Turning Numbers into Decisions

A statistic only matters if it changes a decision. This lesson connects the spreadsheet skills and summary numbers from this department to the real calls a scouting-strategy team makes: building a pick list, prepping a match, and reading published metrics without getting fooled by them.

Build a Pick List, Not Just a Ranking

After qualifications you need an ordered pick list of teams you'd want as alliance partners. Don't sort by a single number. Build a table — one row per team, one column per metric that matters this season:

• Average game pieces scored (center)
• Standard deviation of that score (consistency)
• Average penalties drawn (risk)
• Yes/no for key abilities: can it climb? can it play defense?

Then weigh them on purpose. A high average with a huge standard deviation is a gamble; a slightly lower average with rock-steady consistency is often the safer elimination pick. The numbers inform the order — your strategy call sets it.

Quantitative and Qualitative Data Both Matter

• Quantitative data is countable: pieces scored, cycle times, climb success rate.
• Qualitative data is what a scout watches: driver skill, robot reliability, whether a mechanism looked one hard bump from failing.

The best picks come from both. A sheet can tell you a robot averages eight cycles; only a human in the stands can tell you its intake jammed twice and barely recovered. When a number and a scout's eyes disagree, don't pick one — go find out why they diverged.

Read Published Metrics Critically

The FRC community publishes derived metrics so you don't recompute everything yourself. The classic is OPR (Offensive Power Rating) on The Blue Alliance. OPR uses linear algebra across alliance scores to estimate each team's average point contribution — it's not just your average match score halved, because it factors in how strong your partners were.

Useful, but know what it assumes before you trust it:

1. It assumes scoring is linear and additive — each robot's contribution simply sums.
2. With non-linear scoring, defense, or robots competing for the same objective, accuracy drops. OPR rewards scoring, so a strong defender can look weak.
3. Early in an event it's noise. Give it at least four or five matches before reading anything into it.

As The Blue Alliance puts it, OPR can supplement but never replace proper scouting. EPA (Expected Points Added) on Statbotics is a newer model in point units that's more predictive over a season and splits into auto, teleop, and endgame components — but it's still an estimate, calibrated to be wrong a fair share of the time. Use it as a sanity check on your own data, not a substitute for it.

A Simple Decision Workflow

1. Collect clean data — one row per match, consistent columns.
2. Summarize with center and spread per team.
3. Cross-check against scout notes and published metrics like OPR/EPA.
4. Rank into a weighted pick list, not a one-column sort.
5. Decide, then revisit as new matches land.

Done right, data work doesn't hand you the pick — it makes the pick obvious.