---
name: trade-evaluation
description: This skill should be used when the user asks "is this trade fair", "should I accept this trade", "evaluate this trade", "who wins this trade", "am I getting fleeced", or presents a specific offer (received or about to be sent) for judgment. Runs a two-sided assessment weighing each team's standings and positional needs on top of raw player value. Not for hunting new trade partners (trade-analyzer trade-finder), the back-and-forth after a reply (trade-analyzer trade-negotiation), or browser execution (roster-ops propose-trade).
---

# Trade Evaluation: Two-Sided Assessment

Judge a specific trade from **both** rosters' perspectives. Raw player value is the start, not the verdict — value only counts when it reaches a starting lineup.

## Step 1: Load context

Read `leagues.md` from the project root first — the fields that matter here are scoring, starting slots, roster/bench size, playoff weeks, trade deadline, and whether trades face a veto vote. If the file is missing or blank, ask and suggest running `fantasy-league-setup:league-config`. If more than one league is defined, use the one marked `(default)`.

Gather both full rosters, both records/standings, and the exact terms. Without the other team's roster only half an evaluation is possible — say so and evaluate provisionally.

## Step 2: Who originated the offer? — ask this first

**An incoming offer is adversely selected.** The proposer built it after looking at both rosters and concluded it helps them. That is not bad faith; it is what proposing means. It does change the prior:

- **Offers you received**: require a clearly positive edge, not merely an even one. "Even" plus adverse selection is a small loss on average.
- **Offers you constructed**: even is genuinely fine, and mutual benefit is the normal shape of a good trade.

This is the step most trade advice skips, and it flips the verdict on marginal incoming offers.

## Step 3: Raw value pass

Assign each player a rest-of-season value from projections the user supplies, a consensus source read from the browser (name it and the date), or reasoned judgment (label it as such). Sum both sides.
**Reading data with computer use.** If browser automation is available (Claude in Chrome or equivalent) and the user is already logged into their platform, read the pages directly instead of making them paste — league rosters, the free-agent pool, standings, transaction history, and any rankings site they have open. The session rules from `roster-ops` apply unchanged: the user's session is the auth, never ask for or type credentials, use the UI rather than platform APIs, act at human pace, and stop and hand back on any login or captcha screen.

**This skill is read-only.** Reading a page needs no confirmation, but never click anything that changes a roster, files a claim, or sends an offer from here — that is what the `roster-ops` skills and their confirmation gates exist for.

Interpret the gap in three bands rather than one:

- **Under ~5%** — genuinely indistinguishable. Decide on lineup fit alone.
- **5–15%** — a real edge you cannot measure confidently. Take it when the lineup-delta test agrees.
- **Over 15%** — one side is being fleeced. Say so bluntly, in whichever direction it runs.

Widen these bands to match measured disagreement between projection sources when more than one is available.

## Step 4: Lineup delta — the real test

For **each** team, compute the optimal weekly starting lineup before and after, using this league's slots:

- **Lineup delta = projected starting-lineup points after − before.** A WR4 traded away costs a 3-WR team real points and a 2-WR team nothing.
- **Bench players are not worth zero** — that shorthand is only true in very shallow leagues. With 6–8 bench spots, a plausible starter carries roughly 20–35% of his projection in option value: injury insurance, bye coverage, and trade currency. Only genuinely unplayable pieces round to zero. This matters most on 2-for-1s, which **transfer risk** as well as value: after any consolidation, ask "if my starter at this position goes down, who starts?"
- Run byes and the configured playoff weeks separately from the season average when the deadline is near — a September trade is about 12 weeks, a November trade about 3.
- **Injury status is usually the largest in-season value swing.** Price a hurt player as games missed × per-game value, discount for re-injury risk and ramp-up, and note that a player returning in week 13 can be worth far more to a contender than his season-average value implies.

Translate the result into something decidable: roughly **+1 point per week ≈ +1% win probability per game** against a typical lineup. A +1.2/week delta over five remaining weeks is about +0.06 wins — real, but small enough to weigh against execution risk.

## Step 5: Standings and rivalry context

- **Contender**: values the configured playoff weeks and ceiling; should pay a premium for the best player and for playoff-schedule advantages.
- **Bubble team**: values the next 3–4 weeks; weight near-term delta 2:1. A deal that wins rest-of-season but loses the next three can eliminate them before it pays.
- **Eliminated/rebuilding**: in keeper formats, values keeper-eligible pieces (route to `draft-strategy:keeper-evaluation`). In pure redraft there is little reason to trade at all.
- Determine these from playoff probability and seeding scenarios, not record alone — a 7-3 team locked into a bye and a 7-3 team fighting for the last spot should behave very differently.
- **Are you arming a rival?** Check two things the raw math misses: whether this team is competing with you for a playoff spot, and whether you play them in the remaining weeks. Upgrading the lineup of the team chasing your seed — or the team you face next Sunday — is a real cost.

## Step 6: Verdict

Deliver **Accept / Decline / Counter**, a confidence level, and two or three sentences of reasoning. If Counter, say what change flips it, then suggest `trade-analyzer:trade-negotiation`. Report both sides' win conditions honestly — mutual benefit is what fair trades look like, and saying so builds the credibility that gets future deals accepted.

If the league uses veto votes, note when a fair deal *looks* lopsided, since appearance is what gets vetoed.

## Worked example (fictional)

"Gridiron Gazette," week 10. User (7-3, contender, thin at RB) **receives an offer**: get RB Quincy Marsh + WR Deion Marsette, send WR Cassius Bell. Opponent is 4-6 (bubble), starts 3 WR, RB-surplus.

- **Origination**: they proposed it, so the bar is a clearly positive edge, not break-even.
- **Raw value**: user +~20% — above the 15% band, so this genuinely favors the user on paper.
- **Lineup delta, user**: Marsh upgrades a dead RB2 (+4.1/wk); losing Bell drops WR2→WR3 (−2.9/wk); Marsette is bench depth (+0 to the lineup, but real option value). Net **+1.2/wk**, and Marsh's playoff schedule is soft.
- **Lineup delta, opponent**: Bell upgrades their WR3 (+3.4/wk); Marsh was their bench RB3; Marsette was their WR5 (−0.4/wk). Net **+3.0/wk**.
- **Rivalry check**: they're 4-6, three games back, and not on the user's remaining schedule. Arming them is low-risk.

> **Accept (moderate-high confidence).** It clears the adverse-selection bar — the raw-value edge is above 15% in your favor and it fixes your RB slot. But be clear-eyed about why they offered: they gain more per week than you do (+3.0 vs +1.2), because near-term lineup points are exactly what a bubble team needs. Accept because it's positive for you, not because you won.

Note the verdict does not revert to the raw-value ledger the moment lineup delta is inconvenient — it reports both and says which one drove the decision.