Each-Way Strategy: Field Size, Odds and Race Type Filters

The three filters that decide whether each-way is even sensible

I keep a notebook of races I should not have bet. It runs to about thirty pages now, and after seven years of running place-market analysis I can tell you that ninety percent of the entries in that notebook fail on the same three checks. The field was too small. The price was too short. The race shape did not justify the bet. Three filters. That is the whole strategy in three words.

The standard punter reads an each-way race the wrong way round. They start with the horse — its form, its trainer, its course record, the chance the day looks right. Then they decide whether to back it each-way as an extension of having decided to back it at all. That sequence loses money systemically. The professional sequence is the reverse: start with the race, decide whether it is the kind of race where each-way works structurally, and only then decide whether the horse in it is the kind of horse that makes the bet worth striking. The race comes before the selection. The selection comes before the stake.

Two numbers frame the problem. UK Flat racing averaged 8.90 runners in 2025 across the year, while National Hunt averaged 7.84. Roughly 26% of all British races sat at the 5 to 7 runner band in the same period — the band where each-way pays only two places. The shape of the British calendar is structurally weighted toward the part of the rulebook that punishes the each-way punter. Most each-way bets are made on races that should not have been each-way bets at all, because the field-size filter alone screens them out. That is the entry point to the framework that follows.

The framework is not a set of magic numbers that produces a profit. It is a decision sequence — three filters applied in order, with each filter capable of vetoing the bet regardless of what the next ones would have said. The point is to stop being seduced by attractive prices in unattractive races. The point is to stop placing bets the rulebook does not give you a fair chance of winning.

Filter 1: field size, the structural filter

Field size is the first filter because it determines whether the each-way market structurally exists in a usable form. Everything else — the price, the race profile, the day’s expectations — sits subordinate to this single number. Get this wrong and no amount of analytical work on the horse can recover the bet.

The rulebook gives you the tiers. Five to seven runners pays two places at a quarter. Eight runners and above pays three places at a fifth on non-handicaps and at a quarter on handicaps of twelve runners or more. Sixteen or more handicap runners pays four places at a quarter. The numbers are uniform across UK firms because the standard rulebook is uniform. The question is what they mean for your bet.

In small fields, each-way is structurally wrong for most prices. A 6/1 horse in a seven-runner non-handicap pays a place fraction of 1.5 to 1 — a £10 place stake returns £25 if it places. That return is not nothing, but you have to finish first or second to collect anything and the implied place probability you need to break even is around 40%, which is heroic for an outsider profile and unnecessary for a favourite (you would back it to win at better return). Small fields tilt the each-way punter into either backing favourites for diminished return or longshots that need second-or-better against the maths. Both are losing approaches across a season.

In medium fields — eight to eleven runners on handicaps, eight or more on non-handicaps — the structure is workable but unspectacular. The fifth fraction is the constraint. A 12/1 horse pays a place at 12/5, or 2.4 to 1. A placed bet returns £34 from a £20 outlay. Acceptable arithmetic, but not the bet where each-way earns its keep across a season. This tier is where most of the British racing programme actually sits, and it is the tier where the discipline of “would I take this win-only at the same price” becomes the right counter-test. If the win-only bet is the better expected value, the each-way structure is paying tax for the place insurance.

The tier where each-way structurally rewards the punter is 16 plus runner handicaps, where four places at a quarter give the fourth-place finisher a meaningful return. A 25/1 horse that finishes fourth pays £72.50 from a £20 outlay — the bet has cleared cost on a finish position that nothing else in the each-way universe rewards meaningfully. This is the tier the framework prioritises. It is also the tier that the British calendar deliberately rations to its bigger meetings, which forces a calendar-aware strategy across the season. The realistic distribution of field sizes across the British programme and the strategic case for skewing toward better-attended fixtures is covered in the article on the two-speed Premier and Core each-way market.

The filter is binary in practice. If the field will be 16 or more runners at the off in a handicap, the each-way structure is on. If it is 12 to 15 in a handicap, the structure is acceptable at the right price. If it is 8 to 11 in a handicap or any non-handicap of eight or more, you need the price filter to do real work before the bet earns its place on the slip. If it is 5 to 7 runners, you are essentially looking at win-only with optionality on second. Below five, no each-way market exists at all.

Filter 2: the odds band where each-way breathes

The price filter is the second layer. Even in a structurally strong field, the price of your selection has to sit in the band where the each-way maths produces meaningful return. Short prices kill the place leg. Very long prices stretch the win probability beyond what is realistic. There is a band in between where each-way actually breathes.

The lower bound is structural. A 3/1 horse on the place leg at a quarter pays back 75p plus stake — barely above stake itself. The case for backing such a horse each-way rather than win-only is essentially nil, because the place dividend is too small to compensate for the win-leg drag. The band where each-way starts to add value is roughly 8/1 or longer, depending on the place fraction. At a quarter, an 8/1 chance pays a place fraction of 2 to 1 — £30 back per £10 placed. That is the lower threshold where the each-way structure begins to function as designed.

The upper bound is harder to set because it depends on what you think the realistic win probability looks like. At very long prices — 40/1, 50/1, 66/1 — the win leg becomes essentially a lottery ticket and the place leg is doing most of the structural work. The bet stops being an each-way in the conventional sense and becomes a place-only bet wearing each-way clothing. That is not wrong, but it is a different bet from what the structure was designed for. The cleaner answer in that range is often to bet the horse to place directly on an exchange, where the place market is priced independently of the win odds.

The band where the value sits, in my analytical experience, is 10/1 to 25/1. Industry data on extra-places promotions and the underlying value mathematics indicates that the +EV edge against the bookmaker concentrates in genuine 16/1-plus shots in competitive handicaps — and the same logic extends downward into the 10/1 to 16/1 range when the field is large and the place fraction is a quarter. Outside that band, the each-way structure tends to either pay too little (short prices) or pay too rarely (extreme longshots). Inside it, the structure does what it was built to do.

The price-filter discipline is to recognise that a fancied selection at 5/1 is not a candidate for each-way regardless of how strong the form looks. The right bet on that horse is win-only, possibly with a stake larger than you would otherwise have placed because you are not splitting it across two legs. Conversely, a longshot you fancy at 22/1 in a 16-runner handicap is exactly the each-way candidate the framework is built to identify, and the strike should follow.

One observation that crystallises the price-band point. The way each-way is sold to recreational punters in advertising — as “insurance” or “two-way action” — is structurally backwards. The bet is at its weakest as insurance, because the insurance is most expensive (in lost win expectation) when the win probability is high and the place dividend is correspondingly small. The bet is at its strongest as a positive-expected-value play in races where the win probability is moderate but the place probability is meaningfully higher than the price implies. Each-way is not insurance. It is a place-leveraged positional bet, and the price band where it pays is narrower than the market would have you believe.

Filter 3: handicap, non-handicap and competitive shape

The third filter is the race type itself, which sits on top of the field-size and price filters and either amplifies or dampens the value the first two filters have identified. Race type is not just a label. It tells you about the structural competitiveness of the field, the realistic distribution of finishing positions, and the operator’s implicit liability calculation for the place pool.

Handicaps win this filter consistently, for the reasons the rulebook itself reflects. A handicap is a race in which the official handicapper has assigned weights to bring the contenders theoretically closer together. Even allowing for the imperfection of the handicap process, the structural intent is to produce a competitive field where multiple horses have realistic claims, and the rulebook rewards this by paying a quarter the win odds at twelve runners or more rather than the fifth that applies to non-handicaps. Premier Fixture turnover per race rose 2.7% in 2025 versus 2024, while Core Fixture turnover per race fell 8.6% — a two-speed market that reflects the punter’s recognition that the better-quality competitive racing at premier fixtures is where the each-way value lives.

Non-handicaps lose this filter on average. Group races, Listed races, conditions races, novice races — all carry the fifth fraction and the structural disadvantage of typically having one or two horses the market identifies as the likely winners, locking up the front of the place pool. The exception is the deep-field non-handicap — a 14 or 16-runner maiden, a large summer Listed race — where the breadth of the field starts to mimic the handicap dynamics even though the fraction stays at a fifth. Those are uncommon and worth flagging when they appear on the card.

Within the handicap category, competitive shape matters. A 16-runner handicap headed by a 9/4 favourite is structurally different from a 16-runner handicap headed by an 8/1 favourite. In the first race, the favourite is doing most of the work of locking up one of the paid places, leaving the punter fighting for second, third or fourth with the rest of the field. In the second race, the favourite is a much less certain place finisher, and the punter’s selection has a wider opening to one of the four places without contending against an entrenched favourite. The framework prefers the second race shape — the genuinely competitive handicap — over the first.

Richard Wayman at the BHA has framed the Premier-Core split deliberately: “We wanted to make our best racing better and use that as our tool to grow interest in the sport. Part of that was making the racing as good as we could, so we invested in the racing and we have seen a real upside on that.” The strategic intent is to concentrate the best racing in fewer fixtures, which has the side effect of concentrating the best each-way structural settings in those same fixtures. A weekly each-way strategy that does not over-index on Premier fixture handicaps is leaving structural value on the table.

When the three filters line up

The framework is not about applying each filter in isolation. It is about identifying the races where all three filters give the green light simultaneously — and these races are rarer than punters assume.

The full-stack each-way race looks like this. A handicap of 16 or more runners. A selection priced 10/1 to 25/1. A competitive race shape with no entrenched short-priced favourite. Premier fixture preferred but not strictly required. Place terms at a quarter the odds, four paid places (or more, if an extra-place promotion is in force). This is the race the framework is built to identify, and on a typical week in the British calendar there are perhaps four to eight such races worth striking.

Two of those, in a typical Cheltenham or Aintree week, sit at the festival itself. The 28 races of the 2025 Cheltenham Festival all ranked in the top 31 most-bet-on horse races of the year in the UK, and the handicap races among them are exactly the structural profile the framework targets. A festival week therefore concentrates the framework’s bets in a relatively narrow window. Off-festival weeks demand patience — sometimes nothing on the Saturday card meets all three filters and the discipline is to leave the day alone.

The opposite case — when all three filters reject — is just as important to recognise. A 7/1 chance in an 11-runner non-handicap fails the field-size filter for premium structure, fails the price filter for the lower threshold and fails the race-type filter for fraction. Three rejections, regardless of how good the horse looks on the form. The right answer is no bet, or a small win-only at most. Strategy is not magic. It is consistent rejection of bets that do not pass the structural criteria.

The middle case is where most of the actual decision work happens. Two filters pass, one fails marginally. A 14/1 chance in an 11-runner handicap — fields-size filter borderline (handicap band, but at the lower end of acceptable for the fifth fraction), price filter passing comfortably, race-type filter passing. Bet or no bet? Here the answer depends on the wider context: is there an extra-places promotion that bumps the structural value? Is the win price particularly generous against other firms’ equivalents? Is the form profile of the horse a strong place candidate rather than a likely winner? The marginal-filter bets are the bets where analytical depth matters. The full-pass and full-fail bets answer themselves.

Premier vs Core: where the each-way punter actually lives

The British Horseracing Authority’s deliberate split of the fixture programme into Premier and Core categories has fundamentally reshaped where structural each-way value lives in the UK calendar. Understanding that split is essential to running the framework consistently across a season.

The numbers tell the story. Premier Fixture turnover per race rose 2.7% in 2025 versus 2024, while Core Fixture turnover per race fell 8.6%. The BHA itself has framed the divergence as a behavioural response: “This preference for our highest profile fixtures is undoubtedly linked to the impact of affordability checks with there being fewer larger staking customers, who have either stopped betting or are placing their bets elsewhere, and have been only partially replaced by more recreational punters betting in smaller stakes, primarily at the bigger meetings.” The serious money is concentrating at premier fixtures. The casual money is dispersing across the rest of the programme.

For the each-way framework, this concentration matters in two ways. First, the field sizes at Premier fixtures are larger on average — Premier Flat fixtures averaged 11.02 runners in 2025 and Premier Jumps 9.41, both meaningfully above the Core averages of 8.54 on the Flat and 7.63 over Jumps. Larger fields mean the four-place tier is more accessible and the three-place tier is more reliably populated. The structural filter passes more often at premier fixtures than at core fixtures, by the simple arithmetic of field count.

Second, the operator behaviour around premier fixtures is more promotionally aggressive. Extra-places promotions, enhanced place terms, best-priced offerings — these cluster at premier fixtures because the marketing budget is concentrated where the customer attention is. The promotional overlay amplifies the structural advantage, which compounds the case for premier-fixture bias in the framework.

What the framework cannot do is rescue the Core fixture programme. Core Fixture handicaps with the right field-size, price-band and race-type profile do exist, and when they appear they are worth taking — but they are rarer per fixture than at premier fixtures, and the operator promotional support is thinner. The honest counsel for the each-way framework user is that perhaps two-thirds of your bets will sit at Premier fixtures, perhaps a quarter at major Core handicaps that pass the filters anyway, and the residual at the genuinely outlying races that nonetheless align all three checks.

Risk controls: stake split, bank size, sample variance

Strategy without bankroll discipline is theatre. The mechanical structure of the framework has to be wrapped in a stake regime that survives the variance the structure produces, which means the bankroll question is not secondary — it is constitutive of the strategy.

The starting point is the size of the bank itself. Each-way betting at the strike rate the framework can realistically produce — somewhere between 25% and 35% of bets placing, with maybe 6% to 10% actually winning — has variance that requires a bank large enough to survive the inevitable losing runs. Health Surveys for England historically recorded a problem-gambling rate of 2.8% among horse-race bettors, which is structurally lower than several other gambling categories but still material. The framework user has to set the bank size honestly: an amount that can lose its full value without affecting the punter’s daily life or financial obligations. Anything less and the bet sizing decisions get distorted by the wrong incentives.

Within the bank, the unit stake should be sized so that the worst plausible losing run does not exceed ten or twelve percent of the bank. For an each-way strategy with the variance profile the filters produce, that typically means unit stakes of roughly 1% to 2% of bank per each-way bet. A £2,000 bank therefore sustains £20 to £40 each-way unit stakes — £40 to £80 total outlay per bet. Stakes meaningfully above that level expose the bank to variance shocks the strategy cannot recover from over a reasonable timeframe.

The stake split between win and place legs is the simplest decision: equal stakes per leg, the standard each-way structure. There are sophisticated stake-weighting arguments for over-weighting the place leg in races where the framework identifies particularly strong place candidates, but those arguments tend to break down in execution because the operator does not allow non-standard each-way stakes at most fixed-odds firms. The exchange permits independent leg sizing, and the place-overweighting variant of the framework can be built there for the punter who has crossed that operational bridge.

Sample variance is the last bankroll-relevant consideration. The framework’s edge, where it exists, is small in percentage terms and is realised across hundreds of bets, not dozens. Twenty bets is too few to know whether the framework is working — the variance band on twenty each-way bets is wide enough that a winning system can look like a losing one and vice versa. A hundred bets begins to settle. Five hundred bets gives you a meaningful read on whether the approach is producing positive returns. Punters who recalibrate the strategy after twenty bets are reacting to noise, not signal.

Three worked each-way decisions through the framework

Three concrete examples to show the framework applied to specific decisions, with the actual reasoning rather than the headline conclusion.

Race one. A 19-runner Cheltenham handicap on Festival day three, with a horse priced 16/1 that has won at the course in the same conditions before. Field-size filter: 16-plus runner handicap, four places at a quarter, full pass. Price filter: 16/1 sits in the band where the each-way structure does meaningful work, full pass. Race-type filter: competitive handicap with no entrenched 5/4 favourite, several horses priced 6/1 to 12/1, full pass. All three filters give green lights. With an extra-place promotion likely on the day lifting the paid places from four to five, the structural setting is exactly what the framework is built to identify. Strike the bet, sized at unit stake, accept the variance.

Race two. A nine-runner novice hurdle midweek at a Core fixture, with a 5/2 favourite the punter likes. Field-size filter: nine runners, non-handicap, three places at a fifth, marginal. Price filter: 5/2 sits below the threshold where the each-way structure earns its keep, fail. Race-type filter: non-handicap, fifth fraction, structural disadvantage relative to a handicap of the same size. Two clear fails on the framework. The bet, if struck at all, is win-only — the each-way structure is throwing money at a place dividend that would not justify the lost win expectation.

Race three. A 12-runner Saturday handicap at a Premier fixture, with the punter’s selection priced 11/1 and a non-runner declared in the morning dropping the field to 11. Field-size filter: 12 to 15 runner handicap tier holds at 11 runners — actually drops to the 8 to 11 tier, which is three places at a fifth rather than a quarter. The non-runner has materially changed the structural setting of the bet. Price filter: 11/1 in the band where the structure works, passes. Race-type filter: handicap, passes structurally but the fraction loss is the issue. The right answer is to either accept the reduced terms and bet smaller — the bet is no longer the framework’s target profile — or pass entirely and wait for the next race. The decision is contextual: how strong was the framework’s case before the non-runner, and how much of that case rested on the four-versus-three-places distinction.

What the framework cannot do

The framework is a structural filter. It identifies the races where each-way is structurally a sensible bet. It does not — and cannot — tell you which horse to back. Selection is a separate analytical discipline that sits on top of the framework, and treating the framework as a complete strategy is the mistake the most enthusiastic adopters of any system make.

The framework also cannot guarantee profit. A perfectly disciplined application of the three filters across a season can still produce a losing run that lasts months, because the variance band on each-way betting is genuinely wide and the structural edges the framework identifies are small in percentage terms. Anyone offering a system that produces consistent profit on the basis of three filters alone is offering something the maths does not support.

What the framework does is improve the population of bets you take. It moves your stake away from structurally bad races — small fields, short prices, fifth-fraction non-handicaps — and toward structurally good races — large handicap fields, prices in the value band, competitive shapes. Across a sufficient sample, that shift improves the expected return per pound staked, even if the per-bet variance remains substantial. The improvement is the only thing the framework is supposed to deliver, and any expectation beyond that is the user’s projection, not the framework’s promise.

Created by the "Racing Place Betting" editorial team.