A Data-Driven Defense Evaluation Framework

The blind spot worth owning

OPR cannot see defense, and EPA models offense, so a great defender looks mediocre or worse in every public stat. That blind spot is an opportunity: a team that quantifies defense well finds value everyone else’s spreadsheets miss, and defense decides many elimination matches.

What to measure

Defense is inherently relational (its value is in the opponent’s lowered score), so you measure it indirectly and through super-scouts:

• Opponent score suppression: compare the targeted opponent’s scoring in matches with vs without this defender on them. Use TBA score_breakdown to get the opponent’s coral/algae output per match, then difference it. A defender who drops a strong L4 scorer from 6 to 2 coral is enormously valuable even at 0 offensive OPR.
• Time-on-target: super-scout estimate of how much of the match the defender effectively pinned or disrupted its target (a QRScout timer works here).
• Penalty discipline: count fouls drawn against the defender. Effective defense that draws constant penalties is a net negative; track it explicitly.
• Survivability: does the defender avoid getting beached/tipped or stuck? A defender that takes itself out of the match mid-cycle is unreliable.

A simple defense rating

Combine the signals into one comparable score, e.g.:

defense_rating = (avg opponent suppression in points)
               + w1 * (effective time-on-target fraction)
               - w2 * (penalties drawn per match)

Tune the weights to your game; the suppression term should dominate because it is the actual on-field effect. Keep a separate defense ranking next to your offensive picklist, never folded into a single number, because the two answer different questions.

When defense is worth a pick

Defense value is contextual, so decide with data:

• High-ceiling opponents: if the bracket is full of dominant L4 coral machines, a strong defender that can suppress one of them swings matches and is worth a high pick.
• Your alliance already has offense: a third robot that adds marginal offense may be worth less than one that removes a chunk of the opponent’s offense.
• Reliability matters more in elims: a defender that draws penalties or tips over is a liability, since most playoff bracket rounds are single-match (lose and you drop to the lower bracket, lose twice and you’re out) and even the best-of-3 Finals leaves little room for a bad match; weight survivability and discipline heavily.

Scout defense deliberately

Defense rarely shows up in your standard stand-scout taps, so assign it on purpose:

1. Add explicit super-scout fields: who was this robot defending, how effective (1-5), penalties, survivability.
2. Log it per match so you build a sample, not an impression, and avoid recency bias toward the last great or terrible defensive play you saw.
3. In the pre-match brief, name the opponent’s biggest offensive threat and assign which of your robots (or your defensive pick) neutralizes it.

The teams that win on defense are not lucky; they measured a thing the public models ignore and acted on it.

Key takeaways

• Defense is invisible to OPR/EPA, so quantify it via opponent score suppression (TBA score_breakdown diff), time-on-target, penalties drawn, and survivability.
• Keep a separate defense rating from the offensive picklist; its value is contextual on the bracket and your alliance’s existing offense.
• Scout defense deliberately with super-scout fields logged per match to build a sample and avoid recency bias.

This lesson was adapted from learnfrc.com.