Pre-Recorded vs RNG-Generated Virtual Horse Racing

Loading...
The Same Phrase, Two Different Engines
If you walk into a UK betting shop on a Tuesday afternoon and stand by the screens, you are watching one of two completely different products that both call themselves “virtual horse racing”. On the cashier’s left-hand monitor, a six-runner field crosses a finish line that was decided about ninety seconds ago by a random number generator. On the right-hand monitor, a similar-looking eight-runner field crosses a finish line that was decided some years ago, in a real race at a real course, and is now being replayed under different runner names from a fixture library. They look like the same product. They are not.
Contents
The technical name for the left-hand engine is RNG-generated virtual racing. Each race outcome is created fresh by a certified random number generator at the moment the round-loop starts, weighted by a probability model the supplier built. Inspired Entertainment and Playtech are the two big UK names that lean on this approach. The right-hand engine is pre-recorded virtual racing: a stored library of past races, shuffled and dealt out under disguised metadata, with fixed odds calculated against the known historical result. Mohio Gaming is the most visible UK supplier of this pattern.
Editorial commentary from industry publication Past The Wire framed the technological shift cleanly: horse racing is a sport that dates back centuries and has not changed much in its process, but with technological advancement a new form of horse racing has emerged in the shape of virtual horse racing. That single phrase glosses over the gulf between the two engines.
I find punters generally do not care about the engine — until they discover what determines outcomes on the product they are betting on. Then they care a lot. The remainder of this piece is the explanation I usually give them.
Pre-Recorded: A Library of Historical Races on Shuffle
A pre-recorded virtual racing engine works the way a streaming service plays back films from a back-catalogue. The supplier maintains a library of historical race recordings — sometimes thousands of them — each tagged with full ground-truth data: who won, by how many lengths, at what running positions, in what time. The library is the product. Each round-loop, the system draws one race from the pool, swaps in fresh runner names, silk patterns and commentary, and presents it as a brand-new event.
Because the outcome is known the moment the race is selected, the supplier can compute the fixed-odds book before the betting window opens. The market never has to guess what happens next. It has to disguise something that has already happened. Each race is shuffled with cosmetic randomisation — different runner ID numbers, different displayed jockey names, sometimes different overlay graphics — but the underlying finishing order is locked in by the recording itself.
Mohio Gaming runs its product on this model and certifies through GLI Europe BV. Cycle length is roughly three minutes for the six-runner variant and four minutes for the eight-runner one. The market menu fits the model: Exact (Win), Perfecta (Forecast), Trifecta, In First 3, plus markets you do not see on most RNG products — Even/Odd and Over/Under on the finishing number. Those exotic markets work precisely because the outcome is recorded, which makes pricing the unusual derivative simple.
The audit framing is different too. A pre-recorded engine does not need a certified live RNG in the same way; what it needs is a verified library, a tamper-proof selection mechanism, and provable independence between the selection of the race and any external input (like punter bets posted in the window). The lab work focuses on the fixture pool integrity and the random selection step, not on per-race outcome generation.
What pre-recorded engines lose in real-time freshness, they gain in retail fit. A betting shop cashier handing tickets across the counter wants a product where settlement is predictable, the cycle is regular and the odds book is printable in advance. Pre-recorded racing fits that workflow better than an RNG-generated alternative, which is why it lives in cashier flows more than in mobile apps.
RNG-Generated: A Fresh Result Every Round
RNG-generated virtual racing flips the order entirely. The betting window opens with displayed odds that come from the supplier’s published probability weights. The punter places stakes. The round-loop closes the window; only then does the engine pull a seed, generate the outcome from the probability model, and start the animation. The race the punter watches is rendered from the result, not the other way around.
Inspired Entertainment’s Virtual Sports portfolio is the textbook example. Flat, Jumps and Sprint products all run on this pattern, with cycle times generally in the two-to-three-minute band — Paddy Power’s virtual product runs every two minutes, which fits this supplier shape. Playtech’s virtual horse racing engine, powering Coral’s 90% RTP product, follows the same RNG-first logic. The certified RNG sits at the core; the rest of the system is animation, settlement and reporting.
There are two practical consequences. First, no race in an RNG product has ever happened anywhere else. The runner names are fictional; the finishing order is freshly generated; the time and the margin are reactions to the seed, not echoes of a past event. Second, the supplier carries higher computational obligation per round — the engine has to generate, settle and animate inside the cycle window. That is part of why Inspired’s overall cycle does not push much below two minutes; the system needs that time to produce, present and settle each round properly.
The RNG approach also makes some exotic markets easier and others harder. Win, Place, each-way, Forecast and Tricast all sit comfortably on top of a generated probability table — the supplier can price any combination market that maps cleanly to runner finishing positions. Derivative markets that need a known outcome — Over/Under on combined finishing numbers, sum-of-runners odd/even — are awkward on a generated engine because the supplier has to either price them off the model or skip them. Inspired tends to skip them. Mohio tends to offer them. That is the engine showing through the market menu.
RNG-generated is also the model that gets the headline RTP figures. When a punter reads “Inspired Racing: 80% to 92.1% RTP”, they are reading a property of a generative engine. The 90% on Coral is the same — a property of Playtech’s RNG-driven model. Pre-recorded engines tend to publish RTP less prominently because the punter-facing pricing already reflects the known outcome, and the long-run hold emerges from the book rather than from a model.
Where Each Approach Runs in the UK
The split between the two engines tracks roughly with channel. Online UK bookmakers — bet365, Coral, Paddy Power, William Hill, Ladbrokes — almost all lean on RNG-generated virtual racing from one of the major suppliers. Inspired sits behind a large slice; Playtech sits behind Coral; smaller online operators license either supplier or pull from feeds like BetConstruct’s catalogue, which became publicly visible when the firm received its UK Combined Remote Operating Licence covering eight virtual products. Their licensing head described the UKGC approval as another step forward for enlarging the portfolio for their UK customers, with the licence opening fresh distribution opportunities.
UK betting shops, meanwhile, are heavier on Mohio’s pre-recorded product. The cashier-workflow fit drives that. A licensed betting office (LBO) wants screens that loop smoothly, races that settle predictably, and a printed odds book the manager can hand to a punter who asks. Pre-recorded racing answers all three. The shape of UK retail virtual racing is something we have covered in more detail in our piece on Mohio Horse Racing.
A hybrid pattern is becoming more common in 2026. Some UK bookmakers offer both engines side by side in their virtual lobby — an RNG product from Inspired or Playtech running on the main carousel, a pre-recorded product from Mohio or a similar supplier living in a “more races” tab. The punter rarely notices the difference unless they look at the market menu, where Over/Under and Even/Odd give the pre-recorded engine away.
Regulatory treatment is identical for both. Both are virtual sports under the UKGC Combined Remote Operating Licence regime; both must pass certification at an approved test house (eCOGRA or GLI for the bulk of UK suppliers); both attract the Remote Gaming Duty that rises to 40% from 1 April 2026. The licensing regime does not distinguish between generative and recorded — only between certified and uncertified.
What the Difference Means at the Bet Slip
The honest answer is “less than you would think, for most punters”. The displayed odds, settlement timeline, and overall house edge on the two engines land in roughly the same band. A punter playing Win bets on a six-runner virtual race could be on either engine and would not be measurably better off on the day.
The difference shows up in three specific places. First, market menu: pre-recorded engines tend to offer derivative markets (Over/Under, Even/Odd, In First 3) that RNG engines tend not to. If those are markets you want, the pre-recorded product is the place. Second, perceived “form”: an RNG round is genuinely fresh, with no prior state to study, whereas a pre-recorded round is replayed from a real event, but with all identifying data stripped — you cannot study form either way. If anyone tells you they have “got a system” for virtual horse racing, both engine types are equally hostile to that claim. Third, retail vs online presence: if you bet in shops, expect pre-recorded; if you bet online, expect RNG.
If you remember one thing from this piece, make it the operational difference, not the philosophical one. Pre-recorded engines hide an outcome that already exists. RNG engines generate an outcome that does not yet exist. Both end with a horse crossing a line and a bet settling. The lab signature on the door, not the engine inside, is what matters.
If a race is pre-recorded, are the odds still fair?
Yes, in the regulatory sense. The fixed-odds book on a pre-recorded race is computed against the known historical outcome and presented with a transparent overround, certified by the supplier"s test house. The pre-recorded design does not allow the operator to manipulate which race is played; the selection step itself is audited.
Can a punter spot a pre-recorded engine just from the visuals?
Not reliably. Modern pre-recorded engines re-skin the original footage with fresh silk patterns, runner names and overlays. The clearest tell is the market menu — derivative markets like Over/Under on finishing numbers and Even/Odd on the winner only work efficiently on pre-recorded products, so their presence is a hint.
Which approach has higher RTP on average?
Across the major UK suppliers, the published RTP bands overlap. Playtech"s Coral product sits at 90%, Inspired"s RNG band runs 80% to 92.1% by bet type, and Mohio"s pre-recorded products price into similar territory. No reliable rule says one engine type is structurally higher RTP than the other; specific product and bet type matter more.
Articles
Created by the "Horse Racing Bet Game" editorial team.