Board vs email.
The same posts arrived by email 2013-2021 (from:*.aloft@football.cbssports.com); the email copies carry replies people sent from their inboxes that the board may not.
Every metric named in the ALOFT historical metrics spec and backlog, with its dimensions, description, gather status, and the seasons it spans or still needs.
From the metrics backlog — proposed, not yet specced. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
The same posts arrived by email 2013-2021 (from:*.aloft@football.cbssports.com); the email copies carry replies people sent from their inboxes that the board may not.
Posts per season by the commissioner; when the playoff-picture series started and stopped; the weeks it covered.
The owner's weekly playoff previews (2018-2020 at least: Week 8/12/13, Trade Deadline, PLAYOFFS, Semi-Finals, Championship, Bowl writeups) as the training set for the bullpen to take the task over - see docs/aloft/playoff-picture.md once written.
Every poll, its options, percentages and vote counts; which passed; which changed a rule the following season.
Who talked, who went quiet, the trash-talk era vs the transactional era ("On the Block Update" share).
"RE:" chains: the longest arguments, and who they were between.
Every commissioner post and poll touching keepers, waivers/FAAB, scoring (decimal scoring 2020), playoff format, draft order/lottery, dues and payouts - dated, quoted, and attributed. The 2004 board already carries a post announcing the move to two keepers the following season (board post 2, 2004-08-27) and a poll on drop-add timing. (Cited, not quoted: this paragraph is rendered on the public metrics catalog, and post 2 is a member's post, which build-history's privacy guard refuses on a public page.) Cross-check against the constitution text (cbs/constitution.txt) and the Sleeper ruleset changes 2021-2026.
From the historical metrics spec. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Seasons and weeks covered by the commissioner's weekly playoff-picture posts.
Poll count, options, percentages and vote counts from the message board.
kind by rule: subject starts "Commissioner:" / author is the commissioner and subject matches the playoff-picture set; poll != null; subject matches "On the Block Update"; author "Fantasy Journalist"; else chatter.
From the metrics backlog — proposed, not yet specced. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
W/L/T, PF/PA, titles, finals, playoff appearances, seasons, slots held, names used — per human. CBS's own all-time table is wrong for 11 of 27 slots; this replaces it.
Record in games decided by less than 5 points (era-scaled threshold).
Rolling 3- and 5-season win% per coach; the peak window of each career.
Rebuilt from weekly results through the registry (CBS's h2h is slot-based and inherits the misattribution).
Luck proxy: coaches whose finish beats or trails their PF rank, per season and career.
How often the #1 seed wins; how often a sub-.500 team wins (the 2020 champion was 9-7). From brackets + standings.
Pairs ranked by games played and cumulative margin closeness; playoff meetings as a tiebreaker.
Longest win/loss streaks per coach across seasons, and longest playoff-appearance streaks.
Renames per slot and per coach; the coaches who never renamed vs the one with seven names on one slot (Kevin Herndon, slot 8).
Seasons per coach, arrivals/departures per season, share of the league that is "original" (2002) each year. Co-managed seasons flagged (shared).
With N teams a coach's naive title rate is 1/N; compare each coach's realised rate over their tenure.
From the historical metrics spec. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Record in consolation/post-season games that are not bracket games.
titles + runner-ups. Source: champion | runnerUp. Ben Byler 3
Winning percentage, computed as (W + T/2) / (W + L + T).
season totals (CBS-reconciled). Source: Σ C.seasons[].pf/pa | includes unopposed-week points. Terry 20964.2 / 19760.5
Regular-season points-for per game, era-labelled and never pooled across eras.
Number of seasons in which a coach appeared in at least one bracket game; first-round byes count.
Record from bracket/playoff games only.
CBS-style W-L-T summed. Source: Σ C.seasons[].w/l/t | includes playoff games in early eras — label "season totals (CBS)". Terry 160-116-0
Count of regular-season #1 finishes from the year-by-year standings.
Record from regular-season games only, with ties as T.
seasons with a (slot, season) row. Source: C.seasons.length | shared seasons count once for each coach. Terry Herndon = 19 (CBS)
Number of distinct CBS slot IDs held and team names used in a career.
first and last season. Source: min/max C.seasons[].season | gaps allowed (Ben Byler 2002-05, 2008, 2012, 2014-19). Byler span 2002-2019, seasons 12
League Champion. Source: C.seasons[].champion ← finals{y}.winner | 2015 tie-broken by the CBS champion table (flag). Terry 2002,2005,2009,2010
From the metrics backlog — proposed, not yet specced. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
CBS registry + Sleeper users.json display names; the join is by human, and the Sleeper display names need the same alias treatment (tmmercer = Todd Mercer, etc.).
Never compare raw PF across 2023/2024 or across platforms without labelling the era.
Whether Sleeper roster ids map onto CBS slots at all, or only onto coaches. Probably the latter; test.
From the metrics backlog — proposed, not yet specced. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Highest season points by round drafted; the late-round hits list (Puka-class) across both eras.
Per-coach mean points-over-round-median across all boards; consistency year to year (a coach who is good at this should repeat).
Pick 1's owner each year; test the "inverse finish" rule against CBS-era finishes.
CBS boards carry each player's season Total Fpts: points per pick number, per round, per slot, per coach — 18 seasons of it.
Sleeper and later CBS boards carry Elapsed Time: who drafts fastest, whether time-on-clock relates to pick quality.
RB/WR/QB/TE share of early rounds 2003-2026; when the league's QB obsession began.
Longest same-position runs per draft; first QB/RB/WR/TE/K/DEF pick each year; K and DEF drafted before round 10 by season (the 2026 room reached +37.6 picks on K vs ADP).
From the historical metrics spec. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Best and worst Fpts-over-round-median picks per draft, with player, round, coach and season.
Total Fpts over round median across a coach's picks, per season and career.
Sleeper keeper picks by season, flagged via `is_keeper` draft picks.
Mean Total Fpts by pick number and round, per season.
From the historical metrics spec. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Head-to-head record between a coach and an opponent, by coach identity.
Head-to-head record in bracket/playoff games only.
Pairs ranked by games played and cumulative margin closeness; playoff meetings as a tiebreaker.
pairs[a,b].w == pairs[b,a].l and PF/PA mirror. must hold for every pair
From the metrics backlog — proposed, not yet specced. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Season W/L and PF by draft slot 1-14, linear draft, 2021-2026; and CBS-era 2003-2020 if the draft order can be recovered from the draft page (pick 1's team). *Trap:* slot is assigned by finish (lottery + inverse finish), so it is confounded with prior strength — condition on prior-season finish.
Because the draft is linear (no snake) and keeper cost is "round k at your own slot", a slot-1 keeper costs pick 1 while a slot-14 keeper costs pick 14. Measure realised keeper value by draft slot, and whether early-slot coaches keep fewer or better players.
Consecutive-season keeper streaks per (player, coach). *Trap:* coach, not slot — join through the registry; a slot changing hands ends the streak even if the name stays.
Total keeper-seasons per player league-wide; the "franchise player" list.
A keeper costs round k; compare his season points to the median points of players actually drafted in round k that year (from the same board). Measures whether keeping beat drafting at that price.
Keepers lock 48 hours before the draft. When news lands inside that window (Jacobs 2026), who was holding whom — a case-by-case log rather than a rate.
Where each keeper was originally drafted (the late-pick-value work already does this for round 10+): rate at which each origin round produces a keeper, and how long those keepers last.
Of players kept in year N, what share are kept in N+1, N+2… by anyone; by the same coach.
RB/WR/QB/TE share of keepers; how it drifts. Already partly measured 2026 (19 RB / 16 WR / 4 TE / 4 QB) — extend to all seasons.
Count of is_keeper picks by season; distribution 0-4 per coach. *Trap:* 2021 official but a different scoring era (rec 0 → 0.5 at 2024).
Regular-season W/L, PF, playoff rate, and title rate for coaches who kept 0/1/2/3/4, per season and pooled. *Trap:* keeper count correlates with roster strength from the prior year — report prior-season finish alongside it, or the number reads as a cause.
From the historical metrics spec. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Record a team would have if it played every other team each week.
Record in games decided by less than 5 points (era-scaled threshold).
actual W − all-play W scaled to games. Source: W_actual − W_allplay × (games / (teams−1)). positive = won more than the points earned
Difference between a team's regular-season PF rank and its final finish rank.
From the metrics backlog — proposed, not yet specced. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
CBS awards pages were captured but every checked season reads "Commissioner has not set awards" — verify all 19 before assuming.
The owner's own playoff pictures and championship writeups from Gmail 2013-2021 — human-written history the site never stored.
401 CBS "fantasy journalist" articles 2015-2020: index by season, week, matchup, teams named; the "biggest blowout" and "closest game" claims those articles make can be verified against scores.
From the metrics backlog — proposed, not yet specced. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Every title team's draft class and keeper set; players on the most title teams.
in league history.
Weekly points under this league's own scoring, K and DEF included with return yardage.
Players who spent the most seasons on one coach's roster (keepers + redrafts).
From the historical metrics spec. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Best (lowest-numbered) CBS end-of-season power-ranking rank in a single season.
Mean CBS end-of-season power-ranking rank across all seasons with a ranking.
null in Sleeper years - CBS's power number is its own formula and is never imitated; the cross-era series is RANK only
'computed: all-play pct, PF, lineup efficiency' - printed beside every Sleeper-era row; CBS rows print 'CBS power ranking'
rank of all-play (breakdown) pct over regular-season weeks (§5), ties by PF, then by ppts efficiency (fpts / ppts)
Count of seasons in which a coach held the #1 CBS end-of-season power ranking.
the same rank after each regular-season week, from matchups/<w>, for a sparkline; time series yes
From the historical metrics spec. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Biggest margin; closest game; most ties in a season.
the CBS record book (raw/cbs/<y>/year-by-year table 0) lists some of these; where they disagree, show both and say why (CBS's include unopposed weeks and its own tie resolution).
Longest playoff-appearance streak; most consecutive titles.
Longest win streak and loss streak (regular season, across seasons).
All-time leaderboards: titles, finals, seasons, best winning percentage and regular-season #1 finishes.
Season PF high and low (regular season only, per era).
Single-game high and low (team), with opponent, week, season.
From the metrics backlog — proposed, not yet specced. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
Playoff format per season (2 games in 2002, 7 by 2006); how often the bye team wins.
2003-2005 had three divisions; compare playoff qualification patterns.
Median team PF per season 2002-2025 across both platforms; the 2024 half-PPR step is visible and must be labelled.
For each coach-season, record vs opponent and vs median separately; who the second result helped or hurt.
Standard deviation of W and of PF across teams per season; Gini of PF.
Highest and lowest team score each week per season; the all-time single-game highs and lows (CBS record book carries some of these — cross-check).
From the historical metrics spec. Spans cbs-std, cbs-dec, slp-std and slp-half where data exists.
rounds, games, scores, winners. Source: bracket form (in S.finals and G.kind=bracket). BYE rows are not games
S.finals[y]. tieBreak flag shown when present
rules timeline (roadmap). linked, never asserted
W-L-T, PF, PA, division where present, finish order. Source: S.standings[y], S.rows. resolve team → coach via S.rows
derived; constitution by era. denominators
Maximum and minimum team score in each regular week; unopposed weeks excluded.
every game. Source: G filtered by season. ties shown as ties