Date: January 17, 2024

Topic: Statistics in Dataframes

Recall

Use dataframe operations to calculate statistics across the entire dataframe

Notes

Global Statistics

• Operates on entire dataframe
• e.g., mean = df1.mean() → gives the mean of each column
• Others include median, std, sum, prod, mode…

Split the data into windows, where we calculate the average across a time period and compare it to other data

Rolling Statistics

• Take statistics from snapshots/windows in the data
• Can apply the same functions (mean, std, etc) to the rolling window

Usage

• Possible usage could be if day price deviates a lot, it could be buy/sell signal

Pandas (0.20)

• https://pandas.pydata.org/pandas-docs/version/0.20/generated/pandas.DataFrame.rolling.html

<aside> 📌 SUMMARY: Dataframe allow us to calculate various statistics, and we can do global or rolling ones.

</aside>

Date: January 17, 2024

Topic: Analysis Tools

Recall

Bollinger bands are a form of technical analysis. If the price crosses the bands, then we might want to take action

Notes

Bollinger Bands

• Consider 2 s.d. above and below a rolling mean
• If the daily price deviates past the bands, we might want to take action

Using in code

1. Compute rolling mean
2. Compute rolling s.d.
3. Compute upper and lower bands

# Rolling mean
rm = df.rolling(window=20).mean() # Average over 20 days

# Rolling s.d.
rstd = df.rolling(window=20).std()

# Get bands
upper_band = rm + 2*rstd
lower_band = rm - 2*rstd

# Plot the graphs
ax = df['SPY'].plot(title="Bollinger Bands", label='SPY')
rm.plot(label='Rolling mean', ax=ax)
upper_band.plot(label='upper band', ax=ax)
lower_band.plot(label='lower band', ax=ax)

Daily returns is another indicator on how a stock might be performing, with a positive daily return being better.

<aside> 📌 SUMMARY: We can use different indicators to help us see what a stock’s performance is, like Bollinger bands, daily returns and cumulative returns

</aside>