Calculate the “Frog In The Pan” (FIP) momentum metric over one or more rolling windows using either the pandas or polars engine, augmenting the DataFrame with FIP columns.
The FIP momentum is defined as:
For fip_method = 'original': FIP = Total Return * (percent of negative returns - percent of positive returns)
For fip_method = 'modified': FIP = sign(Total Return) * (percent of positive returns - percent of negative returns)
An optional parameter, skip_window, allows you to skip the first n periods (e.g., one month) to mitigate the effects of mean reversion.
Input pandas DataFrame or grouped DataFrame containing time series data.
required
date_column
str
Name of the column with dates or timestamps.
required
close_column
str
Name of the column with closing prices to calculate returns.
required
window
Union[int, List[int]]
Size of the rolling window(s) as an integer or list of integers (default is 252).
252
reduce_memory
bool
If True, reduces memory usage of the DataFrame. Default is False.
False
engine
str
Computation engine: ‘pandas’ or ‘polars’. Default is ‘pandas’.
'pandas'
fip_method
str
Type of FIP calculation: - ‘original’: Original FIP calculation (default) where negative FIP indicates greater momentum. - ‘modified’: Modified FIP where positive FIP indicates greater momentum.
'original'
skip_window
int
Number of initial periods to skip (set to NA) for each rolling calculation. Default is 0.
0
Returns
Name
Type
Description
pd.DataFrame
DataFrame augmented with FIP momentum columns: - {close_column}fip_momentum{w}: Rolling FIP momentum for each window w
Notes
For ‘original’, a positive FIP may indicate inconsistency in the trend.
For ‘modified’, a positive FIP indicates stronger momentum in the direction of the trend (upward or downward).
Examples
import pandas as pdimport pytimetk as tkdf = tk.load_dataset('stocks_daily', parse_dates=['date'])# Single window with original FIPfip_df = ( df.query("symbol == 'AAPL'") .augment_fip_momentum( date_column='date', close_column='close', window=252 ))fip_df.tail()