Millers Prediction League Rules   

How to Take Part

You need to be registered with and logged in.
You seem not to be currently logged in. If you haven't registered, click on "Register".

It is entirely free to enter!

The Idea

The idea is fairly simple; just predict for the upcoming Rotherham United game:

the result (goals for and against)
one Rotherham United player to score a goal (even if you don't think we'll score; own goals don't count and nor do players scoring AGAINST Rotherham United)

You will receive marks according to how well you do (see below), and your points will determine your standings in the league or cup competitions.


You are predicting match scores at FULL TIME (90 minutes), even if there is the possibility of extra time.
However, goals scored by your nominated player in extra time DO COUNT.

Predict via the forms on this website (or by e-mail).

NOTE! Your predictions MUST be in before the scheduled kick-off!

League Games
These are the ones which go towards the league tables. There are two prediction seasons within each football season and, depending on the league settings, promotion and relegation may apply. Play-off games are NOT part of the prediction league proper, but you can predict them for fun.

Cup Games
As in the real thing, prediction cups are knock-out affairs, but usually start with a qualifying round which is in league format across more than one 'real' round. Each cup that Rotherham United compete in will usually have a prediction competition, and in the unlikely event of our being knocked out, the subsequent prediction matches will be those of the team who beat us - a similar principle applies to them. Note though that occesionally 'real' cups have a group stage. If Rotherham United don't make it through that, the group winner will assume prediction team status.

Point Scoring in 2020-2021
(Note that the scoring varies from season to season).

This is how you score points for your predictions. Note that the points are CUMULATIVE - add together each score that applies.

Points for result prediction
5ptsExact score
2ptsResult correct (win, lose or draw)
1ptGoal difference correct
1ptEach team score predicted correctly

Points for nominated player - league matches (own goals don't count)
2ptsEach goal scored

Points for nominated player - cup matches (own goals don't count)
2ptsEach goal scored


Fairly frequently, league positions and head-to-head cup matches result in a tie on points, and we need to find a way of separating the predictors.

Attendance Error (denoted 'AttErr' in the tables)
This is how far away an attendance prediction was from the actual attendance, and is expressed as a percentage. The lower the AttErr, the better the prediction. It is evaluated by subtracting the actual attendance from the prediction, losing any minus sign (so low predictions are no better than high ones), and then dividing by the actual attendance.
For example, if 2000 fans attend, a prediction of 2500 gives an 'AttErr' of (2500 ‑ 2000) ÷ 2000 x 100% = 25.00% (you get the same for a prediction of 1500).
A table made up of multiple matches shows the mean (average) AttErr, ignoring any games where no attendance prediction was submitted.

Result Error (denoted 'ResErr' in the tables)
This is a measure of how far away a score prediction was from the actual score, and is expressed as a percentage. The lower the ResErr, the better the prediction. It is calculated in rather a complicated way! It is equal to double the error in the margin between the teams plus the error in predicting each team's score. If the GH and GA are the home and away goals from the match result and the prediction is PH goals to PA, then
    ResErr = 2 x |(GH ‑ GA) ‑ (PH ‑ PA)| + |GH ‑ PH| + |GA ‑ PA|,
where |...| represents the modulus of an expression (or, in English, ignore minus signs!). The factor of two is a weighting chosen to make the margin more important than how goal-heavy the game was.
One example for you. The game ends 3-1, and the (brave, and very wrong) prediction was 4-5, so GH=3, GA=1, PH=4, PA=5. Thus
    ResErr = 2 x |(3 ‑ 1) ‑ (4 ‑ 5)| + |3 ‑ 4| + |1 ‑ 5| = 2 x |3| + |‑1| + |‑4| = 6 + 1 + 4 = 11.   Eleven is bad. Perfection is zero.
A table made up of multiple matches shows the mean (average) ResErr.

In 2020-2021, ties are broken by comparing records in the following order - most points scored for predicting the result (Rs), most points for goalscorers (Gl), then lowest mean error in attendance (AttEr).

Other Rules

• The webmaster's word is final, irespective of whether it is fair or just.
• In leagues with a division structure, where predictors have not submitted a prediction during the course of a prediction season, they will be ejected from the league. Where applicable, numbers of predictors in each division will be maintained first by reprieving predictors from relegation. Only if further action is required will extra promotion places be granted.
• Ties in cup competitions are decided by the tie-break criteria above followed by, where applicable, total points scored in that competition up to that point, using the same tie-breaker if necessary.
• Where neither predictor in a knockout stage of a cup has submitted a prediction, neither proceeds to the next round and their place is taken by the highest-scoring loser amongst the other ties.
• In cup competitions, points may be scored in replayed matches during the qualifying phase only. During the knockout phase, results are based on the scheduled tie(s) only.
• Joke entries will be discarded - persistent offenders will be banned.