Quant Library · Evidence and gear · Any

Specification-to-Effect Regression

Predict the effect of a product nobody has measured, from the ones somebody has.

A physics-informed Bayesian ridge in which the known term is fixed at its published value and only the unknown ones are learnt, with leave-one-out error printed beside every prediction.

Read the specification (PDF) Run it on the API The evidence and gear family · £599

Inputs

What it takes

Name	Type	Units	Where it comes from
measured products	list of (specification vector, effect, interval)	mixed	the literature
fixed coefficients	dict	-	the terms physics already settles, such as mass
prior sd optional	float	-	how far the learnt terms may roam

Outputs

What it gives back

Name	Type	Units
coefficients and their errors	dict	-
prediction and sd per product	floats	-
leave-one-out error	float	-
gap to the maker's claim	float	-

Method

How it works

Subtract the fixed physical term from the measured effects, so the regression only learns what is left.
2 further steps in the licensed specification

Assumptions

The specification vector contains the things that matter. If it does not, the residual will be large and the leave-one-out error will say so.

Limits

Ten measured products is a small training set; the model exists to bound a claim, not to certify a product.

Validation

What it was measured at

Fitted now, on the shoes with independent measurements, and used to predict the rest.

Training products	10	the ones with literature or physics behind them
Leave-one-out error	0.94 points	on an effect of about 4%
Fixed by physics	mass, at 1% per 100 g	only the rest is learnt
Largest gap to a maker's claim	+1.8 points	Puma Fast-R Nitro Elite 3

In the package

The full specification: inputs, method, assumptions and limits
The validation panel as measured on the day of purchase
A reference implementation extracted from the running source
Perpetual commercial use for one entity

Reference modules:

1 module(s), named in the licensed specification and shipped in the package

Read this before you buy. The platform's own walk-forward tests say the closing line forecasts football better than these models do. What they give you is a coherent price for every market and match, before the market opens, with the band and the working attached. Any model here that has been beaten by something simpler says so in its limits, and the validation panel above is generated from the platform's result files rather than typed in.