Web beta, not GA
Implemented and build-verified code still needs deployment and production-smoke evidence for each release. Open launch gates stay open.
Connect a compatible league. Pick Pilot finds the decision, shows the legal and fresh case, plans the contingency, and hands the final action back to you.
Connect Yahoo or a compatible Sleeper NBA league. Unsupported settings stop or stay disclosed.
Tiers, mock drafts, auction context, and playoff-week games, using rolling form until licensed projections land.
Triage the matchup, waiver, and lineup decisions that have enough fresh evidence to act on.
Scenario Studio. Test the trade or add/drop before you commit, category by category.
Rest-adjusted schedule and matchup leverage, with postponements and stale data called out.
Archive the result, then grade the decision against what was knowable at the time instead of outcome alone.
Sort the week by leverage, lock time, and what changed instead of opening every league and player page in sequence.
League rules, roster state, source age, assumptions, confidence, and explanation sit beside the call. Thin evidence triggers abstention.
Keep the fallback ready for injuries, lineup locks, games caps, waiver order, FAAB, postponements, and schedule changes.
Pick Pilot remains read-only. Copy or open the provider plan and complete the move yourself; no write-back is implied.
A stale recommendation is not a recommendation. Re-run against the new roster, injury, lock, or schedule state.
Review the choice against information available at decision time, and mark postponed, corrected, or incomplete outcomes ungradable.
Every analytics tool works when the wind is at its back. This is what ours does when it isn’t — outages, missing inputs, model drift. No horoscopes, no faked numbers, no silent fallbacks.
Affected decision surfaces show the last-known-good timestamp and use cached rosters only with a stale label. If a required source is too old, the call abstains.
Decision surfaces expose the inputs they use. If minutes data is missing, the result lowers confidence or abstains rather than quietly padding the number.
Matchup briefs carry their model version, latency, and citation count. A narrative without verified numbers is flagged ‘analyst absent’ — never published as fact.
Custom categories, points and roto, FAAB and priority waivers, keepers, games caps, and IL rules are validated from imported settings where available. Unsupported shapes stop or carry an explicit caveat.
We show the math. Your assist leader might carry the worst lineup rating this week because the category matchup isn’t there — and we’ll say so, with sources.
Supported surfaces expose methodology, source age, and selected tables or exports. Weekly PDF/CSV and protected share paths are Elite; universal chart export is not claimed.
The web beta recomputes rolling-form and schedule signals against the league settings it can verify. If a scoring, roster, waiver, or lock rule is missing, the answer must disclose the assumption or stop. League context is expected in this category; the differentiator is carrying that proof through handoff, replanning, and grading.
| Decision | Generic tool | Pick Pilot (league-aware) | Δ win-prob |
|---|---|---|---|
| Trade · Tatum + PF for Lillard | APPROVE · +2.8 rating | HOLD · FT% drag | −7.2% |
| Stream · Wings off waivers · w12 | Suggests top 10 by FPPG | Suggests 3PM + STL only | +4.1% |
| Lineup · Friday 3-game night | Balances all categories | Punts BLK · locks PTS+AST | +2.6% |
| Draft · round 6 pivot | VBD-only ranking | Category leverage + roster fit | +5.3% |
Connect Yahoo in web beta or a compatible Sleeper NBA league in conditional beta; Fantrax is request access. We use the rules the provider exposes and flag the gaps.
Recent rolling form and schedule context are recomputed against verified scoring and roster settings. Licensed rest-of-season projections have not landed yet.
Every trade, add, drop, or lineup scenario is evaluated against your categories and schedule before a recommendation appears — with advantages and risks named in plain language.
Supported categories are scored by how far they sit from a coin flip. The engine names advantages and risks, while unsupported or stale inputs stay visible.
A value-over-replacement curve, tier breaks, a polled or manually reconciled pick board, and your roster’s category mix beside each other.
Implemented and build-verified code still needs deployment and production-smoke evidence for each release. Open launch gates stay open.
Pro is $9 and Elite is $19. No card is charged until Vandly accepts the integration and live acceptance checks pass.
Current projections use recent player form and schedule context. Licensed projections and active dynasty age curves remain roadmap gates.
Leave your email for beta and provider-access updates. Fantrax requests are reviewed separately; joining the list does not guarantee access.
A co-pilot is only useful if it stays honest. These are product commitments, not launch theatre.
Detect, prove, plan, hand off, replan, grade. The manager stays in command from first signal to final review.