Skip to content

EWMA

Fast-vs-slow EWM moving-average crossover signal.

tinycta.ewma

EWMA-based signal generation utilities.

This module exposes helpers built around exponentially weighted moving averages (EWMA) for use inside Polars expression pipelines. Functions operate column- wise and are suitable for DataFrame.with_columns usage in notebooks and batch pipelines.

ma_cross(prices, fast, slow, min_samples=1)

Return the sign of the fast-vs-slow EWM moving-average cross per column.

Computes two exponentially weighted moving averages (EWM) of the input price series using windows fast and slow (interpreted as com=window-1) and returns the sign of their difference. The output is -1, 0, or +1 after the warmup implied by min_samples.

Parameters:

Name Type Description Default
prices Expr

Polars expression containing the price series to transform.

 required
fast int

Length for the fast EWM mean (fast > 0). Typically fast < slow.

 required
slow int

Length for the slow EWM mean (slow > 0).

 required
min_samples int

Minimum number of observations required before EWM values are produced; earlier rows will be null.

 1

Returns:

Type Description
Expr

pl.Expr: An expression yielding -1, 0, or +1 per row after warmup.
Example

prices = pl.DataFrame({"A": [1,2,3,4,5,6,7,8,9,10]}) df = prices.with_columns( ... ma_cross(pl.col("A"), fast=2, slow=6, min_samples=3).alias("sig_A") ... )

Source code in src/tinycta/ewma.py 
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
def ma_cross(prices: pl.Expr, fast: int, slow: int, min_samples: int = 1) -> pl.Expr:
    """Return the sign of the fast-vs-slow EWM moving-average cross per column.

    Computes two exponentially weighted moving averages (EWM) of the input
    price series using windows ``fast`` and ``slow`` (interpreted as
    ``com=window-1``) and returns the sign of their difference. The output
    is -1, 0, or +1 after the warmup implied by ``min_samples``.

    Args:
        prices: Polars expression containing the price series to transform.
        fast: Length for the fast EWM mean (``fast > 0``). Typically ``fast < slow``.
        slow: Length for the slow EWM mean (``slow > 0``).
        min_samples: Minimum number of observations required before EWM values
            are produced; earlier rows will be null.

    Returns:
        pl.Expr: An expression yielding -1, 0, or +1 per row after warmup.

    Example:
        >>> prices = pl.DataFrame({"A": [1,2,3,4,5,6,7,8,9,10]})
        >>> df = prices.with_columns(
        ...     ma_cross(pl.col("A"), fast=2, slow=6, min_samples=3).alias("sig_A")
        ... )
    """
    return (
        prices.ewm_mean(com=fast - 1, adjust=False, min_samples=min_samples)
        - prices.ewm_mean(com=slow - 1, adjust=False, min_samples=min_samples)
    ).sign()