skaters

Home of Laplace, a light, fast, automatic distributional univariate time-series prediction algorithm that also reliably improves the performance of other time-series models (including Prophet). Recommended for non-price economic time-series and available in Python, JavaScript, R, Julia and Rust.

One call, zero dependencies, every prediction a full probability distribution computed online in O(1).

On 894 continuous non-price FRED series, scored one step ahead per series, laplace has the best held-out log-likelihood in the field, ahead of AutoARIMA, AutoETS, SARIMAX, conformal methods, the heavy-tail GARCH-t, and four zero-shot foundation models. It gives up ground only on CRPS, to the CRPS specialists, and on price and return series, where GARCH-t is the better choice. The benchmarks →

Each language is an independent, parity-locked port, checked against the reference to 1e-6 (Python also has a Rust backend for speed). A fitted model serialises to JSON, so one fit on a server resumes unchanged in the browser. The languages →

A calibrated surprise signal comes with the state: each point is scored against the forecast issued for it, so state["z"] is a normalized residual and state["pit"] its probability integral transform. See it live → Turning that signal into streaming anomaly detection with honest false-alarm rates is its own project, timemachines, built on skaters.

pip install skaters. Every prediction is a Dist — a weighted Gaussian mixture carrying the mean, the spread, quantiles, and a density at once. You build models by composition: transforms chain, ensembles nest, and a distributional leaf sits at the bottom (see the Guide).

Quick start

from skaters import laplace

f = laplace(k=3)
state = None
for y in observations:
    dists, state = f(y, state)
    dists[0].mean              # point forecast
    dists[0].std               # uncertainty
    dists[0].quantile(0.975)   # 95th percentile
    dists[0].logpdf(y)         # log-likelihood
    dists[0].cdf(y)            # CDF at y

Every skater returns list[Dist] — one weighted mixture per horizon $h = 1, \ldots, k$. Point forecasts, uncertainty, density evaluation, and quantiles are all facets of the same object.

One general forecaster

skaters exposes exactly one forecaster, laplace. At multi-step horizons (k>1) it is multi-scale by default, mixing instances on decimated clocks by likelihood (opt out with scales=[1]). Everything else is a building block you can compose. ("skater" is the concept — any (y, state) → ([Dist], state) function — not a function name.)

from skaters import laplace

f = laplace(k=1)

A likelihood-weighted ensemble over the full candidate pool — model first, conform last, with the lattice projection on by default, and an Ornstein–Uhlenbeck mean-reversion group in the multi-step (k>1) pool. Specialist behaviour (mean reversion, GARCH-style volatility) is reachable by composition (ou_transform, garch_leaf) when you have a strong prior.

A bijection, not just a forecaster

laplace's calibration state maps each arriving point through its own predictive cdf, z_t = Φ⁻¹(F_t(y_t)) — a causal bijection on paths (Rosenblatt, 1952) whose image is, under calibration, iid N(0,1). Run any forecaster in those coordinates and score it back in the original ones by exact change of variables: on 30 FRED series, ETS, AutoARIMA, GARCH(1,1) and Prophet each gain ≈+2 nats/point and lose to their fronted selves on 30/30 series — converging to laplace plus a few hundredths of a nat. See the diagram computed live, or the anomaly-detection consequences at timemachines (DSPOT 5.2×, RRCF 1.8× in the same coordinates).

Heritage

skaters is a from-scratch rewrite that distils ideas from timemachines, the competition-winning forecasting package by Peter Cotton, and builds on years of experience running live distributional prediction contests at microprediction — where forecasts are scored, continuously, as full distributions rather than points. Sibling packages: precise (online covariance) and humpday (global optimizers).

Get the source

github.com/microprediction/skaters · pip install skaters · Examples