Understanding OPR, DPR, and CCWM

The idea behind OPR

OPR (Offensive Power Rating) estimates how many points a team contributes to its alliance on average, using only final match scores and the schedule. Its key assumption (from The Blue Alliance’s Math Behind OPR): an alliance’s score is a linear combination of its three members’ individual contributions. So if teams A, B, and C scored 50 together, that becomes one equation A + B + C = 50.

Across a whole event you get many such equations, usually more equations (matches) than unknowns (teams). That is an overdetermined system with no exact solution, so OPR is computed with the least-squares method , which finds the per-team values that minimize the total squared error across all matches. More matches give the solver more equations to work with.

DPR and CCWM

Two relatives are computed the same way:

• DPR (Defensive Power Rating) runs the same math as OPR but on the opposing alliance’s score, so it estimates how many of the opponents’ points get attributed to a team. That means lower DPR is better (the team is associated with holding opponents to fewer points) and a high DPR is bad — the opposite of the intuition the name suggests. Treat it with heavy skepticism regardless: DPR is a notoriously weak defense proxy that mostly reflects who a team happened to be scheduled against and statistical noise, not real defensive skill. Judge actual defense by scouting, not by DPR.
• CCWM (Calculated Contribution to Winning Margin) equals OPR minus DPR , an estimate of net contribution to the score difference (how much a team helps you outscore the other alliance).

Strengths

• Free and automatic. TBA computes it for every event with zero scouting effort.
• Year-agnostic. Because it only needs scores, OPR works for almost any season.
• A useful first cut. A team’s OPR is a quick estimate of offensive strength and a sanity check on your scouting.

Serious limitations

The Blue Alliance’s own OPR and You article is blunt about one core weakness, and the wider FRC community has documented others:

• Linearity breaks down. “The more the actual game deviates from the linear model, the less accurate OPR usually is.” Many games have non-linear scoring (bonuses unlocked at thresholds, or actions that require all three robots), which violates OPR’s core assumption, so its predictive accuracy varies a lot year to year.
• It can mis-credit. When robots interfere with each other or strong robots mostly play with strong partners, the math can spread credit in misleading ways, especially with few matches.
• It ignores defense. OPR is built purely from points scored, so a robot playing aggressive legal defense (which can suppress an opponent’s score substantially) gets no credit in OPR. This blind spot is one of the most-cited reasons OPR cannot stand alone.

How to use it well

The community rule of thumb is exact: per OPR and You , “OPR can always supplement, but never replace, proper scouting.” Practical guidance:

• Use OPR to rank-order offensive output quickly and to flag teams your scouting might have missed.
• Before trusting it, remember the two assumptions baked into OPR: that scoring is roughly linear (each extra game piece is worth about the same), and that team contributions are independent — meaning a team scores about the same no matter which partners it is placed with. That second assumption is the one that breaks in practice, because FRC is a team game: defense, only one robot being able to use a scoring location at a time, field or resource contention, and robots that only shine next to a specific partner all make contributions depend on the alliance. The more a given year’s game rewards those interactions, the less OPR captures — so weight it less and scout more.
• Always pair it with scouting for anything OPR cannot see: defense, reliability, consistency, and endgame success rate.

In short, OPR is a powerful, free starting point and a great cross-check, but it is an estimate built on an assumption that does not always hold.

Key takeaways

• OPR estimates per-team scoring contribution from scores alone via least-squares, using more matches as more equations.
• CCWM = OPR − DPR (net contribution to winning margin); DPR is a rough defense proxy computed from opponents’ scores.
• OPR assumes linear, independent scoring and is built only from points scored, so it supplements but never replaces scouting.

This lesson was adapted from learnfrc.com.