Build a trust-weighted best-ball draft board from analyst track records—then a stack-and-ceiling memo.

Top picks

Protocol

1. 1

   Fill the league intake form.

   Capture scoring, lineup, format, snake vs auction (or budget), your Top X, and the Pack A / Pack B lists you will actually paste. Do not ask the model for a board yet. If a setting is missing, leave it UNKNOWN.

   Copy-paste prompt

   Copy

   You are capturing my fantasy football best-ball settings so we can build a draft board later.
   
   Entertainment only. Not betting advice. I make every pick.
   
   Hard rules:
   - Do not invent contest settings, scoring, roster slots, or host.
   - If a field is missing, write UNKNOWN and leave it for me to fill.
   - Do not suggest players, ranks, ADP, stacks, or ownership in this step.
   - This is best-ball (all drafted players can score). It is not a weekly start/sit league.
   - Output a fill-in form only. Restate it as a compact checklist after I fill it.
   
   Produce this intake form (keep the labels; leave blanks):
   
   LEAGUE INTAKE
   - Contest / site (Underdog / Drafters / other):
   - Format name (Best Ball Mania / other) if I list it:
   - Teams in the draft:
   - Scoring (PPR / half / TE premium / paste custom):
   - Roster construction (QB, RB, WR, TE, FLEX counts; bench if any):
   - Advance / playoff weeks I care about (only if I list them):
   - My draft slot:
   - Draft date / timezone:
   - Top X for the overall board (use 200 if I do not specify):
   - Stack / correlation notes I already want (leave blank if none):
   - Pack A list (track-record / accuracy studies I will paste):
   - Pack B list (CURRENT best-ball ADP, consensus, rankings, and ownership I will paste):
   
   Confirm the rules in one sentence, then wait for my filled form.

2. 2

   Gather Pack A (track record) and Pack B (current only).

   Pack A is accuracy studies and finished-season track records from roughly the last three completed seasons (season-long or best-ball—label which). Pack B is this season’s best-ball consensus, rankings, ADP, and ownership—Underdog, Drafters, FantasyPros, ETR, Late Round—dated. Explicitly do not average old ranking boards to invent this year’s board.

   Paste page titles, dates, and the rows you will stand behind. If you cannot paste a source, that source is UNKNOWN. Do not ask the model to “look up” a live board.

3. 3

   Rate analysts Elite / Strong / Average / Avoid from Pack A only.

   Trust tiers come from Pack A evidence only. Current rankings do not belong in this step. Missing evidence → UNKNOWN. Do not invent win rates.

   Copy-paste prompt

   Copy

   Rate fantasy football best-ball analysts and sources from PACK A only.
   
   Entertainment only. Not betting advice.
   
   Hard rules:
   - Use only the track-record studies, accuracy pages, and notes I paste as Pack A.
   - A “3-year lookback” means roughly the last three COMPLETED seasons. Rate who hit top overall, positional finishers, and breakouts—and who whiffed.
   - Do NOT open or recall a live ranking board. Do NOT use last year’s ranks as this year’s board.
   - Do NOT invent win rates, hit rates, ranks, or seasons I did not paste.
   - If evidence is missing for a source, write UNKNOWN and put them on an Insufficient-evidence list. Do not guess Elite.
   - Ignore any current-season rankings that accidentally appear in Pack A.
   
   Trust tiers (use these labels only):
   - Elite — repeated, documented accuracy on the outcomes I pasted
   - Strong — clearly above the pack on the evidence
   - Average — mixed, thin, or league-average on the evidence
   - Avoid — documented misses, or I told you to avoid them
   
   Output:
   1) Table: Source | Tier | Seasons covered in my paste | What they hit | What they missed | Evidence quotes or page titles I pasted
   2) Insufficient-evidence / UNKNOWN list
   3) One-paragraph method note: these tiers will weight CURRENT boards next; they are not a ranking of players
   
   Confirm the rules, then wait for Pack A.

4. 4

   Build a trust-weighted CURRENT overall top-X board.

   Weight Pack B’s current-season consensus, rankings, and ADP by the Pack A tiers. Higher-trust analysts count more. This is not a 3-year average of old boards. Do not invent a player or a rank.

   Copy-paste prompt

   Copy

   Build a trust-weighted CURRENT overall top-X draft board.
   
   Entertainment only. Not betting advice. I make every pick.
   
   Hard rules:
   - Use only (a) the trust tiers we already agreed from Pack A, and (b) the CURRENT-season consensus, rankings, and ADP I paste as Pack B.
   - Do NOT average three years of old ranking boards. Old boards are for analyst trust only.
   - Do NOT invent a player, rank, ADP, bye, injury, or projection.
   - If a player appears in one source and not others, keep them and mark thin evidence.
   - If you cannot place a player from my paste, write UNKNOWN. Skip them rather than guessing a rank.
   - Weight Elite > Strong > Average. Drop or isolate Avoid unless I say to include them as a contrast column.
   - Follow my scoring and Top X from the intake form.
   - No smash spots, news, or film you were not given.
   - Prefer best-ball ADP I pasted over season-long ADP when both appear; label the source.
   - Do not invent tournament ownership or advance rates.
   
   Output:
   1) Overall board 1..X: Rank | Player | Pos | Team (if I pasted it) | Trust-weighted method note | Sources used | ADP if I pasted it | Flags (thin / UNKNOWN field)
   2) Disagreements: players where high-trust sources differ, with the pasted numbers only
   3) Method footer: “Current board. Not a 3-year average of ranks.”
   
   Confirm, then wait for Pack B and the intake restatement.

5. 5

   Make position lists and a master CSV synced to overall.

   Every position list must stay synced to the overall board you just accepted. Do not add names. Use this column map so the Sheets views in the next step work: A Player · B Pos · C OverallRank · D PosRank · E Target (Y or blank) · F Status (Available|Drafted|Removed) · G Notes · H Team · I ADP · J Bye. Pos values for this sport: QB, RB, WR, TE.

   Copy-paste prompt

   Copy

   From the overall board we just agreed, build position lists and a master table. Stay synced to that overall order.
   
   Entertainment only. Not betting advice.
   
   Hard rules:
   - Use only players already on the overall board (or a Pack B row I pasted). Do not add names.
   - Do not change overall ranks unless I pasted a correction.
   - If a position, team, bye, or ADP is missing, write UNKNOWN.
   - No invented tiers beyond what the overall board already said.
   - Pos values must be one of: QB, RB, WR, TE — or UNKNOWN. Do not invent extra position codes.
   
   Output two blocks:
   
   A) Position lists (QB, RB, WR, TE). Each list: PosRank | OverallRank | Player | Team | ADP | Status=Available
   
   B) Master CSV with this header row exactly:
   Player,Pos,OverallRank,PosRank,Target,Status,Notes,Team,ADP,Bye
   - Target blank unless I already marked Y
   - Status default Available
   - Notes only from pasted flags (thin, UNKNOWN, keeper)
   
   If a count does not match the overall board, list the mismatch instead of filling gaps.

6. 6

   Sync views in Google Sheets (and optional Artifact / Notion).

   Put the master CSV on a tab named Players (header in row 1). Status is Available, Drafted, or Removed. Position views and target views must read Status so they stay in sync—do not keep a separate stale list.

   Paste these formulas on view tabs. Then, if you want a Claude Artifact or Notion database, use the power-user prompt.

   Artifact / Notion prompt

   Copy

   Turn the master table into a draft-day board I can filter.
   
   Entertainment only. I make every pick.
   
   Hard rules:
   - Use only the CSV / table I paste. Do not add players or ranks.
   - Status must be one of: Available | Drafted | Removed
   - Views must stay synced to the master Status column. Do not keep a stale position list.
   - If a field is missing, show UNKNOWN.
   - No live sports data and no invented defaults.
   
   Build:
   1) Master table with the columns I provided
   2) Views: Available overall (sorted by OverallRank), Available by position, Targets (Target=Y), Notes (Notes not empty)
   3) A one-line how-to: when a player is drafted, set Status=Drafted; all views update
   4) For Notion: database properties matching the CSV
   5) For a Claude Artifact: a simple filterable table only
   
   Wait for the CSV.

   Sheets formula — available WRs

   Copy

   =FILTER(Players!A2:J, Players!B2:B="WR", Players!F2:F="Available")

   Sheets formula — available QBs

   Copy

   =FILTER(Players!A2:J, Players!B2:B="QB", Players!F2:F="Available")

   Sheets formula — available overall

   Copy

   =SORT(FILTER(Players!A2:J, Players!F2:F="Available"), 3, TRUE)

   Sheets formula — available targets

   Copy

   =SORT(FILTER(Players!A2:J, Players!E2:E="Y", Players!F2:F="Available"), 3, TRUE)

   Sheets formula — available with notes

   Copy

   =FILTER(Players!A2:J, Players!G2:G<>"", Players!F2:F="Available")

7. 7

   Tag Breakout / Flier / Sleeper / Fade from high-trust notes.

   Tags are optional and must be grounded in Elite or Strong current notes you pasted. A name with no evidence stays untagged. Do not invent a sleeper class.

   Tag definitions (locked)

   Tag	Meaning	Grounding rule
   Breakout	High-trust current notes say the player should jump a tier versus a prior completed season you pasted—not a hunch.	Elite/Strong current note plus a prior-season line you pasted. Else UNKNOWN.
   Flier	Late-round or cheap dart. Fine if they bust.	Must already appear on your board. No new names.
   Sleeper	Available later than high-trust current consensus or ADP implies.	Compare only to ADP/consensus you pasted.
   Fade	High-trust current board is below the ADP or hype you pasted. Not a personal dislike.	Needs a pasted high-trust rank or note versus pasted ADP/hype.

   Copy-paste prompt

   Copy

   Tag Breakout / Flier / Sleeper / Fade using high-trust CURRENT notes only.
   
   Entertainment only. Not betting advice.
   
   Definitions (locked):
   - Breakout — pasted high-trust notes say the player should jump a tier vs a prior completed season you pasted. Not a vibe.
   - Flier — late-round or cheap dart; fine if they bust; must appear on my board.
   - Sleeper — available later than high-trust current consensus/ADP I pasted.
   - Fade — high-trust current board is below ADP/hype I pasted. Not a personal dislike.
   
   Hard rules:
   - Ground every tag in a pasted high-trust source (Elite or Strong). If only Average/Avoid said it, mark weak or skip.
   - Do not invent a tag, quote, rank, or player.
   - If evidence is missing, UNKNOWN — do not tag.
   - A player may have zero tags.
   
   Output a table: Player | Tag | Why (quote or paraphrase my paste) | Source tier | Confidence (documented / thin / UNKNOWN)
   
   Confirm the rules, then wait for the board plus the high-trust notes I paste.

8. 8

   Write the stack-and-ceiling best-ball memo.

   The memo uses the intake form, the current board, and tags you already accepted. If you did not paste projected points, the model may not invent VBD or ownership. UNKNOWN stays visible. You make every pick. This is not a weekly start/sit memo.

   Copy-paste prompt

   Copy

   Write a best-ball draft-day memo from my contest settings and the board we already built.
   
   Entertainment only. Not betting advice. I make every pick.
   
   Hard rules:
   - Use only the intake form, trust-weighted board, position lists, and tags I paste back.
   - Do not invent VBD numbers, ownership, stacks, advance rates, or a positional run that is not visible on my remaining board.
   - If I did not paste projected points, do not compute fake VBD. Use rank / ADP gaps I pasted, or write UNKNOWN.
   - Speak in picks and rounds, not weekly sit/start.
   - Stack / correlation ideas only from pairs already on my board or in my paste. Do not invent a new stack.
   - No live news.
   
   Output:
   
   BEST-BALL MEMO
   1) Early-round posture from my board vs best-ball ADP I pasted
   2) Stack / correlation ideas using only pairs already on my board or in my paste
   3) Middle- and late-round ceiling fliers / sleepers
   4) What to do if a run hits a position I need — scarcity on my remaining board only
   5) Ownership notes only if I pasted ownership; else UNKNOWN
   
   End with: “You make every pick.”
   
   Confirm the rules, then wait for the paste-back of settings + board + tags.

9. 9

   Refresh on draft day.

   Re-paste today’s Pack B only (dated best-ball ADP, consensus, and notes). Keep Pack A trust tiers unless you paste new accuracy evidence. Do not average old boards. Do not invent injuries or ownership. After each pick, set Status to Drafted or Removed so every FILTER view updates. Anything you did not paste is UNKNOWN.

10. 10

    Do not convert this board into a weekly lineup.

    Best-ball scores drafted players without a weekly start/sit. Do not hand this board to the weekly lineup recipe. If you also play season-long weekly leagues, use the season-long draft-strategies recipe for that draft and the weekly lineup recipe after kickoff. This best-ball board is not a live injury feed.

When not to

• Skip this recipe if you have no Pack A track-record evidence and no Pack B current board to paste. A name list is not a board.
• Do not average three seasons of old ranking boards and call it this year’s draft. Old boards rate analysts. Current boards rank players.
• Do not invent win rates, ADP, ownership, stacks, or ranks. Missing evidence is UNKNOWN.
• Do not treat this as the season-long weekly lineup recipe. Best-ball has no weekly sit decision.
• Do not treat the chat as a live injury, weather, or betting feed. This is not betting advice and not a substitute for the contest’s rules.
• Entry fees may apply on Underdog, Drafters, and similar hosts. Play only where legal. Never stake money you cannot afford to lose.
• Do not paste other managers’ private league messages, login cookies, or “export my whole league” tokens.