ANALYSIS GUIDE
Every number on the site comes from the same calculation: the teams' attack and defence strength, the expected goals that come from it and the probability of every scoreline. Alongside there is an AI match summary. Below is what each number is and how to read it.
THE BASIS OF THE CALCULATION
Expected goals (for a match)
What it is: How many goals we expect each team to score in this match on average (e.g. home 1.7, away 1.1). In statistics it is written λ (lambda). Every percentage on the site is calculated from these two numbers.
If both numbers are high, expect a high-scoring match. If the gap is large (1.9 to 0.8) the favourite is clear; if it is small (1.3 to 1.2) the match is balanced.
A team's attack and defence strength
What it is: In the 5 big European leagues, strength worked out from the quality of chances a team created and conceded in its last 38 league matches (75%) and the real goals it scored and conceded (25%). The main driver of expected goals.
Attack 1.20 = creates 20% more chances than the league average; defence 0.80 = concedes 20% fewer chances than average. In the Süper Lig and cups there is no chance data, so only goals are used.
xG (chance quality)PRO
What it is: Expected Goals: the probability that a shot becomes a goal. A 0.30 shot means that of 10 similar shots, 3 are goals. The sum over all of a team's shots shows how many clear chances it created.
If the value of the chances is higher than the goals scored, the team is unlucky or a poor finisher; that usually corrects itself. If lower, the team is very efficient or lucky; that usually doesn't last.
MATCH PROBABILITIES
We give 13 different probabilities for each match. The small number next to a percentage is the same probability written as odds: 1 ÷ probability (e.g. 50% → 2.00). It is not a bookmaker's odds.
Match resultFREE
What it is: Home win · draw · away win. Also written 1 · X · 2. The three add up to 100%.
The result with the highest percentage is the calculation's favourite; if all three are close, the match is balanced.
Total goals (3+?)FREE
What it is: Will the match have 3 or more goals in total, or at most 2? Also known as “over / under 2.5”.
If the two teams' expected goals add up high, the 3+ probability rises. If both score and concede few, “at most 2” stands out.
Both teams to score?FREE
What it is: Yes = both teams score at least once · No = at least one doesn't. Also known as “BTTS yes / no”.
If both teams' expected goals are high, “yes” is likely. If one side's attack is very weak, “no” stands out.
One of two results (doesn't lose)FREE
What it is: Two of the three results together: “home doesn't lose” (win or draw), “away doesn't lose”, “no draw”. Also known as “double chance”.
Because it covers two results at once, its percentage is always higher than that of a single result.
Handicap result
What it is: One of the teams is given a virtual goal advantage or disadvantage from the start (e.g. −1). Also known as “Asian handicap”.
−1 requires the favourite to win by at least 2. It gives an idea of how big a margin the favourite could win by.
Half-time / full-time · Correct score · Goal range
What it is: How the first half and the match finish together, exact-score probabilities and total-goal ranges (0–1, 2–3, 4+).
All come from the same scoreline probabilities and never contradict each other. Because there are many options, each one's percentage is low.
TRUST & TRANSPARENCY
Decision clarity (1–10)FREE
What it is: The favourite's percentage in this match is compared with matches from the league's last 8 months. 10 = the league's clearest 10% of matches (e.g. favourite 62%), 1 = the most balanced 10% (e.g. 36%-33%-31%).
Clarity ≠ accuracy. 10 does NOT mean “it will certainly come off” — only that “in this match the calculation favours one side more clearly than in most matches in the league”. The Past accuracy below shows real accuracy.
Past accuracyFREE
What it is: How often the favourite won in past matches in the same clarity band.
“18 matches · 64%” = there were 18 matches in this band in the league and the favourite won 64% of them. With few matches the result can swing; with fewer than 10 it is not shown at all.
Pick explanation badgeFREE
What it is: A short reason why the featured pick was chosen: expected-goal gap, strength-rating gap, recent matches, past accuracy.
Summarises the logic behind the percentages; shows the calculation is not a “black box”.
Error score (Brier score)
What it is: How close were our percentages to the real results? On the Track record and Past results pages.
Low = good, 0 = perfect. Saying 33% for every result on the match result scores 0.667. The real comparison is the league's historical average; current values are in the Track record every week. If it rises, the calculation is getting worse.
Confidently-wrong score (log-loss)
What it is: Similar to the error score but punishes being very confident and wrong more heavily. Low = good.
If the calculation says “90%” and it doesn't happen, this score jumps. It catches overconfidence.
Consistency (calibration)
What it is: “When we say 70%, does it really happen in about 7 of 10 matches?” Gap = what happened − the percentage we gave.
A gap near zero is good. Negative means we are too bold (it happened less often); positive means too cautious (it happened more often). A gap larger than 5 points deserves attention.
COMPARISON WITH BETTING ODDS
On Pro, bookmakers' odds are shown next to our percentages. This is for information only; it is not betting advice.
GapPRO
What it is: A comparison of our percentage with the betting odds: odds × our probability − 1.
+6% = our calculation gives this result a slightly higher probability than the odds imply. It does NOT mean profit; both the calculation and the market can be wrong.
Model odds and betting odds
What it is: Model odds = 1 ÷ our probability. Betting odds are the odds offered by bookmakers.
If our probability is 50%, model odds are 2.00. If the betting odds are also around 2.00, the two are saying the same thing.
TEAM INDICATORS
Form / Streak
What it is: Results of recent matches (W/D/L) and winning or losing streaks.
Shows the team's trajectory. A long streak points to a team in or out of form.
Consistency
What it is: How much the number of goals a team scores varies from match to match.
Low = every match is similar, consistent. High = it may score a lot one match and none the next; harder to predict.
Fatigue
What it is: Rest time between matches and fixture congestion.
Little rest can lower performance. It matters in cup and European weeks.
Home / away
What it is: A team's performance at home and away, separately.
Some teams are completely different at home; the overall average can hide this.
Goal-minute heatmap
What it is: In which 15-minute period of the match goals are scored.
Shows teams that score or concede late.
Comebacks
What it is: How often a team turns around or rescues a match after trailing at half-time.
If high, the team doesn't give up until the end.
Referee tendencyFREE
What it is: The referee's tendency to give cards and penalties.
A strict referee can raise the number of cards; a penalty tendency affects the goal expectation.
Card / corner average
What it is: Average cards and corners over the last 10 matches.
Gives an idea of how rough or attacking the match could be.
League table
What it is: The two teams' current standings.
The overall picture and motivation (relegation, title race).
AI SUMMARY
AI match summaryPRO
What it is: A newspaper-preview-style summary written by AI after scanning current news on the internet: injuries, motivation, tactics, key players, possible scenarios.
Tells what the numbers don't — adds current information (news, line-ups) that the calculation can't see. It complements the percentages; it doesn't change them.
Data source label
What it is: Where each statistic comes from (football-data, Understat, API-Football).
If there is a “Warning” label, the data is missing or estimated — be cautious in that match.