The Science of Score Predictions: Why Crowds Beat Pundits

An Ox, a Fair, and the Birth of an Idea

In 1906, the statistician Francis Galton attended a livestock fair in Plymouth, England. Visitors were invited to guess the weight of an ox after it had been slaughtered and dressed. Nearly 800 people paid sixpence each to enter — farmers, butchers, and plenty of people who'd never so much as lifted a leg of lamb.

Galton, who was sceptical about the judgement of ordinary people, collected the entry slips after the competition and ran the numbers. What he found surprised even him. The crowd's median guess was 1,207 pounds. The ox actually weighed 1,198 pounds — an error of less than one per cent. When the mean was later calculated, it came out to 1,197 pounds. Off by a single pound.

No individual had been that accurate. But the crowd, collectively, had nailed it.

Galton published his findings in the journal Nature in 1907 under the title Vox Populi — the voice of the people. It was the first rigorous demonstration of what we now call the wisdom of crowds.

How Does It Work?

The principle is elegantly simple. When a large group of people independently estimate the same thing, their individual errors tend to cancel out. Some guess too high, others too low, but the average converges on something remarkably close to the truth.

James Surowiecki, who popularised the concept in his 2004 book The Wisdom of Crowds, identified four conditions that make it work:

Diversity of opinion — participants bring different perspectives, knowledge, and reasoning
Independence — people form their own views without being swayed by others
Decentralisation — individuals draw on their own local knowledge and experience
Aggregation — there's a mechanism to combine individual guesses into a collective answer

Sound familiar? That's essentially what a predictions league is. Each participant makes their own call. Nobody copies anyone else. And the results are tallied together. It's Galton's ox experiment, replayed every matchday.

Pundits: Confident, Entertaining... Average

If collective predictions are powerful, how do the so-called experts measure up?

Not brilliantly, as it turns out. Analysis of BBC pundit Mark Lawrenson's Premier League predictions over 13 seasons showed he correctly called around 51% of match outcomes. His successor Chris Sutton has posted similar numbers. That's barely better than flipping a coin weighted slightly towards the home team.

The issue isn't a lack of knowledge — it's the nature of the task. Football is a low-scoring, high-variance sport. A single deflected shot, a questionable red card, or a goalkeeper having the game of their life can overturn any prediction. Pundits are also human, which means they're vulnerable to the same biases as the rest of us: overrating big-name clubs, anchoring to recent form, and letting personal loyalties cloud their judgement.

A 2025 study published in PLOS ONE tracked crowd predictions across an entire Premier League season by asking participants to forecast the score of every match. The results were striking: the collective predictions consistently outperformed every individual participant. The crowd also showed a measurable bias — overestimating the "big six" clubs and underestimating promoted sides — but even with that distortion, the group forecast was more reliable than any single person's.

Why Score Predictions Are Harder (and More Interesting) Than You Think

Predicting who wins a football match is one thing. Predicting the score is another level entirely.

When you move from a three-way outcome (home win, draw, away win) to predicting the actual scoreline, the number of realistic possibilities explodes. A match could end 1-0, 2-1, 3-2, or 0-0. Each scoreline has its own probability distribution, and even small shifts — a penalty given or not, an injury to a key player — can change the landscape completely.

This is where crowd predictions become especially interesting. Individual score guesses are almost always wrong in their specifics. But the pattern of those guesses — the clustering, the averages, the spread — reveals genuine information about what the crowd collectively expects. That signal is often more informative than a single pundit's confident declaration.

It's also why predictions leagues reward consistency over bold calls. You don't need to nail the exact score every week. You need to be less wrong than everyone else over time. And the research suggests that people who think probabilistically — weighing outcomes rather than picking favourites — tend to do better.

The Limits of the Crowd

The wisdom of crowds isn't infallible. It breaks down when the conditions Surowiecki identified are violated — particularly independence. If everyone watches the same highlight reel, reads the same preview, or follows the same tipster on social media, the diversity of the group erodes. You're no longer getting 500 independent opinions; you're getting the same opinion 500 times.

Researchers have found that even mild social influence — showing participants what others have guessed — can significantly degrade collective accuracy. The crowd becomes a herd, and herds are famously bad at finding the truth.

This is one reason why well-designed predictions leagues tend to keep individual picks private until after the deadline. It preserves that crucial independence that makes the crowd smart.

What This Means for Your Predictions

So what can you take from all this?

Trust your own judgement. You don't need to be a tactical analyst or a statistics guru. The research shows that even casual fans, pooling their gut instinct with others, produce forecasts that rival sophisticated models. Your knowledge of the game — however you acquired it — has value.

Think in ranges, not certainties. Rather than committing to a single scoreline, consider the range of likely outcomes. Is this a match where both teams are likely to score? Is a low-scoring draw a real possibility? Thinking this way naturally pushes you towards more accurate predictions.

Don't follow the herd. If everyone in your predictions league is picking the same result, there might be value in going the other way — especially if you have a genuine reason to disagree. Remember, the crowd is smartest when it's diverse.

Play the long game. One week's results mean very little. Over a full tournament or season, the predictors who perform best aren't the ones with the flashiest single-week scores. They're the ones who are consistently, boringly, slightly less wrong than everyone else.

ScorePick is built on the idea that predictions are better together. Join a league, back your judgement, and find out whether the crowd — or the pundits — really know best.