Save the rule, source and cutoff with the result
A research result is difficult to assess if its data selection and calculation cannot be reconstructed. Record the exact rule, time window, source version and treatment of missing entries. Preserve unsuccessful variations as part of the research history. Reproducibility makes a claim easier to inspect; it does not make the claim correct.
Three checks worth keeping
- Write the inclusion rule before calculating the headline.
- Store the data cutoff and transformation version.
- Keep the full trial history when comparing many candidate rules.
Put the question to work
Give a reviewer enough information to disagree for a concrete reason.
Use the research notebook to organise the next step, then explore the worked example.
No platinum-tier service, exclusive system or independently audited research is advertised.
Continue your racing research
For a related racing resource, explore PTPTips horse racing ratings and form. Check the source for changing race information.
Terms used in this guide
- Denominator
- The count of opportunities used to form a rate.
- Coverage
- The population and events for which a method actually supplies an output.
- Selection bias
- Distortion associated with how observations enter the analysed sample.
- Prospective record
- A record made before the outcome it will later be evaluated against.
- Counterargument
- A plausible alternative explanation that can be compared with the preferred one.
- Reproducibility
- Enough documented inputs and rules for another person to reconstruct the analysis.
Related reading in the network
- horse racing sample size — Make denominators and sample selection visible before interpreting a racing statistic.
- compare horse racing tipsters — Compare records using the same races, observation period and reporting rules.
- survivorship bias racing statistics — Examine the quiet exclusions behind impressive racing stories and selective charts.