CORTENSCORE
TRCreate account
Menu
Create account
NO SIGN-IN NEEDEDFree calculatorsEditor deskGlossarySpor Toto column count18+ · For information only
Home/Methodology
TRANSPARENCY · DATA LIMITS

Methodology
and data limits

An approach that makes the coverage, timing and uncertainty of the data visible.

Short answer

Cortenscore computes the probability implied by a price as 100 / decimal odds, makes the margin visible and always shows historical rates as numerator / denominator with the sample size. Price, observation and inference are different things; we do not infer a definite cause from a price move or a future result from a past percentage.

Cortenscore calculates implied probability by taking the inverse of the odds, and shows the numerator and denominator together for historical result rates. A price, an observation and an inference are different things. The period of data used, the number of matching records and any gaps can all affect a result.

What we show

For a single price, implied probability is 100 / decimal odds. In a three-way 1X2 market the three prices are converted separately and added up; the part of the total above 100% is the excess built into the prices. The opening-to-closing change is (latest price − starting price) / starting price × 100. A historical result rate should always be presented as occurrences / records examined, with the sample size. You can try the calculation in the probability and margin tool.

What we don’t do

You cannot infer a definite cause from a price movement, a future result from a single historical percentage, or full coverage from incomplete data. Different sources may record the same match at different times. Missing examples, narrow filters and information from after the match leaking into the analysis can distort a historical result. The sample size and uncertainty guide explains this limit with a numerical example. Our own pre-registered replication is an example of a hypothesis that did not hold up.

Corrections policy

If you spot a calculation or wording error, let us know with the page address, the date and a short note. Reports are reviewed by the Cortenscore editorial team; if the error is confirmed, the page is corrected.

  • The “Last updated” date at the end of the corrected page is renewed.
  • When a correction changes a number or a result, a short note explaining what changed is added to the page; the old wording is not silently removed.
  • Research articles also publish negative or non-replicating results; we do not select only results that look successful.

Last updated: 28 October 2026 · Responsible: Cortenscore editorial team · Spotted an error? Report a correction.