How Couchcall works
What this is
Couchcall answers one question for people with no team in the game: which game is worth your evening? Every day, for every league we cover, three things: picks (upcoming games ranked by the Couchcall score), the hipsters’ choice (the game the main list ignores, with the reason), and verdicts on finished games that never reveal a result.
The Couchcall score
0 to 100, deterministic. Closeness comes from the betting market, with the bookmakers’ margin removed (in soccer the draw is set aside and the two sides compared, in hockey a regulation draw counts as half a win each because overtime follows). Before a market opens on a soccer or hockey game, closeness comes from the two teams’ points per game and the line says so. Quality is the strength of both teams with the weaker one weighted. Stakes are rankings, table position and what the game decides. Rivalry is a curated table per league. Pace is the expected total, form is the last five results. The number is our estimate of how much a neutral will get from sitting down for this one.
A game becomes a pick at 40. Below that we show the best available and point you to the slate.
The number is six weighted terms, the same in every league:
| Term | Weight | What it is |
|---|---|---|
| closeness × quality | 55% | the square root of closeness times quality, so a one-sided game between strong teams and a coin flip between weak ones both land in the middle |
| stakes | 17% | what the result decides |
| rivalry | 10% | a curated table, weighted |
| pace | 8% | the expected total |
| form | 5% | the last five results |
| upset spice | 5% | a game where a team with something at stake is not a clear favourite |
The method by sport
The inputs mean the same thing everywhere, but each sport reads them differently: a point spread in American football, a three-way market with the draw in soccer, overtime in hockey.
American footballfigures from NFL; every league page lists its own
Closeness. The win probability comes from the point spread: final margins land within about 13.5 points of the line in two games out of three, and that turns the line into a probability. When there is a moneyline instead, its two prices are used with the bookmaker's margin removed. Before any line exists, the difference between the two teams' power ratings stands in for the spread, with home field worth 2.5 points. Closeness is 4p(1−p). A coin flip scores 1, a 90% favourite 0.36.
Quality. Each team's power rating is mapped onto 0 to 1 between −10 and 10, roughly the weakest and strongest sides in the NFL. Quality is 60% the average of the two and 40% the weaker team, so a mismatch is pulled down by its weak link.
Stakes. Stakes are 45% the two teams' power-rating ranks in tiers (top four, eight, twelve, sixteen), 35% their playoff probability, 10% a division or conference game, and 10% how deep into the season it is.
Pace. Pace is the over/under total mapped from 40 points (0) to 70 (1); with no line, 0.5. Form is the share of points from each team's last five, averaged. Upset spice is 1 when a top-ten team by power rating is favoured but priced under 75%, half when the underdog is top ten, else 0.
Verdict. After the game the verdict comes from the win-probability chart: how much it swung, how often the lead changed, and how late it was still in doubt, with the fourth quarter weighted up. That excitement number earns one to five stars: one star, plus one at 3, 5 and 7, plus one for a comeback from under 10% or an upset, plus one for overtime or a close finish. Four or five stars is worth it, three is maybe, one or two is skip.
Without a chart, a cautious number comes from a published excitement index or, failing that, the final score. The verdict then carries the mark score only.
Soccerfigures from Premier League; every league page lists its own
Closeness. The draw is set aside and the two sides are compared: closeness is 4p(1−p) on the probability that the home side beats the away side, from a consensus of bookmaker prices with the margin removed. A coin flip scores 1, a 90% favourite 0.36. An earlier version measured uncertainty across home, draw and away; because a draw is always fairly likely, nearly every match looked equally open, so it ranked nothing. Before a market opens, the probability comes from the two teams' points per game, pulled towards the league average early in the season, with home advantage worth roughly 60/40, and the game page says the number is a rating estimate rather than a price.
Quality. Each club's Elo-style strength rating is mapped onto 0 to 1 between 1300, roughly a weak second-tier side in one of Europe's big leagues, and 1700, a contender in one of its top flights. Quality is 60% the average of the two and 40% the weaker club, so a mismatch is pulled down by its weak link. A club without a rating is treated as middling.
Stakes. Each position in the table carries a weight for what it decides (title, Champions League places, relegation): first place weighs 1.0; places 2–5 weigh 0.8; places 18–20 weigh 0.8; the three places either side of those weigh 0.4, the rest nothing. Stakes are 50% the average of the two weights, 30% their product (so it rises when both sides have something to lose) and 20% how far the season has run. A promotion or relegation play-off scores at least 0.9, because it decides a season on its own.
Pace. Pace is the expected total, from the over/under line when there is one, else from the two teams' scoring and conceding rates, mapped from 2 goals (0) to 3.5 (1). With neither, pace is 0.4. Form is the share of points from each team's last five, averaged. Upset spice is 1 when a team whose table place decides something (zone weight 0.8 or more) is priced under 45% to win, read from the market only, never from the rating estimate.
Verdict. After the game the verdict comes from the goal timeline, because we have no win-probability series for the Premier League. Each goal adds 0.9, each lead change 1.5, each equaliser 1.0, each goal from the 80th minute 0.8, each stoppage-time goal 0.6, each goal of a comeback 1.5, each red card 0.4, and a one-sided finish takes one away. That number earns stars at 4, 6 and 8. One more star comes from a two-goal comeback, an upset (a favourite priced at 60% or more lost) or a lead change, and one more from a goal from the 85th minute in a game decided by one goal or level. Four or five stars is worth it, three is maybe, one or two is skip. A one-sided finish caps the verdict at maybe, so a late flurry cannot turn a rout into a recommendation.
Hockeyfigures from Liiga; every league page lists its own
Closeness. Hockey is judged home against away with the regulation draw split in half, because overtime follows and is close to a coin flip. The probability is the consensus of several bookmakers with their margin removed. Before a line is posted it comes from the two teams' points per game, pulled towards the league average early in the season, with home ice worth about four percentage points of win probability. Closeness is 4p(1−p) on that two-way number. A coin flip scores 1, a 90% favourite 0.36.
Quality. Strength is points per game in the league table, mapped onto 0 to 1 up to 2.4, and pulled towards the league average in the first weeks, so three wins from three do not read as dominance. Quality is 60% the average of the two and 40% the weaker team.
Stakes. Each position in the table carries a weight for what it decides (playoff, relegation): places 1–6 weigh 1; places 7–10 weigh 0.5; places 17–17 weigh 0.8. Stakes are 50% the average of the two weights, 30% their product (so it rises when both sides have something to lose) and 20% how far the season has run. A promotion or relegation play-off scores at least 0.9, because it decides a season on its own.
Pace. Pace is the expected total, from the over/under line when there is one, else from the two teams' scoring and conceding rates, mapped from 4.5 goals (0) to 7.5 (1). With neither, pace is 0.4. Form is the share of points from each team's last five, averaged. Upset spice is 1 when a team whose table place decides something (zone weight 0.8 or more) is priced under 45% to win, read from the market only, never from the rating estimate.
Verdict. After the game the verdict comes from the goal timeline, because we have no win-probability series for Liiga. Each goal adds 0.35, each lead change 1.5, each equaliser 1.0, each goal from the 55th minute 0.8, each overtime goal 0.6, each goal of a comeback 1.5, and a one-sided finish takes one away. Overtime adds 1.0 and a shootout 0.8. That number earns stars at 5.5, 7.5 and 9.5. One more star comes from a two-goal comeback, an upset (a favourite priced at 60% or more lost) or two lead changes, and one more from a goal from the 55th minute or overtime in a game decided by one goal or level. Four or five stars is worth it, three is maybe, one or two is skip. A one-sided finish caps the verdict at maybe, so a late flurry cannot turn a rout into a recommendation.
Verdicts
Finished games are rated worth it, maybe or skip from how the game moved: the swing of the win-probability chart where one exists, the goal timeline where it does not. Verdicts are listed in start order and never ranked against each other. They give no score, no winner, no margin and no overtime: nothing that gives away what happened. Game pages hide the result behind a control you choose to open.
Sources
Fixtures, results and tables come from the leagues’ own feeds, from open football-data projects and from public statistics sites. Each league has its own adapter and some have a fallback, so one feed failing stops one league, not the site. The mix is open feeds and paid APIs, and it is still changing. Prices are a consensus of several bookmakers, shown as probabilities with the margin removed and never linked. Strength comes from public power and Elo ratings and from the league tables themselves. The kinds of sources, and how to choose among them, are on the build-your-own page. Thank you to the maintainers of the open ones.
Some verdicts rest on less: SHL goals are known only by period, and a Veikkausliiga match whose goal record is missing or does not add up is judged from the score alone and says so.
TV listings
Pick your market at the top of any page: Finland, Sweden, Germany, the UK, the US, or Other for the league’s own site. The choice stays in this browser; on a first visit we guess it from your time zone. Without scripts every market shows in one line.
Outside the US a listing is who holds the rights this season, from a table we keep by hand: league by league and market by market, the service, whether it shows every game or a selection, and how sure we are. UK broadcasters may not show Saturday 3pm kickoffs live, and the game page says so. In the US it is the network listed for that game in the published schedule. Each game page has the whole table under Where to watch.
Every row carries a review date. Once a month a check compares the table with the leagues’ own broadcaster lists and with Wikipedia’s, and a person applies the changes. Corrections sent through the corrections form are reviewed monthly too.
By league
Each league has its own page with what the score is made of there this season, the inputs it runs on and its verdict rules: College football, NFL, 2. Bundesliga, Championship, Premier League, Champions League, Bundesliga, La Liga, Veikkausliiga, NHL, SHL, Liiga. And how to build this yourself, with the kinds of sources to use and an agent skill on top.
Which parts matter to you
Not every neutral watches for the same thing. What kind of fan are you? maps seven ways of watching, from the xG nerd to the one who wants to see people crash into each other, to the parts of the score that serve each.
Conventions
American leagues read away at home with away-home scores; European leagues read home v away with home-away scores. Times are shown in each league’s own zone, and in yours when scripts are on.
Couchcall is made in Helsinki.