa_positive + f2_cf_posive + f3_change_roa + f4_accruals + f5_leverage + f6_current_change + f7_change_shares + f8_gross_margin + f9_asset_turn
  ) %>% 
  arrange(f_score) %>% 
  ungroup() %>% 
  mutate(f_score_rank = rank(-f_score))
  

```


```{r}
rankings <- full_join(earnings_yield %>% select(ticker, earnings_yield_rank),
                      piotroski %>% select(ticker, f_score_rank )
                      ) %>% 
  mutate(
    average_rank = (earnings_yield_rank + f_score_rank) / 2,
  ) %>% 
  arrange(average_rank)
```

```{r}
linear_weight <- function(length) {
   len <- length
    avg <- 1/len
    min <- avg/2
    inc <- 1.5 * avg
    weights <- seq(from = min, by = inc, length.out = len)
    rescale <- 1 / sum(weights)
    return(rescaled_weights <- sort(weights*rescale, decreasing = TRUE) )
}


portfolio <- rankings[1:30,] %>% 
  mutate(
    equal_weight = 1/nrow(.),
    linear_weight = linear_weight(nrow(.))
  )
```


```{r}
get_asset_summaries <- function(ticker) {
  get_asset_summaries_proto <- function(ticker) {
    jsonlite::flatten(jsonlite::fromJSON(glue::glue("https://query2.finance.yahoo.com/v10/finance/quoteSummary/{ticker}?modules=summaryProfile"))[[1]][[1]][[1]]) %>% 
      mutate(
        ticker = ticker
      ) %>%
      select(ticker, everything())
  }
  purrr::map_df(.x = ticker, ~ get_asset_summaries_proto(.x) )
}
```


```{r}
asset_summaries <- get_asset_summaries(portfolio$ticker) 
  
full_portfolio <- full_join(portfolio, asset_summaries)
```