Relativities

Relativities is Effective’s Python library for comparative rating. Run the same risks through your rating program and competing programs, then analyze the premiums in pandas. Use the results to investigate competitive position, explain pricing gaps, and assess proposed changes by segment. Start with a market basket or a book of business represented in pandas tables. Each program maps those shared facts to its carrier’s inputs and calculates premiums locally through the rslc engine. Results retain submission IDs, rating statuses, and each rater’s outcomes for analysis in your own notebook.

Start with a pricing question

Question	Workflow
Where does our program cost more or less than the competition?	Compare carriers, then summarize by your own segments
Which rating rules explain a gap?	Explain a premium and compare inputs, factors, and coverage premiums
How does a proposed change affect our competitive position?	Measure a rate change against the current program and a competitor
How sensitive is the result to a deductible or mileage assumption?	Price a deductible change or find a mileage tipping point
Does the conclusion depend on missing facts?	Fill unknown facts and report the assumptions and rating success counts

The examples use personal auto. The data model lets you retain your own columns, tables, and analysis dimensions.

Rate one risk

Install Relativities and download the example data before running this page.

from pathlib import Path

from relativities import Rater

with Rater.load(Path("relativities-examples/example-auto.zip")) as rater:
    result = rater.rate(
        {
            "drivers": [{"age": 42}],
            "vehicles": [{"annual_mileage": 5000}],
            "coverages": {"collision": False},
        }
    )
result.to_pandas()[["id", "status", "annual_premium"]]
#    id     status  annual_premium
# 0   0  succeeded           400.0

For a reusable comparison, rate your first book, then compare carriers on a common basis.

How the pieces fit

Noun	What it is
`Rater`	A rating algorithm compiled from RSL, with its own inputs and outcomes
`Program`	A rater, the schema it accepts, and the `prepare` that builds its inputs
`Book`	Named pandas tables checked against a line’s schema
`Study`	A recipe for versions of one book: scenarios and sampled draws
`Market`	Named programs to compare

Verb	What it does
`load`	Gets a rater or a program, by catalog ID or local path
`download`	Writes a catalog bundle to a local path
`rate`	Rates carrier inputs, a book, or a study
`explain`	Shows the inputs and each calculated value behind one result
`with_…`	Returns a changed copy, and keeps the original

For carrier-specific inputs, Rater.rate returns a RatingResult; call to_pandas() for analysis. Program.rate and Market.rate return DataFrames directly. Each layer adds its key columns to the result:

You rate	With	Rows are keyed by
Book	Program	`submission_id`
Study	Program	`submission_id`, `scenario_id`, `draw_id`
Book	Market	`submission_id`, `program`
Study	Market	`submission_id`, `scenario_id`, `draw_id`, `program`

Make the comparison defensible

Each carrier defines its premium term, coverages, and outcome columns. Choose a common basis before comparing premiums. Retain unrated submissions and their reasons in your exhibit, and distinguish underwriting eligibility from rating status. Compare carriers walks through these checks. Use Prepare a book to preserve reported facts and your own segment dimensions. Use Validate a program to check the filing version, mapping assumptions, and independent rating examples. Premium comparisons measure price position. Claims, expenses, and customer response require their own evidence for profitability or conversion analysis.

Build and inspect raters

RSL expresses rating calculations, input constraints, tables, rounding, and source citations. Its compiler checks the implemented algorithm; source review and independent examples establish fidelity to the filing. Package a program to distribute that rater with its prepare, schema and examples. The Python reference, PPA book schema, and result reference describe the API contracts. The examples use synthetic households and rating algorithms. They illustrate analytical methods and package behavior rather than market premium levels.

Get started

Comparative pricing

Prepare and validate

Concepts

Rater development

Python reference

Data reference

Start with a pricing question

Rate one risk

How the pieces fit

Make the comparison defensible

Build and inspect raters

​Start with a pricing question

​Rate one risk

​How the pieces fit

​Make the comparison defensible

​Build and inspect raters

Start with a pricing question

Rate one risk

How the pieces fit

Make the comparison defensible

Build and inspect raters