Goal

The goal is to evaluate an investment strategy suggested by an investment advisor in February 2016. The strategy plans to invest in 20 Swiss stocks and ETFs with a focus on dividend payout (dividend strategy). The constituents are equal weighted (5%) and rule based rebalancing takes place monthly. Dividends are not reinvested.

symbolsEUR <- c("MUV2.DE",
                "ALV.DE",
                "BAS.DE",
                "C029.F") # SPI ETF

symbols_namesEUR <- c("Muenchner Rueck",
                      "Allianz",
                      "BASF",
                      "SPI ETF")

symbolsCHF <- c("ROG.SW",
                "NESN.SW",
                "SGSN.SW",
                "SCMN.SW",
                "ZURN.SW",
                "CFR.SW",
                "BCVN.SW",
                "WARN.SW",
                "SPSN.SW",
                "CLN.SW",
                "BAER.SW",
                "ORON.SW",
                "CMBN.SW",
                "0DZH.L") # SMI ETF

symbols_namesCHF <- c("Roche",
                      "Nestle",
                      "SGS",
                      "Swisscom",
                      "Zurich Insur",
                      "Richemont",
                      "Banque Cant Vaud",
                      "Warteck Invest",
                      "Swiss Prime Site",
                      "Clariant",
                      "Julius Baer",
                      "Orior",
                      "Cembra",
                      "SMI ETF")

pricesCHF <- tq_get(symbolsCHF, from = "2016-01-01") %>% mutate(close_chf = close)
pricesEUR <- tq_get(symbolsEUR, from = "2016-01-01")

dividendsCHF <- tq_get(symbolsCHF, from = "2016-01-01", get = "dividends") %>% mutate(dividends_chf = dividends)
dividendsEUR <- tq_get(symbolsEUR, from = "2016-01-01", get = "dividends")

# VALOR: 1065278 / ISIN: IE0008471009
# https://www.finanzen.ch/etf/historisch/ishares-ii-plc---ishares-core-euro-stoxx-50-ucits-etf-eur-dist/swx/1.1.2015_6.6.2019
pricesfinanzenCHF <- read_csv("data/iShares_EURO_STOXX_50_CHF.csv") %>% 
  mutate(symbol = "iSharesSTOXXDiv", date = dmy(Datum)) %>% 
  select(symbol, date, 
         open = Eröffnung, high = Tageshoch, low = Tagestief, close = Tagestief, 
         volume = `Umsatz (St.)`) %>% 
  filter(date >= "2016-01-01") %>% 
  mutate(close_chf = close)

dividendsfinanzenCHF <- tibble(symbol = "iSharesSTOXXDiv",
                               date = date(c("2016-12-30", "2017-12-29", "2018-12-31", "2019-05-31")),
                               dividends = c(1.23, 1.14, 1.29, 0.49)) %>% 
  mutate(dividends_chf = dividends)

# VALOR: 42746037 / ISIN: LU1812092168
# https://www.finanzen.ch/etf/historisch/lyxor-stoxx-europe-select-dividend-30-ucits-etf---dist/fse/1.1.2015_6.6.2019
pricesfinanzenEUR <- read_csv("data/Lyxor_STOXX_Europe_Select_Dividend_EUR.csv") %>% 
  mutate(symbol = "LyxorSTOXXSelDiv", date = dmy(Datum)) %>% 
  select(symbol, date, 
         open = Eröffnung, high = Tageshoch, low = Tagestief, close = Tagestief, 
         volume = `Umsatz (St.)`) %>% 
  filter(date >= "2016-01-01")

dividendsfinanzenEUR <- tibble(symbol = "LyxorSTOXXSelDiv",
                               date = date(c("2016-12-30", "2017-12-29", "2018-12-31", "2019-05-31")),
                               dividends = c(0.83, 0.76, 0.81, 0))

eurchf <- Quandl("ECB/EURCHF", start_date="2016-01-01")

pricesEUR_CHF <- pricesEUR %>% 
  left_join(eurchf %>% rename(eurchf = Value), by = c("date" = "Date")) %>% 
  mutate(close_chf = close*eurchf)

pricesfinanzenEUR_CHF <- pricesfinanzenEUR %>% 
  left_join(eurchf %>% rename(eurchf = Value), by = c("date" = "Date")) %>% 
  mutate(close_chf = close*eurchf)

dividendsEUR_CHF <- dividendsEUR %>% 
  left_join(eurchf %>% rename(eurchf = Value), by = c("date" = "Date")) %>% 
  mutate(dividends_chf = dividends*eurchf)

dividendsfinanzenEUR_CHF <- dividendsfinanzenEUR %>% 
  left_join(eurchf %>% rename(eurchf = Value), by = c("date" = "Date")) %>% 
  mutate(dividends_chf = dividends*eurchf)

# putting it all togethet
prices <- bind_rows(pricesCHF, pricesEUR_CHF, pricesfinanzenCHF, pricesfinanzenEUR_CHF) %>% 
  arrange(symbol, date) %>% 
  group_by(symbol) %>%
  mutate(close_chf = na.locf(close_chf)) %>% 
  ungroup()

dividends <- bind_rows(dividendsCHF, dividendsEUR_CHF, dividendsfinanzenCHF, dividendsfinanzenEUR_CHF)

names_lut <- tibble(symbol = symbolsEUR, names = symbols_namesEUR) %>% 
  bind_rows(tibble(symbol = symbolsCHF, names = symbols_namesCHF)) %>% 
  bind_rows(tibble(symbol = "LyxorSTOXXSelDiv", names = "STOXX SelDiv ETF")) %>% 
  bind_rows(tibble(symbol = "iSharesSTOXXDiv", names = "STOXX Div ETF"))

prices %>% saveRDS("data/prices.RDS")
dividends %>% saveRDS("data/dividends.RDS")
names_lut %>% saveRDS("data/names_lut.RDS")

Performance analysis

prices <- readRDS("data/prices.RDS")
dividends <- readRDS("data/dividends.RDS")
names_lut <- readRDS("data/names_lut.RDS")

Check that we have 20 titles and all dividends. Actually we only need monthly (end-of-month) returns.

# # how many stocks do we have?
# prices %>% group_by(symbol) %>% count() %>% ungroup() %>% count()

# # did all pay dividends?
# dividends %>% group_by(symbol) %>% count() %>% ungroup() %>% count()

# #which is the one that did not pay?
# prices %>% select(symbol) %>% anti_join(dividends) %>% unique()
# # 0DZH.L is xtrackers SMI

price_returns <- prices %>% 
  group_by(symbol) %>%
  tq_transmute(select     = close_chf, 
               mutate_fun = periodReturn, 
               period     = "monthly", 
               col_rename = "Ra") %>%
  mutate(date = ceiling_date(date, "month")-1) %>% 
  ungroup()

price_returns %>% 
  left_join(names_lut, by = "symbol") %>% 
  ggplot(aes(y = Ra, x = reorder(names, desc(names)), color = names)) +
  geom_hline(yintercept = 0, color = "red") +
  geom_boxplot(color = "black", alpha = 0.5) +
  geom_point() +
  scale_y_continuous(labels = scales::percent) +
  coord_flip() +
  labs(title = "Price returns of selected stocks/ETFs", subtitle = "Jan 2016 - June 2019, monthly returns, arithmetic, excl. dividends",
       x = "Stock", y = "return",
       caption = "Data: Yahoo Finance, graph: mgei.github.io") +
  theme_tq() +
  theme(legend.position = "none")

weights = rep(1/20, 20)

portfolio_returns <- price_returns %>% filter(date > date("2016-03-01")) %>% 
  tq_portfolio(assets_col  = symbol, 
               returns_col = Ra, 
               weights     = weights, 
               col_rename  = "Ra",
               rebalance_on = "months")

portfolio_returns %>% mutate(symbol = "Portfolio") %>% 
  ggplot(aes(y = Ra, x = reorder(symbol, desc(symbol)), color = symbol)) +
  geom_hline(yintercept = 0, color = "red") +
  geom_boxplot(color = "black", alpha = 0.5) +
  geom_point() +
  scale_y_continuous(labels = scales::percent) +
  coord_flip() +
  labs(title = "Price returns on equal weightes stocks/ETFs portfolio", subtitle = "Jan 2016 - June 2019, monthly returns, arithmetic, excl. dividends",
       x = "", y = "return",
       caption = "Data: Yahoo Finance, graph: mgei.github.io") +
  theme_tq() +
  theme(legend.position = "none")

# SMI
url <- "https://www.six-group.com/exchanges/downloads/indexdata/hsmi.csv"
GET(url, write_disk("data/temp.csv", overwrite = T))
data <- read_csv2("data/temp.csv", skip = 2)

smi <- data[-1:-4,] %>% 
  rename(Date = "SYMBOL") %>% 
  mutate(Date = as.Date(Date, "%d.%m.%Y")) %>% 
  mutate_if(is.character, funs(as.numeric(.))) %>% 
  select(date = Date, SMI)

# SPI
url <- "https://www.six-group.com/exchanges/downloads/indexdata/hspipr.csv"
GET(url, write_disk("data/temp.csv", overwrite = T))
data <- read_csv2("data/temp.csv", skip = 2) 

spi <- data[-1:-4,] %>% 
  rename(Date = "SYMBOL") %>% 
  mutate(Date = as.Date(Date, "%d.%m.%Y")) %>% 
  mutate_if(is.character, funs(as.numeric(.))) %>% 
  select(date = Date, SPI = SPIX)

benchmark <- bind_rows(smi %>% mutate(symbol = "SMI") %>% select(symbol, date, close_chf = SMI),
                       spi %>% mutate(symbol = "SPI") %>% select(symbol, date, close_chf = SPI))

benchmark %>% saveRDS("data/benchmark.RDS")

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

benchmark_returns <- benchmark %>% 
  filter(year(date) >= 2016) %>% 
  group_by(symbol) %>%
  tq_transmute(select     = close_chf, 
               mutate_fun = periodReturn, 
               period     = "monthly", 
               col_rename = "Ra") %>%
  mutate(date = ceiling_date(date, "month")-1)

SMI_cumret <- benchmark_returns %>% filter(date > date("2016-03-01")) %>% 
  tq_portfolio(assets_col  = symbol, 
               returns_col = Ra, 
               weights     = c(1,0), 
               col_rename  = "wealth",
               rebalance_on = "months",
               wealth.index = T)

SPI_cumret <- benchmark_returns %>% filter(date > date("2016-03-01")) %>% 
  tq_portfolio(assets_col  = symbol, 
               returns_col = Ra, 
               weights     = c(0,1), 
               col_rename  = "wealth",
               rebalance_on = "months",
               wealth.index = T)

portfolio_value <- price_returns %>% filter(date > date("2016-03-01")) %>% 
  tq_portfolio(assets_col  = symbol, 
               returns_col = Ra, 
               weights     = weights, 
               col_rename  = "wealth",
               rebalance_on = "months",
               wealth.index = T)
  
bind_rows(portfolio_value %>% mutate(return = "Portfolio"),
          SMI_cumret %>% mutate(return = "SMI"),
          SPI_cumret %>% mutate(return = "SPI")) %>% 
  bind_rows(tibble(date = date("2016-03-01"), wealth = 1, return = c("Portfolio", "SMI", "SPI"))) %>% 
  ggplot(aes(x = date, y = wealth-1)) +
  geom_line(aes(color = return, size = if_else(return == "Portfolio", 2, 1))) +
  geom_hline(yintercept = 0, color = "red") +
  geom_hline(yintercept = max(portfolio_value["wealth"])-1, color = "red") +
  scale_x_date(date_breaks = "3 months", date_labels = "%b %y") +
  scale_y_continuous(labels = scales::percent, breaks = seq(0,1,0.025)) +
  scale_size_continuous(range = c(0.5, 2), guide = F) +
  labs(title = "Portfolio performance over time",
       subtitle = "Jan 2016 - June 2019, monthly returns, arithmetic, excl. dividends",
       x = "", y = "Cumulative return", color = "",
       caption = "Data: Yahoo Finance, graph: mgei.github.io") +
  theme_tq() +
  theme(axis.text.x = element_text(angle = 90))

price_returns %>% filter(date > date("2016-03-01")) %>%
  left_join(names_lut) %>% mutate(symbol = names) %>% 
  bind_rows(portfolio_returns %>% mutate(symbol = "Portfolio")) %>% 
  bind_rows(benchmark_returns %>% filter(date > date("2016-03-01"))) %>% 
  group_by(symbol) %>% 
  summarise(mu = (1+mean(Ra))^12-1, sigma = sd(Ra)/sqrt(1/12)) %>% 
  ggplot(aes(x = sigma, y = mu, color = symbol, size = factor(symbol %in% c("Portfolio", "SMI", "SPI")))) +
  geom_point(alpha = 0.7) +
  geom_text_repel(aes(label = symbol)) +
  scale_size_discrete(range = c(3,7)) +
  scale_x_continuous(labels = scales::percent, breaks = seq(0,1, by = 0.025)) +
  scale_y_continuous(labels = scales::percent, breaks = seq(0,1, by = 0.025)) +
  labs(title = "Risk return diagram", subtitle = "Mar 2016 - June 2019, annualized values, portfolio equal weighted and monthly rebalanced",
       x = "Risk (Volatility, sigma)", y = "Return (mu)",
       caption = "Data: Yahoo Finance, graph: mgei.github.io") +
  theme_tq() +
  theme(legend.position = "none")

total_portfolio <- 10000000 # 1 million CHF
total_per_title = total_portfolio/(prices %>% group_by(symbol) %>% count() %>% ungroup() %>% count() %>% pull())

shares <- prices %>% 
  group_by(symbol, year(date), month(date)) %>% 
  filter(date == max(date)) %>% 
  ungroup() %>% 
  select(symbol, date, close_chf) %>% 
  mutate(date = ceiling_date(date, "month") - 1) %>% 
  mutate(shares = total_per_title/close_chf)

dividends_paid <- dividends %>% 
  mutate(date = ceiling_date(date, "month") - 1) %>% 
  group_by(symbol, date) %>% 
  summarise(dividends_chf = sum(dividends_chf)) %>% 
  ungroup() %>% 
  left_join(shares, by = c("symbol", "date")) %>% 
  mutate(dividend_payout = dividends_chf*shares) %>% 
  group_by(symbol, year = year(date)) %>% 
  summarise(dividend_payout = sum(dividend_payout)) %>% 
  ungroup() %>% 
  mutate(dividend_return = dividend_payout/total_per_title)

dividends_paid %>% 
  left_join(names_lut) %>% mutate(symbol = names) %>% 
  ggplot(aes(x = year, y = reorder(symbol, desc(symbol)))) + 
  geom_point(aes(size = dividend_payout, color = symbol)) +
  geom_label(aes(label = number(dividend_payout, big.mark = "'")), hjust = 0, nudge_x = 0.1) +
  geom_label(aes(label = scales::percent(dividend_return, accuracy = 0.1)), hjust = 0, nudge_x = 0.45, size = 3) +
  scale_x_continuous(limits = c(2016, 2019.6)) +
  scale_size_continuous(range = c(0, 10)) +
  labs(title = "Dividends", subtitle = "Portfolio",
       x = "", y = "",
       caption = "Data: Yahoo Finance, graph: mgei.github.io") +
  theme_tq() +
  theme(legend.position = "none")

Overall perspective

How did we perform? We shall take into account:

• appreciation of portfolio value
• dividends recieved
• bank/broker flat fee (40 basis points)
• income tax (on dividends, assume flat 27%)
• wealth tax (on portfolio valuation, assume flat 0.8%)

Initial sum invested 10 million CHF.

initial_sum <- 10000000
flatfee_rate <- 0.004
income_tax_rate <- 0.27
wealth_tax_rate <- 0.008

dividends_paid_yearly <- dividends_paid %>% 
  group_by(year) %>% 
  summarise(dividend_payout = sum(dividend_payout))

table_data <- tibble(portfolio_value = initial_sum, year = 2015) %>% 
  bind_rows(portfolio_value %>% 
              mutate(portfolio_value = wealth*initial_sum) %>% 
              group_by(year = year(date)) %>% 
              filter(date == max(date)) %>% 
              ungroup()) %>% 
  left_join(dividends_paid_yearly, by = "year") %>% 
  mutate(fee = if_else(year>2015, portfolio_value*flatfee_rate, NA_real_),
         income_tax = dividend_payout*income_tax_rate,
         wealth_tax = if_else(year>2015, portfolio_value*wealth_tax_rate, NA_real_),
         cash_in_pocket = dividend_payout - fee - income_tax - wealth_tax,
         change_in_portfolio_value = portfolio_value-lag(portfolio_value),
         total_performance = cash_in_pocket+change_in_portfolio_value) %>% 
  select("Year" = year, "Portfolio value" = portfolio_value, 
         "Total dividends paid" = dividend_payout,
         "Bank/broker fee" = fee,
         "Income tax" = income_tax,
         "Wealth tax" = wealth_tax,
         "Cash in hand after fee and tax" = cash_in_pocket,
         "Change in portfolio value" = change_in_portfolio_value,
         "Total performance after fee and tax" = total_performance) %>% 
  mutate(Year = as.character(Year) %>% str_replace("2015", "initial"))

options(knitr.kable.NA = "")

table_data %>% 
  kable(format.args = list(big.mark = "'"), digits = 0) %>% 
  kable_styling(bootstrap_options = c("striped", "hover", "condensed")) %>% 
  row_spec(0, bold = T)

Is it worth it?