Adjusted Close from yfinance is not the same as total return from Yahoo Finance

[Allen-JThomas]

I am trying to calculate the monthly and annual percent return for a fund like VSMPX. My understanding is that the Adjusted Close from yfinance takes into account splits and dividends.

To get the percent return I am doing:

  hist = ticker.history(start="2020-01-01", end=None, interval="1d", auto_adjust=True)
  # Calculate daily, monthly, and annual percentage returns
  daily_returns = hist['Close'].pct_change()
  monthly_returns = hist['Close'].resample('M').last().pct_change()
  annual_returns = hist['Close'].resample('Y').last().pct_change()
  
  # Save the returns to CSV files
  daily_returns.to_csv("VSMPX_daily_adjusted_returns.csv")
  monthly_returns.to_csv("VSMPX_monthly_adjusted_returns.csv")
  annual_returns.to_csv("VSMPX_annual_adjusted_returns.csv")

My annual return is the following:

2016-12-31 00:00:00-05:00,0.12694628348918013
2017-12-31 00:00:00-05:00,0.21189241900874567
2018-12-31 00:00:00-05:00,-0.05155696029590995
2019-12-31 00:00:00-05:00,0.3079293379439798
2020-12-31 00:00:00-05:00,0.21010764518963132
2021-12-31 00:00:00-05:00,0.2574150697504418
2022-12-31 00:00:00-05:00,-0.19501641528697433
2023-12-31 00:00:00-05:00,0.26013908084513204

However, when I check the total return percentage on the Yahoo Finance website, my adjusted return percentages are off by a little.

For example, you can see that my 2019 annual return is 30.79%, but on Yahoo Finance, it is 30.82%. I realize this is a small descripency but I want 1) Understand why there is a difference 2) I noticed every other total return is the same as Yahoo Finance, so concerned I am doing something wrong.

Good question and also wait for an answer

[ValueRaider]

Hard to answer without Yahoo's source code. Can you get a third opinion from other source e.g. Morning Star?

[vdbergh]

I find it confusing that hist['Close'] actually refers to the adjusted close price and not to the real close price. IMHO yfinance should supply both.

EDIT: I am now seeing that this behavior is configurable through the auto_adjust parameter to history. Sorry for noise!

Incidentally: I tried to compute the adjusted close prices for https://finance.yahoo.com/quote/HYLD.L/history/?period1=1352793600&period2=1728108169 following the explanation here https://help.yahoo.com/kb/SLN28256.html but the numbers I got were also slightly off.

[ValueRaider]

The adjustment from a dividend is 1.0 - [ dividend / close day before ] . Accumulate them by multiplying. Any differences are rounding errors.

[deeleeramone]

Total returns typically assumes reinvestment on the payment date, which would create a slightly different value because of the compounding effect.

Instead of discounting the price, the base unit (1 share) increases by the multiplier, base unit * (1 + (div/close)).

The next dividend received, all else equal, would be a slightly larger amount.

The adjusted price level would reflect the increase in the base unit (1 share), not because of any cash flow, but because of compounding from reinvestment.

At the end of the holding period, the exit price would be the closing price multiplied by 1 + however many fractional shares accumulated from reinvestment.

[ValueRaider]

Instead of discounting the price, the base unit (1 share) increases by the multiplier, base unit * (1 + (div/close)).

Valid way to adjust but Yahoo does reverse.

[Allen-JThomas]

@deeleeramone @ValueRaider Thanks for the answer! Although, I am not quite sure if I follow. Would it be possible to post a code snippet to calculate total return using yfinance?

[jharemza]

Calculating Total Return with and without DRIP

Below is an example of how to calculate total returns. First we calculate total returns without dividends reinvested (DRIP) and second we redo the calculation with DRIP "turned on."

Total Return without DRIP

Formula for Total Return without DRIP

The total return over a period (start date to end date) is calculated as:

$$Total\ Return = \frac{\left(Final\ Closing\ Price + Sum\ of\ Dividends\ Received\right)}{Start Closing Price} - 1$$

Where:

• Final Closing Price = Last market traded price on the end date.
• Start Closing Price = Last market traded price on the start date.
• Sum of Dividends Received = Sum of all dividends paid during the period.

Example Code for Total Return without DRIP

import yfinance as yf

# Specify the stock and date range
ticker = 'AAPL'
start_date = '2020-01-01'
end_date = '2023-01-01'

# Fetch historical data
stock_data = yf.Ticker(ticker)
hist = stock_data.history(start=start_date, end=end_date, actions=True)

# Adjusted closing prices
close = hist['Close']

# Dividends
dividends = hist['Dividends']

# Initial and final adjusted prices
initial_price = close.iloc[0]
final_price = close.iloc[-1]

# Total dividends received
total_dividends = dividends.sum()

# Total return calculation
total_return = ((final_price + total_dividends) / initial_price) - 1
print(f"Total Return: {total_return * 100:.2f}%")

Total Return with DRIP

Formula for Total Return with DRIP

To calculate total return with DRIP:

1. Track how many additional shares are purchased using dividends.
2. Reinvest dividends at the adjusted closing price on the day the dividend is paid.
3. Update the total number of shares held over time.
4. Calculate the final portfolio value using the total shares held and the final adjusted price.

$$Total\ Return \left(DRIP\right) = \frac{Final\ Portfolio\ Value}{Initial\ Portfolio\ Value} - 1$$

Where:

• Final Portfolio Value = Total shares × Final adjusted price.
• Initial Portfolio Value = Initial shares × Initial adjusted price.

Example Code for Total Return with DRIP

import yfinance as yf

# Specify the stock and date range
ticker = 'AAPL'
start_date = '2020-01-01'
end_date = '2023-01-01'

# Fetch historical data
stock_data = yf.Ticker(ticker)
hist = stock_data.history(start=start_date, end=end_date, actions=True)

# Start with 1 share
initial_price = hist['Adj Close'].iloc[0]
shares = 1

# Initialize variables
total_shares = shares
for date, dividend in hist['Dividends'].iteritems():
    if dividend > 0:
        # Adjusted closing price on dividend date
        price_on_date = hist.loc[date, 'Adj Close']

        # Calculate additional shares purchased with dividend
        additional_shares = dividend * total_shares / price_on_date

        # Update total shares
        total_shares += additional_shares

# Final adjusted price
final_price = hist['Adj Close'].iloc[-1]

# Final portfolio value
final_portfolio_value = total_shares * final_price

# Initial portfolio value
initial_portfolio_value = shares * initial_price

# Total return with DRIP
total_return_drip = (final_portfolio_value / initial_portfolio_value) - 1
print(f"Total Return with DRIP: {total_return_drip * 100:.2f}%")

[ValueRaider]

Adding dividends to the adjusted price is total nonsense.

[jharemza]

@ValueRaider You are correct. My bad. Adjusted above comment.

[ValueRaider]

It's still total nonsense btw.

stock_data = yf.Ticker(ticker)
hist = stock_data.history(start=start_date, end=end_date, auto_adjust=True)
hist['Close'].iloc[-1] / hist['Close'].iloc[0]

That's your total return. That's it.

[deeleeramone]

It's still total nonsense btw.

stock_data = yf.Ticker(ticker)
hist = stock_data.history(start=start_date, end=end_date, auto_adjust=True)
hist['Close'].iloc[-1] / hist['Close'].iloc[0]

That's your total return. That's it.

Total returns is calculated by adjusting the base unit multiplier (here, number of shares owned), not by subtracting a dividend payment from the closing price. Subtracting the payments from the closing price distorts the answer because of an artificial beginning price which never happened, and it assumes linear market behavior not reflected in reality. It also does not account for compounding via reinvestment, this is the important part.

Example of Total Return
An investor buys 100 shares of Stock A at $20 per share for an initial value of $2,000. Stock A pays a 5% dividend the i > investor reinvests, buying five additional shares. After one year, the share price rises to $22.

To calculate the investment's total return, the investor divides the total investment gains (105 shares x $22 per share = $2,310 current value - $2,000 initial value = $310 total gains) by the initial value of the investment ($2,000) and multiplies by 100 to convert the answer to a percentage ($310 / $2,000 x 100 = 15.5%). The investor's total return is 15.5%.

[Allen-JThomas]

@deeleeramone can you help illustrate with a code snippet?

[Steve8291]

I think the confusion is from the way that Yahoo calculates "Adj Close."
See: https://quant.stackexchange.com/questions/42445/total-returns-from-adjusted-close-prices
https://help.yahoo.com/kb/SLN28256.html
Regular "Close" price is what you want because it still takes into account splits. The "Adj Close" is going to reduce the close price whenever there is a dividend distribution. The non-DRIP calculation comes out correct if you use:
hist = stock_data.history(start=start_date, end=end_date, interval='1d', auto_adjust=False, actions=True)
And then add the dividends back in as @jharemza said with:
total_return = ((final_price + total_dividends) / initial_price) - 1
You can test the output with an online calculator
Any minor differences are just rounding errors.

[jizongFox]

Hello,

I remarked that @Allen-JThomas 's issue caused minor inconsistency (often less than 0.1%) on ticker VSMPX, between the webpage 'total return' and the computed one from adj close. I also see discussions above with different alternatives.

I am drawing your attention to this matter again due to significant differences between the webpage total return and the computed one.

Taking the popular vdy.to as an example, @Allen-JThomas gives the returns as the following:

import yfinance as yf

ticker = yf.Ticker("VDY.TO")
hist = ticker.history(
    start="2016-01-01", end=None, interval="1d", auto_adjust=True, actions=True
)
# Calculate daily, monthly, and annual percentage returns
annual_returns = hist["Close"].resample("Y").last().pct_change().dropna()
print(annual_returns * 100)

with output:

2017-12-31 00:00:00-05:00     7.426081
2018-12-31 00:00:00-05:00   -12.706048
2019-12-31 00:00:00-05:00    19.242564
2020-12-31 00:00:00-05:00    -3.723435
2021-12-31 00:00:00-05:00    33.946094
2022-12-31 00:00:00-05:00    -0.866885
2023-12-31 00:00:00-05:00     4.623778
2024-12-31 00:00:00-05:00    **18.193654**
2025-12-31 00:00:00-05:00     5.533703
Freq: YE-DEC, Name: Close, dtype: float64

Method without DRIP offers a year return of 20.21% for the year of 2024, while the other one gives 20.47%

Compared with the total performance for that year (21.42%), I observed an inconsistency of more than 3%, persistently appeared in all years, for the conventional methond (first one).

The questions is then natural, should we trust the comptuation based on adj close? How to obtain a growth curve which matches/correlates well with Growth of 10,000 diagram from morningstar website (at least the key statistics should be matched)?

An example of these three prices in 2024 is plotted as below. Thank you for your attention and potential help.