`Momentum`#

Description#

Momentum(k) returns the raw price displacement over k steps:

\[ \text{Momentum}[t] = x[t] - x[t-k] \]

This is exactly Diff(k). It exists as a separately-named class because TA-Lib calls this indicator MOM and traders look for "momentum" in the API. Internally Momentum is a thin subclass of Diff -- the implementation (delay buffer + subtraction) is shared, not duplicated.

Parameters:

window_size (int, positive): the lookback k.
start_policy (str, optional): "strict" (default), "expanding", or "zero". Same semantics as Diff.

When to use which#

You want...	Use
TA-Lib parity (writing `MOM` in published strategies)	`Momentum(k)`
Any other context	`Diff(k)`

Both produce identical output bit-for-bit (matches talib.MOM to 0.0 -- exact integer arithmetic).

NaN handling#

Policy: propagate. Input NaN values are stored in the lookback. Output is NaN at any index where the function's positional formula references a NaN input; recovery happens once the NaN slides out of the lookback.

Examples#

Usage example#

import numpy as np
import plotly.graph_objects as go
from plotly.subplots import make_subplots
from screamer import Momentum

np.random.seed(0)
x = np.cumsum(np.random.normal(size=200))
mom = Momentum(window_size=20)(x)

fig = make_subplots(rows=2, cols=1, shared_xaxes=True,
                    row_heights=[0.55, 0.45], vertical_spacing=0.08)
fig.add_trace(go.Scatter(y=x, name="price"), row=1, col=1)
fig.add_trace(go.Scatter(y=mom, name="Momentum", line=dict(color="red")), row=2, col=1)
fig.update_layout(title="Change over the last 20 steps (Momentum)",
                  margin=dict(l=20, r=20, t=60, b=20),
                  legend=dict(orientation="h", yanchor="bottom", y=1.02, xanchor="right", x=1))
fig.update_yaxes(title_text="price", row=1, col=1)
fig.update_yaxes(title_text="momentum", row=2, col=1)
fig.show()

Reference#

See Diff for the full documentation. Equivalent to numpy.diff(x, n=1, prepend=NaN)[::k]-like indexing or TA-Lib's MOM.

Momentum#