Form and scoring filters: the team-history family

Five ways to read a team's recent past

A team has won its last five. Every preview column knows it, and so does Pinnacle. Whether that streak is worth anything to you is the question this article ends on. First, the tools that let you see it.

Form filters answer one question from five angles: what has this team actually been doing lately? They sit in the Pro filter tier (see the plans page for what your plan unlocks), and together they form the largest family in the catalog: Result streak, Last N result rate, Scoring streak, Scoring rate and Goals average. Two read results, three read goals. Each one scans a team's match history at the moment a matchup is evaluated and matches or rejects the game based on what it finds.

They are worth learning as a group, because they share most of their machinery: venue scoping, side resolution, sample-size guards, lookback windows. Get those right once and the whole family opens up.

Results: streaks and rates

Result streak matches when a team is on an unbroken run of a chosen result. The result metric is broader than win-or-lose: win, draw, loss, and their negations (non-win, non-draw, non-loss). You set a minimum streak length from 1 to 50, and optionally a maximum. The maximum matters more than it looks. A bare "minimum 3" catches everything from a quiet little run to a freak twelve-game tear; "between 3 and 5" isolates streaks at the stage you may actually want them: established, but not yet the only thing anyone talks about.

Concrete use: find home teams on at least 5 straight wins, then study in a backtest, not in your imagination, whether backing or fading them pays at the prices Pinnacle actually offers. Or flip the metric. A non-loss streak of 8 or more finds sides that are hard to beat even when they do not win, a very different profile from a pure winning machine.

Last N result rate softens the streak into a percentage. Instead of demanding consecutive results, it computes the rate of a result metric over the team's last 3, 5, 10, 20, 50 or 70 matches and checks it against your min and max. A team that went 13 wins, 2 draws and 5 losses over its last 20 shows a 65% win rate, no streak required. Typical use: home teams with a win rate of at least 60% over their last 20, which captures sustained quality that a streak filter loses the moment a single draw interrupts the run.

Goals: three lenses on scoring

Scoring streak is the goals cousin of Result streak. It counts consecutive matches where a goal condition held: goals scored, goals conceded, or the match total, compared against a threshold with "at least" or "under". Scored and conceded thresholds run from 1 to 3; total thresholds are 4 or 5. A team that has conceded at least 1 goal in each of its last 6 matches has a leaky defense the filter can see directly, useful raw material for a totals strategy.

Scoring rate does for goals what Last N result rate does for results: the share of the last 3, 5, 10, 20, 50 or 70 matches meeting the goal condition, with a required minimum and an optional maximum. "Scored at least 2 in 70% of the last 20" describes a genuinely potent attack. "Total of 4 or more goals in at least 60% of the last 10" describes matchups that tend to be open, worth knowing before you touch an over.

Goals average drops thresholds entirely and takes the plain average of goals scored or conceded per match over the last 10, 20, 50 or 70, matched against your range. An average of 2 or more conceded over 20 matches is a stable, hard-to-fake signal of defensive trouble. It is the smoothest of the three goal lenses, since one 5-0 disaster cannot dominate a 20-match average the way it dominates a highlight reel.

The shared machinery

Venue scope. Every form filter can look at all venues, at home matches only, or at road matches only. This is not a cosmetic option. Plenty of teams are effectively two different teams depending on where they play, and for a home matchup, home-only form is often the sharper signal: it measures the exact context the upcoming game will be played in instead of blending it with road performances that tell another story. The price is a thinner sample. A home-only window of 10 reaches much further back in the calendar than an all-venues window of 10.

Which team. The filter's side can be home or away, but also favorite or underdog. Favorite status is resolved from the moneyline: whichever side the market prices shorter. Pick'em matchups, where the line shows no favorite, are excluded; a favorite-side form filter simply will not match them.

Minimum sample size. The rate filters accept an optional minimum history length, and you should treat it as mandatory in spirit. A 100% win rate over 3 matches tells you almost nothing; three results measure luck at least as much as quality. With a minimum sample size set, teams with thin histories (a newly promoted side in its first weeks, a team barely present in your venue-scoped window) are skipped rather than matched on a mirage.

The window trade-off. Short windows of 3 or 5 are reactive: they see the new-manager bounce and the injury crisis quickly, and they see noise just as quickly. Long windows of 50 or 70 are stable but stale; a 70-match lookback can reach into a previous season, a previous squad, a previous identity. For most uses 10 or 20 is the honest compromise, and a 20-match window with a sample guard beats a 5-match window without one nearly every time.

The caveat that pays for the whole article

Recent form is the most public information in sports. Every casual bettor, every preview column, every model knows who won its last five. Pinnacle prices that knowledge into the line: a team on a seven-match winning streak is quoted accordingly, and backing it blindly means paying full price for a fact everyone owns. Backtest a bare "5+ wins" filter and you will usually watch the market's efficiency stare back at you.

Form filters earn their keep in combinations, where they sharpen a situation instead of defining it. A form condition plus an Odds range price window plus a venue scope describes something far more specific than the headline streak: say, road underdogs at generous prices against a home side whose home-only form is worse than its overall record suggests. Specificity is where edges hide.

Two warnings before you stack, though. Each form filter costs 2 to 4 resource points against your plan's cap, so a five-condition form tower is expensive before it has proven anything. And every added condition shrinks your backtest sample: pile on form clauses until the backtest looks perfect and you have probably carved out an overfit corner that exists only in the past. Two or three deliberate conditions, tested honestly, beat six clever ones.

Where to go next

• Filters overview: the full catalog, tiers and resource costs, so you can place the form family in context.
• Variance and sample size: why small windows lie, in filters and in track records alike.
• Head-to-head and scheduling filters: the other Pro-tier history filters: direct meetings, rest and schedule congestion.