How We Think About Betting on Horses (And Where the Odds Work Against You)

We spend a lot of time staring at horse racing data. Over a decade's worth. Nearly 190,000 runners across more than 24,000 races. And after a while, patterns start to emerge that have nothing to do with which horse is faster and everything to do with how the betting system itself is built.

We thought it would be useful to share what we've learned. Not as advice — we're not financial advisers — but as an honest look at how the numbers work, and where punters tend to give away more edge than they realise.

Every Bet Starts in the Red

Most people think about horse racing odds the same way they think about sports odds. You see a price, you decide if it's fair, you back your judgment. But horse racing in New Zealand works a little differently under the hood, and it's worth understanding why.

The NZ TAB runs two systems side by side. There's fixed odds, where you lock in a price before the race — the same way you'd bet on the Warriors or the All Blacks. And there's the older tote system, where every dollar wagered on a particular bet type gets pooled together. Before winners are paid, the TAB takes a percentage. That percentage — called the takeout — is how the TAB and the racing industry stay in business.

The Commerce Commission's 2023 determination on the TAB-Entain partnership confirmed what industry watchers have long understood: takeout rates run from roughly 15% on straightforward win bets to around 25% on something like a First4. Put another way, for every $100 that goes into a First4 pool, only about $75 ever comes back to punters.

To give that some context, a typical sports bet carries a bookmaker margin of about 5%. Horse racing asks you to overcome three to five times that before you've even picked a horse. You're starting every race behind.

Fixed Odds vs the Tote

Entain has said that about 90% of win and place betting now happens through fixed odds rather than the tote. This shift happened because fixed odds give certainty, and punters strongly prefer knowing what they'll get paid. Most people are already using fixed odds without giving it much thought, and that's probably a good thing.

With fixed odds, the price you see is the price you get. The tote works differently — your final dividend depends on how much money flows into the pool right up until the race starts. If a wave of late money lands on your horse, your payout shrinks.

The tote does have one genuine advantage: if you're backing an outsider the public doesn't fancy, and there aren't many other people on it, your share of the pool can sometimes work out better than the fixed odds that were available. But in most situations, for most bets, fixed odds give you something valuable: certainty. And when you're trying to make decisions based on value, knowing what you'll actually get paid matters.

The Place Bet Problem

A place bet feels safer. Your horse just needs to finish in the top two or three, and you collect. And the numbers back up that feeling — looking at every New Zealand thoroughbred race in our database going back to 2016, a horse priced between $1.00 and $2.50 places 80.5% of the time. (That's based on actual starting prices from historical NZ races, not theoretical probability — it's what really happened.) Four times out of five, you get paid.

But here's why that still doesn't make place betting a better strategy.

In the place pool, the TAB takes its cut and then splits what's left across two or three winning positions instead of one. The pool is smaller to begin with and gets divided more ways. But the bigger issue is how the public bets. Most punters gravitate toward the favourite in the place pool because "it'll definitely run top three." That floods money onto the same short-priced horses and compresses the payout. Meanwhile, the horses at longer odds that might represent real value have so few backers that the dividend still doesn't reflect their true chance of placing.

The practical reality, based on a decade of data, is this: if you've done your homework and identified a horse that the market has mispriced, you'll almost always do better backing it to win than backing it to place. The place pool just eats more of whatever edge you've found.

This doesn't mean place betting is bad. It means it's mathematically less efficient for someone who has a genuine opinion about which horse should win.

Exotics: The Funnel Gets Narrower

A quinella asks you to pick the first two horses home in any order. A trifecta asks for the first three in exact order. A First4 asks for the first four.

Two things happen at the same time with these bets, and they compound each other.

One, the takeout goes up. Win pools take about 15%. Exotics like quinellas and trifectas range from about 19% to 25%. First4 pools are at the top of that range. The TAB's share grows with each step you take away from a simple win bet.

Two, the number of possible outcomes detonates. The average New Zealand field over the last decade has about 8 runners, though Saturday metro meetings regularly draw fields of 10 to 12, which makes exotics significantly harder. For an 8-horse field, that's 8 possible winners for a win bet — but 336 possible trifecta combinations in exact order, and 1,680 for a First4. In a 12-horse field, those numbers jump to 1,320 and 11,880.

You're not just facing a higher commission. You're trying to solve a much harder puzzle, and every time you box extra horses to widen your net, the ticket cost climbs with it. A four-horse boxed trifecta is $24. A five-horse box is $60. The cost of covering your bases grows faster than your chances of being right.

This is why exotic bets get marketed so heavily. The guaranteed pools, the big-number payouts, the dream of turning a few dollars into a few thousand. They're fun to play. We play them ourselves sometimes. But we don't treat them as a strategy, because the arithmetic says they're not one.

What the Data Actually Says About Favourites and Longshots

There's an old finding in horse racing research called the favourite-longshot bias. It was first documented in 1949 and has been confirmed in study after study across every racing country since. The basic idea: punters overbet longshots and underbet favourites, relative to how often they actually win.

We checked this against our own database — every New Zealand thoroughbred race from August 2016 through to June 2026, 24,050 races and 189,000 individual runners. The pattern is impossible to miss.

The win rate falls away relentlessly. Horses at $1.00–$2.50 win roughly 45 times out of 100. Horses at $20–$50 win about 4 times. Horses at $50 or longer win less than twice in a hundred starts. Across the full decade, only 135 of the 8,614 runners in that bottom band managed to win.

Now, a $2.00 chance that wins 45% of the time returns about $90 for every $100 you put through it. A $35 chance that wins 4% of the time returns about $140 — but that's before the takeout, and over the full dataset, the net result is the same: both lose money long-term. The shorter-priced horse just loses it more slowly.

There's an interesting New Zealand-specific nuance here, too. Research from Massey University has found that the favourite-longshot bias is strongest when betting opens and gets substantially corrected by sharp money in the last half hour before each race. The odds you see just before the jump are generally more reliable than the odds posted an hour earlier. The professionals are doing their work late, and the market gets more efficient as a result.

How We Think About Value — Two Real Examples

Let's walk through a couple of actual races to show how we think about this.

The Two Sets of Hands in the Tote

There's something about tote betting that most people never think about, and it quietly makes the maths worse.

In a fixed-odds market, if a professional punter spots value early — say, grabbing $2.20 about a horse that firms into $2.00 — the bookmaker absorbs that cost. You, as a regular punter taking the market price, aren't directly paying for the sharp's profit.

In a tote pool, everyone's money sits in the same bucket. The professional's profit is extracted from the exact same pool your money is in. There are two deductions at work: the TAB takes its percentage off the top, and then skilled bettors draw their edge from whatever's left. The effective loss rate for a casual punter in a tote pool is nearly always worse than the headline takeout figure suggests.

So What's the Approach That Makes Sense?

None of this means you can't win. It just means you have to be clear-eyed about what you're up against and disciplined about how you bet.

We focus on win betting. The takeout is lowest, the maths is simplest, and the relationship between how good we think a horse is and what we get paid is clearest. If we can't make the win side work, we won't make anything else work either.

We look for value, not winners. The point isn't to pick the horse that crosses the line first. It's to find horses whose real chance of winning is higher than the price suggests. A $10 shot that we rate as a $7 chance is a bet — we're getting paid for $10 on something worth $7. A $2.50 chance we rate at $3.00 is a pass, even if it wins. That $2.50 will slowly drain the account.

We're picky about what we bet on. On a busy Saturday there might be 40 or 50 races across the country. We don't have a genuine opinion in most of them, and neither does anyone else. Across the decade, the top three horses in the betting have won 68.5% of all races — more than two out of three. If we're betting on a horse outside that range, we want a clear reason why. Fewer bets, bigger when the conviction is real.

We lean toward fixed odds. Locking in the price removes one variable we can't control. If we think a horse is going to firm, we'd rather secure the price early than hope the tote pool doesn't flood at the last minute.

We treat exotics as fun, not a plan. We like a quinella as much as anyone. But we don't pretend it's a strategy. The maths on a 25% takeout combined with the sheer number of possible outcomes is not something you outwork.

We keep our stake sizes consistent. If you're betting seriously rather than just having a casual flutter, staking a small, steady percentage of your bankroll on each bet helps ride out the losing streaks. The edge, if you have one, shows up over hundreds of bets — not in any single race. Betting too much on one horse, even one you're confident about, turns a run of bad luck into a wipeout.

We Love Horse Racing

We love horse racing. The sport, the data, the puzzle of it. And plenty of people get genuine enjoyment from having a bet — that's part of the experience, and it should be.

But if you're going to bet, it helps to know the landscape you're betting into. Across a decade of NZ racing, the favourite has won less than a third of the time. The top three in the market have taken nearly 70% of all races. Horses at $50 or longer have won 135 times out of more than 8,600 starts.

The TAB takes a meaningful slice of every dollar. Place pools are structurally less efficient than win pools. Exotics are aggressively priced relative to the difficulty of getting them right. And the favourite-longshot bias means that, all else being equal, the horses the public ignores tend to be worse value than the ones they back.

The way through is the same as it's always been: be disciplined, be selective, focus on finding horses the market has got wrong, and bet accordingly. You won't win every race. You don't need to. You just need the numbers to work in your favour over time, rather than the other way around.

Betting on horse racing should be for fun and within your means. Never bet more than you can afford to lose. If gambling is causing problems for you or someone close to you, free and confidential support is available through the Gambling Helpline on 0800 654 655 or at gamblinghelpline.co.nz.

Sources: NZ Commerce Commission TAB Authorisation Amendment Determination (June 2023); Griffith, R.M. (1949), "Odds Adjustments by American Horse-Race Bettors"; Thaler, R. & Ziemba, W. (1988), "Parimutuel Betting Markets: Racetracks and Lotteries"; Snowberg, E. & Wolfers, J. (2010), "Explaining the Favorite-Longshot Bias." NZ race data drawn from 24,050 races, 189,054 runners (August 2016–June 2026), sourced from loveracing.nz.