Football prediction results and record

How the football prediction results are graded

A tip enters the record when its match is within 40 hours of kick-off and still unplayed, with the model chance and the best price at that moment. After full time it is graded from the score.

Match result and goal markets on 90 minutes plus stoppage time, half-time markets on the half-time score, correct score on the exact final.

Postponed, abandoned and cancelled matches make every tip void. The grading runs automatically a few minutes after the feed marks a match finished. And a later official correction to a score is applied by hand when it is noticed or reported.

Eight markets are frozen for every rated match. Match result, double chance, over 1.5, over or under 2.5, both teams to score, correct score, half-time / full-time and first-half goals. That is a deliberately wide net.

A record built only from the picks that happened to win is a sales page. And a record that lets the editor choose which markets to count after the results are in is only slightly better. The daily accumulator tickets are stored with their legs and settled the same way.

Reading the hit rate against the price

The hit rate on its own says little, because the markets have different base rates. Over 1.5 goals lands three times in four whatever the model says, correct score one time in eight.

The number that matters is whether each market's hit rate sits near the chance the model gave its picks, and whether the flat-stake return at the frozen prices is positive. Both are shown per market in the table for any day and in the side cards for the last 30 days.

A market that lands 62% of its picks at an average price of 1.80 returns about 12% on turnover. The same hit rate at 1.55 loses 4%, and the two look identical on a hit-rate-only scoreboard. That is why so many prediction sites publish only that.

Why the record starts when it starts

Football prediction results published here start on the first published matchday, with no entries before it. A backtest on past seasons could be shown, and it would look better than any live record, because a model tuned on the past fits the past.

The honest comparison between prediction sites is a live, dated, graded list with prices, which is what the dated pages hold, one per day, reachable from the day navigation above. Early weeks have few decided tips in some markets and the per-market figures swing accordingly.

The 30-day figures settle as the record grows, and the daily chart shows how quickly.

Anyone comparing this record with another site's should ask the other site the same question: when did it start, and can every day since be opened?

Accumulators and the prediction of the day in the record

The 2, 5 and 10 odds tickets are built once in the morning and stored with their legs and prices. After the last leg finishes, each leg is marked won or lost and the ticket is settled. A postponed leg is void and the ticket is recalculated without it.

Settled tickets appear on their day's results page and on the accumulators page, including the ones that lost, which over any month is most of the 10 odds tickets. The prediction of the day is frozen in the same way and graded as a single.

Its tip appears in the day's list under its market, so it carries no separate scoreboard that could be polished.

Football prediction results today, yesterday and on the daily chart

Football prediction results for today appear as matches finish, with each frozen tip graded from the final score within minutes of the whistle; yesterday's results are the first complete day and open from the date list.

The daily chart plots the hit rate for each of the last 30 days against the number of tips settled, so a 70% day on six tips reads differently from a 55% day on sixty.

A run of bars below 55% means little unless the 30-day figure follows it down, in which case the grading bars for the affected market are reviewed and any change is noted on that market's page.

The methodology page describes how the model is checked. And the per-market cards on the right give the 30-day hit rate for the eight frozen markets so that a weak market is visible without opening a single day.