NSBA4 ยท Modeling

Prediction-market models for NSBA season 4. Game lines (win prob / spread / total) and the player ratings that drive them.
๐Ÿ“ How the Model Works
Very detailed, end-to-end explanation of the whole pipeline: seed ratings, the Elo-style update, the tossup-contest simulation, home-court ban, line-making, and player props โ€” with the formulas.
๐Ÿ“ฐ Week 5 Recap โ€” science bowl + reskins (public)
Hub linking the science-bowl recap + NBA/NFL/Soccer/Baseball reskins, xGets coverage analytics, plus the refreshed playoff picture & category kings. "The Model Gets Cooked" โ€” two double-digit favorites lost outright (โˆ’9 by 108, โˆ’13 by 18), the losers negged themselves out of it, mingle/Yunyi & cryo the new #1 seeds. Share one link: /nsba/week5.html.
๐Ÿ“ฐ Week 4 Recap โ€” science bowl + reskins + ratings (public)
Hub linking the science-bowl recap + NBA/NFL/Soccer/Baseball reskins, xGets coverage analytics, and SofaScore-style player ratings (plus a soccer match-ratings cut). "The Blowout Week" โ€” mingle 144โ€“2, Fez 140โ€“34, xpoes two-man. Share one link: /nsba/week4.html.
๐Ÿ“ฐ Week 3 Recap โ€” 6 flavors + xGets + Game Centers (public)
Hub linking the science-bowl recap + NBA/NFL/Quant/Soccer/Baseball reskins, now with xGets coverage analytics, plus NFL & Baseball play-by-play Game Centers. "The Contenders Crack." Share one link: /nsba/week3.html.
๐Ÿ“ฐ Week 2 Recap โ€” all 4 flavors (public)
Single shareable hub linking the science-bowl recap + NBA / NFL / Quant reskins. "Order Restored" โ€” favorites went 4โ€“1, jonathan's 7 negs, mingle's 130, roshan POW. Share one link: /nsba/week2.html.
๐Ÿ“ฐ Week 1 Recap (public)
The shareable, analytical writeup โ€” bonus conversion, category sweeps, the Harry-G no-show, market movements, players of the week, Week 2 outlook. Lives at /nsba/week1-recap.html (not under /nsba4, so sharing it doesn't expose this hub).
๐Ÿ€ Week 1 as NBA (for fun)
The same recap reskinned as an NBA writeup โ€” buckets, free throws, shot zones, DNPs. Same data and names, zero science bowl. Public at /nsba/week1-nba.html.
๐Ÿˆ Week 1 as NFL (for fun)
The recap as an NFL writeup โ€” scoring drives, red-zone %, offensive phases, inactives, bye weeks. Same data and names. Public at /nsba/week1-nfl.html.
๐Ÿ“‰ Week 1 as Quant/IB (for fun)
The recap as a trading-desk P&L report โ€” books, monetization/capture, asset-class attribution, blow-ups, a League Table. Bloomberg-terminal look. Public at /nsba/week1-quant.html.
โšฝ Week 1 as Soccer (for fun)
The recap as a matchweek review โ€” goals, clinical finishing, areas of the pitch, giveaways, squad omissions, no fixtures. Broadcast-pink look. Public at /nsba/week1-soccer.html.
โšพ Week 1 as Baseball (for fun)
The recap as a box-score wrap โ€” runs, RISP/clutch hitting, at-bats by category, rally-killers, late scratches, off days. Red ballpark look. Public at /nsba/week1-baseball.html.
โ™Ÿ Week 1 as a Chess Olympiad (for fun)
The recap as an Olympiad round โ€” 4-board team matches (4โ€“0, 2ยฝโ€“1ยฝ), board-by-board results, converting winning positions, blunders, board-1 defaults, byes, board medals. Gold tournament look (and yes, ChessFun plays chess). /nsba/week1-chess.html.

Current โ€” Week 6

Week 6 Game Lines
Win prob, spread, O/U for all six Week-6 games, off post-Week-5 ratings + most-recent lineups. Model-only (no platform market open yet โ€” these ARE the numbers to open them from). BYE: cryo, xpoes (uncs). JSON: week6_lines.json ยท CSV: week6_lines.csv

Prior โ€” Week 5

Week 5 Game Lines
Win prob, spread, O/U for all six Week-5 games, off post-Week-4 ratings + most-recent lineups. Model-only (no platform market open yet โ€” these ARE the numbers to open them from). Spreads are deliberately widened (GAIN dial) vs the honest margins on the props page; the AMM self-corrects. JSON: week5_lines.json ยท CSV: week5_lines.csv
Week 5 Player Props
Per-player projected tossups (xTU) / points (xPts) with O/U lines + projected final score. BYE: dan.k.memes, ChessFun. JSON: week5_props.json

Prior โ€” Week 4

Week 4 Game Lines
Win prob, spread, O/U for all six Week-4 games, off post-Week-3 ratings + most-recent lineups. Model-only (futures blend dropped after the Round-3 backtest). LOW CONFIDENCE โ€” direction was ~58% over Weeks 2โ€“3. JSON: week4_lines.json
Week 4 Player Props
Per-player projected tossups (xTU) / points (xPts) with O/U lines + projected final score, full slate. JSON: week4_props.json
Player Ratings (post-Week 3)
Per-player, per-subject ability (ฮธ) after Weeks 1โ€“3 โ€” the input the simulation runs on. JSON: ratings_week3.json

Past โ€” Week 3

Week 3 Game Lines
Win probability, point spread, and over/under. Post-Week-2 ratings โ†’ tossup simulation โ†’ home category ban; auto-blends with the platform market once it opens. Released game-by-game โ€” (Stitch)ยฒ @ Connor Chang up first; the rest follow as more stats land.
JSON: week3_lines.json
Week 3 Player Props
Per-player projected tossups (xTU) and points (xPts) with O/U lines + a projected final score, off post-Week-2 ratings. (Stitch)ยฒ @ Connor Chang up first. JSON: week3_props.json
Player Ratings (post-Week 2)
Per-player, per-subject ability (ฮธ) after Weeks 1โ€“2 (direct buzz attribution). The input the simulation runs on. JSON: ratings_week2.json

Past โ€” Week 2

Week 2 Game Lines
Win probability, point spread, and over/under for each Week-2 game. Model-only: post-Week-1 ratings โ†’ tossup simulation โ†’ home category ban.
JSON: week2_lines.json
Week 2 Player Props
Projected per-player tossups and points for each game, with over/under points lines.
JSON: week2_props.json
Player Ratings (post-Week 1)
Per-player, per-subject ability (ฮธ). Seed = combine + draft; updated by Week-1 buzz results (direct attribution). These are the input the simulation runs on.
JSON: ratings_week1.json
๐Ÿˆ NFL Game Center โ€” play-by-play
Drive-by-drive recap of all six games using the real chronological sequence โ€” early leads, the decisive runs, turnovers, comebacks. /nsba/week1-nfl-gamecenter.html
โšพ Baseball Game Center โ€” play-by-play
Inning-by-inning recap of all six games from the box score โ€” rallies, lead changes, rally-killers, the moment each turned. /nsba/week1-baseball-gamecenter.html

Props & Season

Season-Record Prop Markets
Player/team season-long prop markets (records, over/unders).
Draft Grades
Per-team draft evaluation from the preseason combine + value model.

Archive

Week 1 Game Lines
The opening Week-1 lines (preseason: combine + draft + market futures, before any game data).
JSON: game1_lines.json
Tokyo Night ยท regenerated each week. Lines are model output for proofreading, not final market prices.