A date in a historical odds file does not necessarily tell you when the quote became available. The fixture's start time, the source's update time and your collection time can all be different. Reconstructing a pre-match decision requires knowing which clock each field represents.
Keep three timestamps separate
Suppose a fixture is scheduled for 20:00 UTC. A source updates its price at 18:03 and a researcher records that price at 18:05. These are three distinct facts: event time, quote time and observed time. A single column labelled “date” loses the distinction.
If the source does not expose an update time, record it as unknown. Substituting your collection timestamp does not recover the missing information. A screenshot can document what you saw, but the time it was captured still does not prove when the price was first published.
Use a record another researcher could inspect
- Stable fixture identifier, competition and scheduled kick-off
- Bookmaker, market, selection and line where applicable
- Source update time and collection time, each with a time zone
- Decimal quote and market availability status
- Pre-match or in-play classification
- Source location, dataset version and notes about unknown fields
An ISO 8601 value such as 2026-09-21T18:05:00Z explicitly records UTC. If you retain local time, include its offset: 21:05:00+03:00 represents the same instant. Avoid matching records using clock numbers alone when the sources operate in different time zones or observe different daylight-saving rules.
Reconstruct the information available at the cutoff
If the study claims to evaluate a decision made at 18:00, the record first observed at 18:05 cannot automatically be used as an input. You would need evidence that the relevant information was already available at 18:00. The same principle applies to line-ups, statistics and later corrections to fixture metadata.
When examining the archive in OddsTips, check what each visible time field means and note any information that is unavailable. This record specification is a research method, not a claim that a particular product exposes or exports every listed field.
Preserve corrections without rewriting the past
A rescheduled kick-off can make a previously correct fixture time look wrong. Keep the correction time and its reason alongside the earlier record instead of silently overwriting the history used in a published study. That lets you distinguish the information originally available from the information now considered accurate.
Before reporting a result, select one observation and trace it back through these fields. Can another person tell which price was known, from which source, at the claimed decision time? If not, the unresolved timing issue belongs in the research limitations. A newer file is not necessarily a better reconstruction of an earlier information state.
Sources and method
Calculations and scenarios are original teaching examples, not observed match data, product performance or predictions. Sources support the stated concepts and methods. This article was prepared with AI assistance; sources and calculations were checked.
- scikit-learn: data leakage — accessed 14 September 2026
- Football-Data: data coverage and collection notes — accessed 14 September 2026