Historical Betting Data, Sports Betting Data & Horse Racing Data on Betfair

Historical betting data is one of the most valuable resources for anyone building racing/sports betting models, analysing betting markets, or researching horse racing performance.

Betfair provides access to a wide range of historical betting data, including horse racing data, market prices, betting volumes and exchange activity. Whether you’re building predictive models, backtesting strategies or conducting betting market research, there are multiple data sources available.

Historical Betting Data on Betfair

Once you’ve decided to start building a model or strategy for betting on Betfair, you’ll need to source some historical data to understand either the sport or racing code that you’re looking to bet into and the behaviour of those markets on the exchange.

In some problem spaces there can be significant costs to sourcing data historically, and while this is the case for some areas, Betfair tries as much as possible to provide historical datasets to customers as easily and cheaply as possible.

Sports Betting Data

Sporting data is generally easy to come by for major leagues, though the most detailed data is usually pay-walled and may not be available historically. Some sports have freely available and accessible API services whilst others a mixture of free and paid services.

Some of the most useful available paid API data sources include (correct as at June 2026):

There are also several free crowdsourced resources available for use with APIs or other automations:

• AFL – https://jimmyday12.github.io/fitzRoy/
• NRL – https://github.com/beauhobba/NRL-Data
• NFL – https://www.nflfastr.com/
• NHL – https://pypi.org/project/nhl-api-py/
• MLB – https://pypi.org/project/MLB-StatsAPI/
• NBA – https://github.com/swar/nba_api
• Soccer – https://fbref.com/en/
• Tennis – https://www.tennisabstract.com/

Racing Betting Data

Thoroughbred, harness and greyhound racing data can be more challenging to source than sport data, as it is usually heavily monetised and licensing arrangements can mean that it is difficult to share.

The most used API sources for Australian racing data are:

Betfair Australia customers can email the automation team to request free access to the Topaz API. However, the Isolynx GPS data is currently not available for Betfair users through the Topaz API.

The racing bodies that publish this data are:

• Racing Queensland – https://www.racingqueensland.com.au/industry/greyhound/greyhound-sectionals
• Greyhound Racing Victoria – https://www.grv.org.au/racedata/

Betfair CSV files

For those just starting out, csv files (Excel) are the friendliest of starting points, where race fields, starting prices and results can be assessed in bulk using common tools like pivot tables.

Promo Files – https://promo.betfair.com/betfairsp/prices

Betfair in the UK provides daily racing csv files containing race fields and race results going back to 2008. While this gives a timely and enormous dataset, care should be taken when using this data as the data:

• Is completely in British Pounds regardless of the market country
• Excludes New Zealand racing
• Excludes from the calculation of min and max prices any bets under a payout of 100GBP
• Incorrectly uses the term event_id in place of market_id

It does provide a timely (less than 24-hour lag) and mostly complete collection of racing results and BSP data.

Notes:

• MORNING_WAP is not useful for AUS racing so can be ignored
• The dates are in UTC so Australian racing will be 1 day behind

Automation Hub Data Listing

The Automation Hub is maintained by the Betfair Australia Automation team where blocks of data are published on a varying basis including:

• Australian and New Zealand Racing (monthly blocks)
• Australian Sporting Leagues (season blocks updated weekly)
  • AFL
  • NRL
  • BBL/WBBL
  • NBL
• Betfair Hub Racing Prediction Models (monthly blocks)
• Betfair Hub Racing Tipster Results (monthly blocks)

Betfair Stream Files

The Betfair Stream API feeds many tools like the mobile app and desktop site, and so historical recordings of it can be very useful for creating and backtesting wagering strategies.

These recordings are referred to as “Stream Files” and are unstructured JSON data containing the market’s activity from start to finish with varying levels of granularity.

Betfair UK makes these files available with a lag of 5 or so days through their historic data website where these files can be purchased. The levels are referred to as:

• BASIC (free files with prices updated at 1-minute intervals and no volume information)
• ADVANCED (paid files with prices updated at 1-second intervals)
• PRO (paid files with the highest level of granularity)

Betfair Australia customers interested in accessing these files should contact the Betfair automation team (automation@betfair.com.au) before purchasing.

If you have a Betfair API Live Key, then it is possible to setup software to record your files of interest directly to your local machine for further analysis.

How do I analyse the stream files?

The stream files contain only delta values (meaning changes in the market) rather than a full image of an order book at any one time. Therefore, analysis of these files must be done programmatically rather than through manual inspection (like an excel file).

There are several online tutorials available to assist you in this process and primarily, these tools will help you to:

• Build order books by adding placed bets and subtracting matched and cancelled bets
• Build traded volume ladders by adding matched bets
• Build the BSP ladder by adding bets placed into the pool

Key watchouts for analysing these files are:

• All volume information is provided in British Pounds
• Crossmatching/virtual bets are not included
• Competition / Metadata / Market Catalogue information is not provided
• Runner finishing position is not provided
• Sport scores as displayed on the website are not available

What next?

Once you’ve decided the sport or racing code you want to focus on, then get stuck into the data and start looking around and exploring. You’ll need to spend some time understanding the data and the problem space before wagering with a model can even be considered, but that’s just part of the journey and the fun!

Disclaimer: The data sources referenced here are done for information only and do not constitute a recommendation from Betfair