Every prediction game has the same problem. People pick the favourites, week after week, and then act surprised when Nottingham Forest beat Manchester City or Bournemouth put three past Arsenal. These results are not freak accidents. They happen regularly, and if you are not accounting for them, you are leaving points on the table.

The word 'upset' makes underdog wins sound exceptional. They are not. They are a feature of the Premier League, built into the structure of the competition. And understanding how often they happen - and why - gives you a real edge in any prediction league.

How often do underdogs actually win?

Let us define terms. An 'underdog' here means a team from the bottom half of the table playing against a team from the top half. Using data from the last ten Premier League seasons, the numbers are clear.

Bottom-half teams win roughly 22-25% of matches against top-half opposition
That figure rises to around 28-30% when the bottom-half team is playing at home
Away from home, bottom-half teams still win about 15-18% of the time against top-half sides
Draws account for another 25%, meaning the favourite only wins about 50-55% of these fixtures

Read that last point again. When a top-half team faces a bottom-half team, the favourite wins barely half the time. The other half of the time, it is a draw or an upset. That is a massive chunk of results that most predictors just ignore.

Why favourites lose more than expected

Home advantage is still powerful

This is the single biggest factor behind underdog wins. A team sitting 16th in the table is a different proposition at home compared to away. Their crowd lifts them, they know their pitch, and they play with more confidence. When you see a bottom-half team hosting a top-six side, do not automatically hand three points to the visitor.

We covered this in detail in our piece on how home advantage shapes predictions. The short version: home teams win about 46% of all Premier League matches, regardless of league position. That alone should make you cautious about backing away favourites.

Fixture congestion and squad rotation

This is where the calendar becomes your friend as a predictor. Top teams in the Champions League or Europa League play midweek European matches, domestic cups, and then have a league game at the weekend. That is three matches in seven or eight days, and the fatigue is real.

When Manchester City play a Champions League knockout tie on Wednesday and then visit Crystal Palace on Saturday, they are not the same team. Key players get rested, legs are heavy, and concentration dips. These are the weeks where upsets cluster.

Pay attention to the midweek schedule. If a top team has a big European match before or after a league fixture, their chances of dropping points go up significantly.

The 'nothing to lose' effect

Relegation-threatened teams in the second half of the season play with a desperation that quality alone cannot match. When every match feels like a cup final, players find extra reserves. Managers simplify tactics, the crowd gets behind every challenge, and the underdogs become genuinely hard to beat.

Some of the biggest upsets in recent seasons have come from teams in the bottom three pulling off results against top-four sides. The pressure dynamic flips. The top team is expected to win and plays conservatively, while the underdog throws everything at it.

Early season chaos

The first four or five gameweeks of any season produce more upsets than average. Newly promoted teams are full of energy and optimism. Squads are still gelling after summer transfers. Some managers are trying new formations. The hierarchy has not been established yet.

If you are playing a prediction game from the start of the season, be more generous with underdog picks in August and September. The form table barely exists at that point, and reading form data only becomes reliable after six or seven matches.

Which underdogs are most dangerous?

Not all bottom-half teams are equally likely to pull off an upset. Some are genuinely poor and will lose most weeks. Others are competitive sides having a rough patch, or newly promoted teams with real quality.

Look for these traits in a potential upset pick:

Strong home record relative to their league position
A recent run of close defeats - losing by one goal suggests they are competitive but unlucky
A new manager bounce - teams often pick up results in the first few matches under new management
Playing against a team with fixture congestion or a big match coming up
Good defensive record even if they do not score much

A team that has been losing 1-0 and 2-1 every week is much more likely to pull off an upset than a team getting hammered 4-0 regularly. The margins matter.

How to use this in your predictions

Do not back every underdog

The point is not to go wild and predict upsets everywhere. That is just as bad as always backing the favourite. The point is to identify the one or two fixtures per gameweek where an upset is genuinely plausible, and factor that into your predictions.

If you look at a gameweek of 10 matches and you have predicted the favourite in all 10, something is probably off. Statistically, 2 or 3 of those results will not go the way the table suggests.

Pick sensible scorelines

When you do predict an underdog win, keep the scoreline tight. A 1-0 win for the underdog is the most common upset result. A 2-1 win is the next most common. You are not going to see the bottom team win 4-0 against a top-six side very often.

This ties into the broader principle that low-scoring games are easier to predict. Upsets tend to be scrappy, tight, and low-scoring. The underdog defends well, nicks a goal, and holds on.

Use draws as a compromise

If you think an upset is possible but cannot quite bring yourself to back it, a draw is a reasonable middle ground. A 1-1 prediction captures the idea that the underdog will compete without requiring them to win outright.

We wrote about when to predict a draw in detail. The short version: when you are uncertain between the favourite and the underdog, a draw is often the most efficient pick.

Think in terms of the full season

Over 38 gameweeks, the predictors who finish at the top are not the ones who got every favourite right. They are the ones who picked up points on the results that everyone else missed. Getting one or two underdog results right per month adds up enormously over a season.

This is the essence of staying consistent across a full season. You are not trying to nail every upset. You are trying to be slightly more accurate than the field, week after week.

A quick checklist for underdog picks

Is the underdog at home? If yes, the upset is more likely.
Does the favourite have a midweek European or cup match? Congestion helps the underdog.
Is it early in the season or late in the season with relegation at stake? Both increase upset chances.
Has the underdog been losing narrowly? Close losses suggest they are competitive.
Does the form table back up the table position, or is the underdog actually in decent form?

If three or more of those boxes are ticked, seriously consider backing the underdog - or at least predicting a draw instead of an easy favourite win.

The best predictors are not the ones who always back the obvious result. They are the ones who spot the matches where the obvious result is not as likely as it looks. Put this into practice on the Premier League prediction game and start collecting the points everyone else is leaving behind.

Keep reading

Home advantage is a huge factor in underdog wins. Read our breakdown of how venue shapes your predictions to understand why.

When you cannot decide between the favourite and the underdog, a draw might be your best bet. Our guide to when to predict a draw explains the logic.

Want to know which scorelines to pick? See our data on the most common Premier League scores and how they apply to upsets.

Why Underdogs Win More Than You Think