Strategy·March 25, 2026·8 min read
How we evaluate trades (and why opinion polls fail)
Modeled outcome shifts, category leverage, and a decision framework that keeps league context and uncertainty visible.
The three questions
For any proposed trade, in order:
- What's the win-probability delta this week? (Monte Carlo, your league rules.)
- What can we responsibly say beyond this week? (The beta has rolling form, not a licensed rest-of-season feed.)
- Does it change which categories you can win? (Category leverage, before vs after.)
Everything else — "Player X has more upside", "Player Y is a buy-low" — is noise unless it changes one of those three answers.
Why opinion polls fail
A context-free winner label leaves important questions unanswered. League categories, roster slots, waiver replacement, standings, deadline, and injuries can change the decision, and modeled percentages should never be copied between leagues.
Illustrative example
Player A (21/8/4 with 2.1 TO) for Player B (17/11/3 with 1.4 TO), in a punt-TO 9-cat.
- Naive analysis: Player A is "better" by counting stats.
- Contextual read: a punt-TO build may not value the turnover edge, while rebounding could matter more. Run the actual rosters and schedule, inspect uncertainty, and compare standing pat or a counter.
- Decision: accept, counter, or reject only after legality, freshness, and the manager's horizon are explicit.
Closing
Use rankings and opinions for discovery. Close the decision with league evidence and a contingency.