Why We Bet Against the Public: Public-Money Bias Explained

One of the two core inefficiencies our entire system is built on is simple to state: popular teams are systematically overpriced. Not because bookmakers misjudge them — because recreational money floods onto the famous name, and the book responds by shortening that side's odds beyond its fair probability. The bettor on the other side of that distortion gets paid a premium for taking the unfashionable position. That's public-money bias, and this post explains how it works, how to measure it honestly, and why the popular version of this strategy — 'always fade the public' — doesn't work.

What public-money bias actually is

Recreational bettors don't bet probabilities; they bet identities. Real Madrid, the Yankees, the Chiefs, the home nation at a World Cup — brand-name sides attract volumes of money far out of proportion to their true winning chances. Casual money also skews structurally: toward favourites, toward overs (goals are fun, 0-0 is not), and toward whatever was on television last week.

A bookmaker facing that lopsided flow has a choice: hold the fair price and carry a large liability on the popular side, or shade the price — shorten the popular team's odds, lengthen the opponent's — and let the margin do double duty. Soft books, whose customer base is overwhelmingly recreational, shade heavily and persistently. The result is measurable: the odds on the popular side imply a probability *above* the true one, and the odds on the unpopular side imply a probability *below* it. Value doesn't need the popular team to lose — it needs its price to be wrong.

Why bookmakers leave the bias in

You might expect competition to erase a known bias. It doesn't, for two reasons. First, a soft book shading a popular side isn't making an error — it's maximising revenue from customers who will bet that side at almost any price. The shaded price loses nothing (those customers weren't price-sensitive) and earns extra margin on every popular-side bet. Second, the customers who *would* exploit the distortion get limited or banned; the ones who don't, stay. The bias is not a bug in the soft-book business model — it *is* the business model, which is why it persists year after year. Sharp books like Pinnacle, who welcome winning customers and balance flow at low margins, show far less of it — the full mechanics are in soft vs sharp bookmakers.

How to measure it — the sharp anchor, not ticket counts

Here is where most 'fade the public' content goes wrong. The popular method — bet against whichever side has 70%+ of tickets — is nearly useless: ticket percentages are self-reported by books, count a €5 bet the same as a €5,000 one, and are already reflected in the price by the time you see them. Fading tickets is fading old news.

The honest measurement is a price comparison. Take the sharpest available price for the match, strip its margin, and treat the result as the reference probability. Then compare each soft book's price against that anchor. Where the soft book's odds on the unfashionable side imply meaningfully less probability than the sharp reference — that gap *is* the public-money bias, expressed in the only units that matter: expected value. No narrative required, no guessing what 'the public' thinks. The bias shows up as a repeatable pattern: the gaps cluster on the counter-side of famous teams, national teams in tournaments, and heavily televised matches.

• Sharp book prices the underdog at 2.30 → de-vigged fair probability ≈ 42%
• Soft book, flooded with favourite money, offers the same underdog at 2.55
• EV = (0.42 × 2.55) − 1 = +7.1% — a genuine public-bias overlay
• Note what we did *not* use: ticket percentages, 'sharp money reports', or an opinion about the favourite

Worked example: the brand-name favourite

A Champions-League-calibre club visits a mid-table side in a league match. The sharp market prices it Home 4.80 / Draw 3.90 / Away 1.72 — de-vigged, the star team's away win is worth about 56% (fair odds ≈ 1.79). A soft book anticipating one-way traffic prices the star team at 1.62 (62% implied — roughly 6 points of pure shading) and pays 5.60 on the home side against a fair ~19%. The soft book's home price carries (0.19 × 5.60) − 1 = +6.4% EV. Nothing about this bet says the mid-table side is likely to win — at 19% they'll lose four times out of five. It says the payout is bigger than the risk is worth, which is the entire game. If EV maths is new to you, expected value betting explained builds it from scratch.

Where naive fading fails

Three honest caveats keep this strategy from becoming a slogan. First, the bias is small — typically worth a few percentage points of price, not a coin-flip advantage. It only converts to profit through disciplined staking over hundreds of bets, and a bettor without variance tolerance and a real sample-size mindset will abandon it in the first losing month. Second, not every popular team is overpriced in every match — the bias must be measured per-price against a sharp anchor, never assumed. Sometimes the public side is also the right side, and the sharp market says so. Third, the counter-side is often a grind: long odds, low win rates, and psychological pressure. That's not a flaw in the method; it's the entry fee that keeps the opportunity alive.

Public-money bias is one of the two inefficiencies our models are pointed at — the other being the information asymmetry between a well-built statistical model and an odds compiler covering hundreds of markets. The two compound: the model estimates the fair probability independently, the sharp anchor sanity-checks it, and the soft book's public-shaded price provides the payout. When all three line up, that's a value bet — regardless of whose shirt is more famous.