Last updated: 21 June 2026
At some point, every serious backer has the same thought. The tips have let you down, the tipsters have let you down, and you start to wonder whether you could just get hold of the raw racing data and work it out for yourself. It is a good instinct. Almost everything genuinely useful in this game starts with data you can trust. The trouble is that nobody tells you where that data actually comes from, or what it costs you once you go looking for it.
So let us answer the question properly, because it is less obvious than it sounds, and the cheapest-looking option is usually the most expensive one in the end.
Where can you get horse racing data? You have two routes: buy it from a licensed provider such as the Racing Post, the Press Association or Timeform, or from a reseller like Betwise, or scrape it from racing websites yourself. Buying is the only legitimate option, because the data is owned and licensed by whoever publishes it. Scraping looks cheaper, but it costs you in time, maintenance and reliability.
Your two real options: buy the data or scrape it
There are only two ways to get racing data into your own hands. You can buy it, or you can scrape it off other websites. Everything else is a variation on one of those two.
The first thing to understand is that racing data is owned. When you look at a racecard on the Racing Post, that data belongs to them. When another site shows you the same information, they have almost always licensed it from a provider to do so. Which means that when you scrape it, you are taking something that is not yours to take. For personal use, behind your own front door, the odds are nobody will ever know, and what you do there is your business. But it is worth being honest with yourself about what scraping actually is before you build your weekend around it.
Why scraping costs far more than it first looks
Scraping is free in the sense that a puppy is free. The real bill arrives later. The moment a website changes its layout, your feed breaks. The moment they tweak their code, your feed breaks again. Unless you are a developer yourself, keeping a set of scrapers alive across a dozen pages that all change on their own schedule turns into a second job, and not a fun one.
That is where the only question that matters comes in: how much is your time worth? If you are spending three evenings a week patching broken scripts instead of studying races, the “free” data has quietly become the dearest thing you own. Work that cost out honestly and you may find that buying clean data is not the expensive option at all. It is the one that gives you your evenings back.
Where to actually buy UK racing data
If you decide to buy, there are three main providers in British racing: the Racing Post, the Press Association and Timeform. All three price differently, and all three can be negotiated with, so it is worth asking rather than accepting the first number you are quoted.
The catch for an individual is the cost. Provider-grade data is priced for businesses, and it can be far too expensive to justify for personal use. That is where a reseller becomes the sensible middle path. A site such as Betwise buys data from the Press Association and is licensed to sell it on to you at a price an individual can actually stomach, with guides to help you get started. You will still need to get your hands a little dirty, because a raw database has no friendly front end. You have to learn how to ask it questions. The payoff is real, though: once you can interrogate the data yourself, you are no longer limited to whatever a system builder decided to let you ask. You can ask it anything.
Raw data is not the same as usable data
Here is the part almost nobody warns you about. Racing data is not all created equal, and raw data can be a long way from trustworthy. Years ago, when Race Advisor ran on Press Association data, it arrived raw, with no checks applied. We had horses recorded with winning times that would have meant they galloped at over 100 miles per hour. We had goings that were simply wrong. The processes we had to build just to catch the errors, before we could do anything useful with the numbers, grew so large they became unmanageable. That experience is exactly why the Race Advisor data now comes from the Racing Post.
The lesson generalises. A number is only worth as much as the checking behind it. A going code that is wrong poisons every speed figure that depends on it. A duff winning distance quietly distorts every rating built on top of it. Before you trust a single clever conclusion, you have to trust the rows it came from, and that cleaning work is invisible, thankless, and absolutely the difference between analysis and guesswork.
You can see what that cleaning buys you the moment the data is ranked properly. Across more than 1.4 million British runners with a valid Betfair SP since 2013, the top horse on Race Advisor’s PR rating wins 33.4% of the time, while a horse ranked fifth or lower in the field wins just 5%. The VDW rating sorts the field the same way, its top-ranked runner winning 26.8%. None of that separation is possible until the rows underneath have been checked first.
And here is the honest half of it. Across every one of those ranks, the A/E, the ratio of actual wins to what the market expected, sits between 1.02 and 1.07. Trustworthy data points you straight at the likely winner. It does not hand you a free edge. It tells you where to look, and the price still decides whether the bet is worth making.
The shortcut: let someone else do the data work
None of this means you have to build a database to bet well. The reason to own the data is to be able to ask it honest questions, and you can reach the same place without the plumbing. The whole point of cleaned, structured data is what it lets you do next: test whether a factor genuinely beats the market using the A/E ratio, read trainers and jockeys by condition rather than by reputation through proper trainer statistics, or weigh several ratings against each other instead of trusting one figure by combining ratings.
That is the work Race Advisor has already done. The dataset behind the ratings covers around thirteen years of British racing, roughly 1.66 million runners between 2013 and 2026, all checked and cleaned before a single rating is calculated. You get the answers without spending your weekends repairing scrapers or arguing with a provider over a licence. If you would rather skip straight to the part where the data earns its keep, put it to work on today’s racecards and let the cleaning, the checking and the ratings run in the background while you study the races that matter.
