AI Football Predictions

About 66.6%. Our record starts in May 2021, and since then our NT Apex model has predicted 273,934 matches, 182,463 of them correctly. Each match's pick is graded against the final score once the game ends, and losing picks stay in the record. Our public track record page shows the all-time figures, a month-by-month breakdown and the last few days of results. You can download the full daily history there as a CSV file, plus a spreadsheet of every match. Each market we rate carries its own confidence score out of 10. A win rate is not a profit rate, and past results do not decide the next match.

No. The free predictions are open to anyone: no account, no email address, no payment details. We publish up to 6 of them each day, and the list is locked in when the day starts. Free picks are capped on purpose. In most competitions they stop at 8 out of 10 on our confidence scale, and in the biggest leagues and competitions they stop at 6.5 out of 10, so they are never our highest-rated predictions. The rest of the day's picks are on All Matches, the full list of the matches we rate that day. They stay locked until you subscribe, and they open to everyone once the match ends.

Yes, and you do not need an account. The yesterday section on this page shows whether the previous day's free picks won or lost. If you want every match and not only the free ones, All Matches goes back over the last three days. Finished matches are open to everyone, so you see each prediction next to its final score. Losing picks are never removed. They stay in the daily and monthly numbers on our track record page, and in the public dataset, which we update every day.

Predictions are published before kickoff, usually as much as four days in advance, and the how it works page explains that timing. This page lists three days of free picks one after another — today, tomorrow, and yesterday with every pick graded — and All Matches goes two days further forward, out to three days ahead. Before kickoff a pick can still be refreshed, as its confidence and its odds move. Once a match has finished, its pick is graded against the final score and counted, whether it won or lost.

Each code is the outcome the model picked, in its betting market's shorthand. These are model ratings, not betting advice.

• 1 home win, X draw, 2 away win.
• 1X home win or draw, X2 away win or draw.
• BTTS whether both teams score, or at least one does not.
• O2.5 over 2.5 goals, three or more, U2.5 two or fewer. The 1.5 and 3.5 markets work the same way.
• HS the home team scores, AS the away team scores, HS2+ and AS2+ twice or more.
• H1 the home team wins by two goals or more, H2 the same for the away team.

On a phone or tablet, All Matches shows the full list.

The confidence score shows how strongly our model rates a pick, from 0 to 10. Every match we publish carries a best tip, the one pick our model puts forward, shown with its score. Read the number as a rating, not a probability, and not the chance of a bet winning. A 9 means the data leans strongly one way. A 5 means it barely leans at all. A pick rated 9 or above is what we call a banker, our highest confidence level, not a recommendation. Since the start of 2026 we have published 1,424 banker picks and 1,025 came in, a 72% win rate, counted separately on our track record page. The day's bankers are collected on Bet of the Day.

Not with certainty, no. A model can measure what the data shows. That is not the same as knowing the result. NT Apex is our own AI model. It reads recent form, head-to-head history, expected goals and odds movement, then rates the main markets, such as match result, total goals and both teams to score. The correct score is published without a rating. Each pick and its score are printed exactly as the model produced them, never changed by hand. A red card or a late goal can still decide a match, and none of that is in the data beforehand. Read how the model works, then judge it against our track record.

We cover more than 700 leagues and competitions worldwide. That includes the Premier League, La Liga, Serie A, the Bundesliga and Ligue 1, plus the Champions League, the Europa League and the Conference League. You will also find Major League Soccer, many second divisions and smaller domestic leagues, and national team tournaments such as the World Cup and the European Championship. The league directory puts the biggest competitions at the top and groups the rest by country. Each competition has its own page with recent results and our predictions, plus upcoming fixtures whenever it is in season. What you see on a given day depends on which competitions have matches.

No. A win rate only counts how many predictions were correct. What a bet returns depends on the odds you take and the amount you stake, and both of those choices are yours alone: we do not take bets and we never place or size one for you. Our track record shows win rates, not profit or return figures. A single match can also be decided by chance, and a deflected shot or a red card can undo a strong prediction. The 66.6% win rate measures the model's best tip across 273,934 settled matches. It does not forecast the next match, or your own results. Only bet what you can afford to lose.

No. You cannot place a bet here, and nothing on this site is betting advice. We are not a bookmaker, we accept no wagers, and we are never on the other side of a bet. We publish statistical analysis and probability-based insight: what our NT Apex model considers likely in a match. Alongside it we publish the public track record of how those predictions performed. No prediction is certain, and betting money carries real financial risk. Betting can also be addictive and cause harm. Any decision is yours alone. The service is for adults who are 18 or older, and our terms and conditions set out what we do and do not provide.

No. We do not sell fixed matches, and we have no inside information about any game. Every pick is a prediction from our NT Apex model about a match that has not been played yet. We publish it before kickoff, settle it against the final result, and keep it in the public track record, whether it was right or wrong. A subscription unlocks the rest of our predictions and the tools that sort and combine them. What it cannot buy is a result. Nobody can know a football result in advance, and that includes us.