A predictions league is the football game that everybody can play: predict the score of every match, collect points when you are right, and the table sorts out who knows the game and who just supports a big club. It needs no transfers, no budget and no lineup news, which is why groups that abandon fantasy football by November are still arguing about a predictions league in May.
It does need an organiser, and that organiser needs to settle a dozen small decisions before the first deadline. This guide walks through each of them - scoring, deadlines, missed predictions, tiebreakers, prizes and the arguments you will otherwise have in the group chat - and finishes with how to run the whole thing in a few minutes a week.
Every player predicts the scoreline of every fixture in the round before it kicks off. Once the results are in, each prediction is scored: the most common system awards points for the exact score and fewer points for getting the result right (home win, draw or away win) with the wrong score. Points accumulate across the season, and the player with the most at the end wins. Most leagues also crown a winner for each month or each block of gameweeks, so a bad August does not end anyone's interest in September.
The format scales from six mates to a whole office. Below six the table is too swingy to mean much; above forty, monthly prizes become the thing people actually chase. Ten to twenty is the sweet spot where everyone knows everyone and the reveal grid after kickoff is worth reading.
The scoring system decides what kind of league you get. The standard is 5 points for an exact score and 2 for the right result, and it is standard because it works: the exact score is worth two and a half times the result, so a brave call pays, but a player who reads the outcomes right every week still finishes above one who chases scorelines and misses.
The variants worth considering, and the ones to avoid:
Deadlines cause more arguments than scoring. A single deadline for the whole round is simple to announce but unfair in practice: modern fixture lists spread a round from Friday night to Monday night, and a player who submits on Sunday has seen results a player who submitted on Friday had not. The fair rule is that each prediction locks at its own match's kickoff, so the Friday game can be revealed and discussed while the Sunday fixtures are still open.
Whatever you choose, apply it without exceptions. The moment one late prediction is accepted 'just this once', every future deadline becomes a negotiation. Software that locks automatically removes the organiser from that conversation entirely, which is the single best argument for not running the league on a spreadsheet.
Somebody will forget a gameweek. Decide now what that means. The strict rule is zero points for every unpredicted match, which is fair and needs no explanation but knocks a player out of contention after one bad weekend. The softer rule is a default prediction - usually a 1-1 draw or a 1-0 home win - applied automatically to anything left blank, which keeps people in the race without rewarding absence, since a default rarely beats a considered prediction over a season.
Reminders solve most of this. A message the day before the first kickoff and another an hour before catches almost everyone; the players who still miss it were not going to predict anyway.
Over a season, two players will finish level. Agree the tiebreaker before it matters:
Most predictions leagues are played for the table and the bragging rights, and the game loses nothing for it. If your group wants a prize, keep the structure simple and announce it before round one. Monthly prizes are the best decision an organiser can make: they keep the players in the bottom half engaged, because a good month is always within reach even when the season is gone.
Calendar months are easier than blocks of gameweeks - everyone knows when September ends - even though some months carry five rounds and others three. If you run a pot, a common split is a set amount for each monthly winner with the remainder divided between first, second and third overall, so that a player who never wins a month can still take the biggest prize. Any money involved is an arrangement between friends, handled by the organiser and never through the site; if your league is for a workplace, check the company's rules on sweepstakes and pools before collecting anything.
Every league eventually meets one of these. Write the answer down in the league description now:
A predictions league runs on attention, and attention runs on the reveal. The moment a match kicks off, everyone's prediction for it should be visible: the grid of who called what, with the exact scores lit up when the final whistle goes, is the thing people screenshot into the group chat. A short weekly summary - who won the round, who called the shock, who has taken the lead - keeps the players who had a bad weekend reading.
Monthly winners, a form table for the last five rounds, and an end-of-season awards message (most exact scores, most draws predicted, worst gameweek) do the rest. The organiser's job is to make the results visible; the football does the rest.
On paper, running a predictions league means collecting ten predictions from fifteen people every week, chasing the late ones, scoring a hundred and fifty results by hand and posting a table. Predictions League does all of that for free: create the league, choose your scoring (5 and 2 by default, or your own), set an optional default prediction for anyone who forgets, and share an invite code. Predictions lock at each kickoff, results sync automatically, the standings and the reveal grid update themselves, and the deadline reminders go out without you. Private leagues by invite code or public ones anyone can join, for the Premier League, the Championship, Europe's big leagues and the international tournaments.
5 points for an exact score and 2 for the right result is the standard for good reason: the exact score is worth enough to reward a brave call, and the result points keep a consistent predictor ahead of a lucky one. Keep it simple; three-tier systems and negative points cause more arguments than they settle.
Per match. Rounds are spread from Friday to Monday, and a single deadline means either everyone predicts before the first kickoff or the late submitters have seen results the early ones had not. Locking each prediction at its own kickoff is fair and needs no announcement.
Decide before the season: zero points for every unpredicted match, or a default prediction such as 1-1 applied automatically. The default keeps people in the race without rewarding absence, since it rarely beats a considered prediction over a season.
Most exact scores over the season first, then most correct results, then a season-long side prediction made before round one (the total number of draws, say). Agree the order before it matters.
Six is the practical minimum for the table to mean anything; ten to twenty is the sweet spot. Larger leagues work too, and monthly prizes become the thing that keeps them alive.
Most leagues expect it, and skipped matches score zero unless the organiser has set a default prediction. Predicting the whole round takes a couple of minutes on a phone.