How Ratings Work

1307scout shows every statistic we can, so you can look at the numbers and draw your own conclusions. Ratings and the other derived stats on this page are there to supplement those numbers, not replace them. A grade is a quick summary to help you decide where to look, and the stats behind it are always shown next to it.

Two ratings, never blended

Every team can have two separate grades. Neither one knows the other exists.

	Scouting rating	Official rating
Built from	Match and pit entries from our own scouts	Match results synced from FRC Events and The Blue Alliance
Measures	What this robot did on its own	About how many points this team adds to its alliance per match (OPR)
Seasons	This season's form only (2026 REBUILT)	Every synced match
Updates	As soon as an entry is saved	Within about a minute of a sync

They're kept apart so a team is rated on whichever data it actually has. A team with a full official record but no scouting coverage still gets an official grade, and when the two grades disagree, that's useful to see.

Letter grades

Both ratings turn into a letter the same way. Each team gets a score (0 to 100 for scouting, points per match for official), then the grade comes from the share of rated teams with the same score or lower. A grade is a ranking among the teams in our database, not a measure against all of FRC.

Grade	Share of rated teams at or below this score	In other words
S	90% or more	Top 10%
A	75% to 90%	Next 15%
B	50% to 75%	Next 25%
C	25% to 50%	Next 25%
D	10% to 25%	Next 15%
F	Under 10%	Bottom 10%
NR	Not rated: fewer than 3 matches of that kind of data (for the official rating, 3 qualification matches)
T	Test team: tagged by staff and left out of the grading

The scouting rating

A team's score is a weighted blend of seven parts, each from 0 to 100. These are the bars in the Rating breakdown on the Team Database. "Ranked" means the share of scouted teams at or below this team on that measure, so 80 means it's level with or ahead of 80% of them.

Weight	Part	How it's measured
32	FUEL scoring	Average FUEL points per match, ranked. Matches where the scout marked the role as Defense or Feeder are left out, so a good defender isn't graded as a bad scorer.
16	Climbing	Average TOWER points per match (auto and teleop), ranked.
16	Reliability	1 − (disabled rate + 0.6 × broke-down rate + 0.4 × stuck rate), never below 0. Not ranked.
11	Shooting efficiency	The average of two things: FUEL per volley (ranked) and scouted accuracy, where 1 to 5 maps to 0 to 100.
11	Robot capability	The team's latest pit entry, scored in the table below. Not ranked.
9	Autonomous	80% average auto points (ranked) and 20% how often it left the starting zone.
5	Fouls avoided	1 − foul or card rate. Not ranked.

Points used

From Table 6-4 of the 2026 game manual. Avg pts / match on a team's card is FUEL plus TOWER points, across all of its matches.

Action	Points
FUEL in an active HUB (auto or teleop)	1 each
TOWER L1 in auto	15
TOWER L1 / L2 / L3 in teleop	10 / 20 / 30

Robot capability scoring

Each pit question answered scores from 0 to 1, and the answers are averaged. Questions left blank are skipped. The robot photo and notes aren't scored.

Question	Score
Highest TOWER level	None 0, L1 0.33, L2 0.67, L3 1
Fits under a TRENCH	No 0, Yes 1
Ground intake	None 0, Front or Side 0.7, Front and side 1
Shooter type	Fixed flywheel 0.4, Other 0.5, Adjustable hood 0.7, Turret 1
Drivetrain	Mecanum or Other 0.5, Tank 0.65, West Coast 0.8, Swerve 1
Max FUEL held	Capacity ÷ 30, up to 1
Max shooting range	Range in feet ÷ 20, up to 1
Uses AprilTag vision	No 0, Yes 1

When data is missing

A whole part has no data: it's skipped and the other weights are rescaled, so a team isn't punished for, say, having no pit entry.
A field is blank in a match entry: FUEL and TOWER count as 0, and an unanswered problem such as "disabled" or "foul" counts as no.
Only Defense or Feeder matches: FUEL scoring has no data, so it's skipped.

Example. A team scores FUEL 80, Climbing 60, Reliability 90, Efficiency 70, Capability 40, Autonomous 50 and Fouls avoided 100.

(32×80 + 16×60 + 16×90 + 11×70 + 11×40 + 9×50 + 5×100) ÷ 100 = 71.2

With no pit entry, Capability is skipped and the remaining weights add up to 89:

(32×80 + 16×60 + 16×90 + 11×70 + 9×50 + 5×100) ÷ 89 = 75.1

The official rating

One number, from results only: the team's OPR, which estimates how many points it adds to its alliance each match. An OPR of 64 means "about 64 points per match, on top of its partners". It's worked out from the qualification matches at each event, then averaged across the team's events:

Score = average of the team's event OPRs, each weighted by matches played × event tier

Results against a tougher field say more about a team, so each event also counts by the tier staff set on it:

Event tier	Match weight
FIRST Championship	2.0×
District Championship	1.5×
Qualifier	1.0×
Off-season	0.75×

Example. A team had an OPR of 60 over 12 matches at a qualifier, and 80 over 10 matches at a District Championship.

Score = (60×12×1 + 80×10×1.5) ÷ (12×1 + 10×1.5) = 1920 ÷ 27 = 71.1 points per match

Playoff matches aren't used, and neither is win rate. Both depend heavily on partners: a robot picked onto a dominant alliance wins and scores a lot without doing much itself. Record and average alliance score are still shown next to the rating as plain stats.

Why OPR. This replaced an earlier score (60% average alliance score + 40% win rate) after both were tested on the full 2026 season. Built from matches before the FIRST Championship, OPR picked the winner of 76% of Championship qualification matches, against 71% for the old score, and its grades matched an independent rating (Statbotics EPA) far more often. Adjusting OPR for how high each event or week scored was tested too, and made it worse.

Reading ratings carefully

Grades are relative. They depend on which teams are in the database, so a grade can change when other teams' data changes, even if this team hasn't played. Early in a season, with few rated teams, they mean very little.
Three matches is a low bar. A rated team may still have very little data behind it, so check the match counts.
A match scouted twice counts once. When several scouts enter the same team, event and match number, the rating averages their entries into one match, and if they disagree on something like "disabled", it counts as half. Entries with no match number can't be paired, so each counts on its own. Tables that say Entries, and the stats on the Alliance View, still use every entry, so they can differ slightly from the rating.
Scouted numbers are estimates. FUEL counts are approximate, and accuracy is a scout's judgment.
OPR separates teams from partners, but not perfectly. With few matches it can't tell a strong robot from a lucky schedule, and it can't see defense. Check the match count beside the grade.
Capability scores describe a robot's design, not how well it performs. A swerve drive scores higher than tank whether or not it's driven well.
Share codes are ranked differently. A guest's grades are worked out among only the teams in that share code, so they can differ from the Team Database.
Test teams don't affect real teams. They're ranked in their own group at every step, so a dummy team can't move anyone's score or grade. Teams with fewer than 3 matches are not graded, but they are still counted when each part is ranked.

Alliance grades and win chance

The Alliance View adds two estimates for a full red and blue match-up. Our Matches shows the same win chance for each upcoming match.

Alliance grade

Each alliance gets a Scouting grade and an Official grade, kept separate like team grades. The three teams' scores are added together, then compared against every possible 3-team alliance from the rated teams, using the same S to F cutoffs as above. An S alliance has a combined score in the top 10% of all alliances those teams could form. All three teams must be rated, otherwise the grade is NR, and a test team counts as not rated.

Win chance

This estimates how likely each alliance is to outscore the other, from two separate sources. A source is only used if all six teams have at least 3 matches in it.

	Scouted	Official
Expected score	The three robots' Avg pts / match, added up	The three teams' OPRs (official rating scores), added up
Uncertainty	The three robots' variances, added up	The average of the three teams' alliance-score variances

A team's variance measures how far its scores swing from match to match, and is increased by 1/n for a team with n matches, so less data means less certainty. It's never treated as less than a 5 point swing. Each source then turns the expected gap into a chance:

Chance = 1 ÷ (1 + e^{−1.702 × gap ÷ √(red uncertainty + blue uncertainty)})

The headline number is the average of the sources available, and it's kept between 5% and 95%. Both sources are shown beside it, because when they disagree, that's worth knowing.

Example. Red is expected to lead by 20 points, and the two alliances' uncertainty adds up to 900, a typical swing of 30 points.

1 ÷ (1 + e^{−1.702 × 20 ÷ 30}) = 76% for red, leaving 100 − 76 = 24% for blue

Treat this as a rough guide:

It hasn't been tested against real results yet, so a 70% isn't proven to mean winning 7 times out of 10.
It only knows past scores. It can't see defense, strategy, match-ups, penalties or a robot that was just repaired.
It treats robots as independent, so it misses alliances that work especially well or badly together.
Scouted points are estimates and leave out penalty points, and OPR is still an estimate from alliance totals.

Playoff Selection ranking

The Playoff Selection page ranks every team at the event that hasn't joined an alliance. It adds no new score: the two ratings stay separate, and the raw stats sit beside them.

Default order. Teams are listed by scouting rating score, highest first. Ties, and teams without a scouting rating, are then ordered by official rating score, then by team number.
Any column can rank the list. Click a heading to sort by it, starting with the best value, and click again to reverse it. Teams with no value for that column go last and show — for their rank.
All season or This event. These buttons change the scouted and official stats. Ratings always cover the whole season, and robot details come from the team's newest pit entry.
Who's listed. A team appears if it's on the event's match schedule or was scouted there. Test teams are left out.
Alliance points. The number on each alliance adds up the season Avg pts / match of its first three teams, and leaves out any team with no scouted data.

Event rankings

Pick an event at the top of the Team Database to see its qualification standings. They're worked out from the synced FRC Events / TBA qualification matches, since the sync doesn't bring in the official ranking table.

Order. Teams are ranked by average ranking points per played match, when every played match reports them. If some don't, teams are ranked by record instead, with a tie counting as half a win. Teams level on that are then ordered by average alliance score, then team number.
Who's listed. Every team on the event's qualification schedule. Teams that haven't played yet go last with no rank.
Other columns. Total RP, W-L-T, matches played (with matches still to come in brackets), average alliance score and margin, the team's OPR at the event with its rank there, and how many scouting entries we have for it at the event.
Limits. The official tiebreakers change every season and aren't copied here, and surrogate matches and disqualifications aren't known. Close places can differ from the official table.

Estimated contribution (OPR)

Official scores cover a whole alliance. OPR (offensive power rating) estimates how much of that one team adds. It treats each alliance's number as the sum of its three robots' numbers, then finds the per-team values that best fit every qualification match at the event.

Every score field gets one. Each number FRC Events or TBA reports for an alliance, such as FUEL counts or foul points, gets its own estimate, alongside the total score.
Steady early in an event. With only a few matches there's no unique answer, so each estimate is pulled gently toward the event average. The pull is worth about one match, and it fades as matches are played.
Needs 3 matches. A team with fewer qualification matches at the event gets no estimate.
Per event, then the official rating. The Team Database shows one column per event, with the team's rank there, and Playoff Selection shows the current event's total, plus its AUTO and teleop FUEL counts (teleop is the total FUEL count minus AUTO when FRC doesn't report teleop on its own). The official rating is the total-score OPR averaged over the team's events, as described above. It never touches the scouting rating.
Limits. It can't tell a strong robot from a lucky schedule after only a few matches. Defense and partners still affect it.
Climbing needs no estimate. FRC reports each robot's TOWER level on its own, so Playoff Selection's official Auto climb and Climb rate are plain counts of the matches it climbed, with its best endgame level. They follow the All season / This event buttons.

Vision estimates

A program on a team laptop reads match video and saves an estimate for each robot. It reads the scoreboard and the clock from the stream overlay, follows each robot's bumpers on the wide shot, and tells robots apart by reading their bumper numbers. These numbers are shown in their own Vision section and never enter a rating.

Bumper numbers are often too small or blurry to read on a broadcast. When a robot's number can't be read, the program compares how the robot looks with how that team's robot looked in other matches of the season where its number was read clearly. Matches from the same event count first, since robots change between events. If that doesn't settle it either, a robot is only identified when it's the one team on its alliance left.

Stat	How it's worked out
Estimated FUEL scored	Every rise in the alliance's score is split between its robots. A robot in its own alliance zone just before counts most, one elsewhere counts a little, and one out of view counts in between. The end of auto and the end of the match are skipped, since TOWER points land then. The split is then scaled to the official FUEL points: the alliance score minus each robot's official TOWER points and the foul points.
Left starting zone (auto)	The robot crossed its alliance-zone line during auto.
Distance driven	Second-by-second movement on the field, skipping gaps and impossible jumps.
Time in each zone	Seconds the robot was tracked in its own zone, the neutral zone and the opponent's zone.
Time stopped	Seconds in teleop, before endgame, where the robot moved less than 6 inches in a second. A long stop can mean a disabled robot.
Close contacts	Separate times the robot came within 40 inches of an opponent.
Identified by number, by look	The share of the robot's tracked time it was identified by reading its bumper number, and by how it looks. The rest was by being the only team left.
Confidence	The share of the match the robot was tracked, cut by up to half when it wasn't identified. Being identified by look counts a little over half as much as by bumper number. Under 50% is shown as low.

Match Flags checks

The Match Flags page runs these checks on each played qualification match. Test teams are left out.

Flag	Level	When it's raised
Team isn't in this match	High	An entry's team isn't in that match's line-up, or the match is past the end of the schedule.
TOWER doesn't match	High	A scouted auto or endgame TOWER level differs from the official result for that robot.
Not scouted	Medium, or Low with vision	No one scouted the robot. It's Low when vision covered it with 50% confidence or more.
Scouts disagree	Medium	Two entries for one robot differ by more than 10 FUEL and 30% of their average, or give different TOWER levels.
Alliance FUEL doesn't add up	Medium	All three robots are scouted, and their FUEL total is more than 20 points and 35% away from the official FUEL points.
FRC Events and TBA disagree	Medium	The two sources have different scores for the match.
Vision saw the robot stopped	Medium	Vision saw a stop of 20 seconds or more, and no scout marked the robot disabled or broken down.
FUEL far from usual	Low	The entry is more than 15 FUEL and 3 spreads from the team's median at the event, with 3 or more other matches. The spread is 1.48 times the median distance from the median, which matches a standard deviation when there are no outliers.
Vision disagrees	Low	Vision's FUEL estimate differs from the scouts by more than 15 FUEL and more than half of the larger number, or it disagrees on leaving the starting zone.
Vision scoreboard read is off	Low	Vision's final score differs from the official score by more than 5, both with and without foul points (the broadcast scoreboard usually doesn't show those until after the match), so its numbers for that match may be off. Not checked when the video ends before the match does.

Who to scout next lists the next three unplayed qualification matches. It highlights the three robots in each with the smallest share of their played matches scouted by a person.

Other derived stats

These appear on the Team Database and Alliance View. They use plain, unweighted numbers, so they won't always match the rating's inputs.

Stat	How it's calculated
Win %	Wins ÷ matches played. A tie counts as played but not as a win, and matches aren't weighted by event tier.
Qualification / Playoff win %	The same, for just those matches. The number in brackets is the match count.
Avg score margin	Average of alliance score − opponent score, across wins and losses.
Score swing	Standard deviation of alliance score, shown from 3 matches. Lower means steadier.
Red / Blue / Station win %	Win % for just the matches on that alliance color or at that driver station.
Biggest win / Biggest loss	The largest positive and negative score margin.
Longest win streak	Most wins in a row, by event date and match order. Ties and losses end a streak.
Last 5	The last five results, by event date and match order.
Teleop FUEL trend	Average of the last 3 entries minus the average of all earlier ones, shown from 5 entries.
Scouting / Official rank	Position by score among rated teams, leaving out test teams. Teams level on score share a rank.
All scouted stats	For every form field: average, low and high for numbers, Yes % for yes/no answers, and each answer's share for choices. Blank answers aren't counted.
Score breakdown	The same summary for everything FRC Events or TBA report per match. These are alliance totals unless marked "this robot".
Estimated contribution (OPR)	One team's share of each alliance number, explained above.
Climb rate	Share of entries with a teleop TOWER level other than None.
Usual role / Main FUEL source	The most common answer, with how often it was chosen.
Combined scouted pts	On the Alliance View, the sum of each team's Avg pts / match.
OPR (pts added / match)	On the Alliance View and Team Database, the team's official rating score, explained above.

The calculations live in team_ratings() and team_official_ratings() in SQL/rebuild_all.sql, and the Team Database applies the final grading. OPR is in js/opr.js, the vision estimates in the vision/ worker, and the flag checks in flags.html. If they change, this page should change with them.