KISS MM Podcast — NSBA4 Mid-Season Stat Pack

📂 Deep dives — full tables & methodology

The run-of-show below is the on-air script. For the complete leaderboards, the math, and the receipts, each area has its own page:

• The Value Board — full WPA + Buzzer Elo ladders, the who-beats-whom matrix, team roles
• Draft Steals & Busts — the full board, all drafted players
• The Market — sharpest bettors, mispriced players, sharp money
• Buzz Science — timing, stealing, discipline & the rebound rule
• Teams, Luck & the Title Sim — Pythagorean, schedule, full chip-odds decomposition

The hook

Open on the punchline: we ran the entire season through a sabermetrics grinder — win shares, an Elo that ranks every player head-to-head, a prediction market, and 30,000 simulated seasons. Three teasers to promise:

• ⭐ The title is a four-way coin flip — nobody's above 14%.
• ⭐ We found the worst throw of the season, and put a number on it: a team that hit 86% win probability and still lost.
• ⭐ The draft's biggest steal and biggest bust — and the bust was a first-round pick.

Every team's chance to win it all

30,000 Monte-Carlo sims of the rest of the season + the single-elim playoff. SOLID (as a model; small-season caveat applies).

Why nobody breaks 15% — the receipts

This isn't the model being wishy-washy. A title requires winning 3 straight single-elimination games, and each is ~50/50 even for the best team — because skill/combine explains only ~6% of a game's margin. External proof: the prediction market went exactly 50% on 60 games. So a dominant team's ceiling is 0.5 × 0.5 × 0.5 ≈ 12.5%. Worked example, mingle/Yunyi:

Talking points

• ⭐ Dark horse — ChessFun. Only ~62% to even make the playoff (the biggest bubble team in the field), but 7.9% to win the whole thing if they get in. SOLID
  Say "ChessFun is the team nobody wants to draw — barely makes it, but if they're in, they're a top-4 title threat."
• All regular season, no trophy — mingle/Yunyi. 100% locked into the playoff, still only 12.2% to win it. Biggest "makes-it-but-can't-win-it" gap on the board. SOLID
• ⭐ West cannibalism. cryo and James W are both title-caliber but in the same conference — so they can knock each other out before the final. That structure quietly suppresses the West's combined title odds. Good "why the bracket matters" beat.
• Round 6 update baked in: mingle beat sumin 102–48 (mingle → 4-1, sumin → 3-2).

Strength of schedule & luck

• ⭐ dan.k.memes has run the hardest gauntlet in the league — and is still 3rd at 3-1. Most impressive record-vs-schedule. Meanwhile anurag has the easiest slate and sits 7th — no schedule excuse. COLOR
• ⭐ Pythagorean luck (baseball classic — wins your points "deserve"): DANK is the luckiest team — 3 wins on a −74 point differential. Ponnor St. Germain is the unluckiest — 1 win, deserved ~2. TINY n (4–5 games)
  Say "DANK is 3-and-1 while getting outscored by 74 on the season. That is not going to hold."

Win shares & the Buzzer Elo ladder

Win Probability Added (WPA) — "win shares for science bowl"

For every buzz, how much it moved the team's chance of winning, summed over the season. SOLID (30/30 games reconcile).

⭐ The Buzzer Elo ladder — everyone ranked head-to-head

A Bradley-Terry/Elo rating built from who actually beats whom on tossups, on the floor together. Validated: on 183 head-to-head duels the higher-Elo player held the winning buzz record 96% of the time. SOLID (overall ladder; 44–63 contests each)

⭐ Stats lie, Elo doesn't

• Patrick L is #14 in raw points-per-game but #53 in Elo — big counting stats piled up against weak fields. The classic empty-numbers guy. SOLID
  Say "Patrick L looks like a top-15 scorer on paper. Head-to-head against real competition, he's outside the top 50. Volume against weak defenses."

The steal, the bust, the breakout

⭐ Edward C — the breakout of the season (one player, four stats)

• Buzzer Elo #2 overall (1759)
• Biggest Points-Above-Expectation in the league: +8.8 tossups over what his rating predicted
• Biggest week-over-week rating rise of anyone (+0.60 Wk1→Wk5)
• Draft steal: taken #25, producing like the #3 pick

⭐ Draft steals

⭐ Draft busts

Fairness note: 9 drafted players have no game data (likely no-shows) and were excluded, not labeled busts. Captains weren't drafted and are out of this entirely.

Biggest riser

• Sohil R gained +214 Elo from his debut to now (chronological Elo) — steepest upward trajectory in the league. COLOR

Speed vs. knowledge

⭐ The rebound rule (the "how do they know that" stat)

On every tossup this season where two people buzzed, the first buzzer was wrong all 179 times. A second buzz only happens because the first person negged and opened the door — and the rebound is converted 76% of the time. SOLID

⭐ Ice in the veins — Lucas W

Lets the question read to 95% of the way through on his correct buzzes, and still converts 76% (19/25). Pure knowledge, zero panic. TINY n (25 buzzes; buzz-position is descriptive)

⭐ The Vulture — stealing other people's subjects

• Roshan A — 53% of his gets come outside his own subjects. TINY n
• Kian D — 40% stolen off other categories. SOLID

Discipline & the cost of a neg

• ⭐ Kian D — 13 straight conversions, 83% accuracy on 30 buzzes. The most disciplined high-volume buzzer. SOLID
• Vishnu M — 30 buzzes without a single neg. SOLID
• A neg costs ~11.2 points, not 4 — you lose 4 and hand the opponent a free tossup+bonus. Worst offender bled ~134 points to opponents on negs alone. SOLID
• Kensuke O — biggest gunslinger, 50% neg rate on his buzzes. TINY n

Who's the go-to guy — and who really beats whom

Note: we can't do a real "rivalry" ledger from raw duels — opponents only play once, so any two enemies share just ~4 subject tossups. What we CAN see cleanly is how teammates divide the subjects, and for true head-to-head we lean on the validated Buzzer Elo below.

Team subject specialists (the pecking order) SOLID

• ⭐ Lucas W is sumin's physics lockdown — 10 physics gets to his teammate's 1. The most one-sided subject role on any team.
• ⭐ GidTheKid2's split personality: Harry G handles physics (8 gets), Monish S handles math — complementary teammates, each owning a lane. (This is the pair I earlier mislabeled as a "rivalry" — they're on the same team.)
• cryo is a true two-headed monster: suzuko owns Bio (9–1 over Edward), Edward owns Math (9–1) and Physics. Almost no team splits value this evenly.

True head-to-head → use the Buzzer Elo SOLID

For "who actually beats whom," the Buzzer Elo (opponents only, validated 96%) is the honest tool: Akhil B tops it, and it can give an expected win-rate for any matchup. Raw one-off duels are too small to crown.

Per-subject Elo kings COLOR

Fun to name, but each is on ~10–14 contests — call them "current leaders," not settled crowns.

What the market knew (and didn't)

Nobody else has this — an internal prediction market on the league, with a full trade ledger.

• ⭐ Sharpest forecaster — dannyridel, 0.094 Brier over 44 resolved markets (the robust pick). The flashier line: "a new rag" has a better Brier (0.068) but on only 19 bets — and is dead last in net worth. SOLID
  Say "Our best forecaster and our worst trader might be the same guy — a new rag calls games beautifully and is somehow broke."
• ⭐ Most overrated by the market — Eric L. The pool made him the outright #1 favorite to lead Biology scoring. He's actually 4th. (Same story: Sohil R priced #1 in math, actually 3rd.) SOLID
• ⭐ The market can't call a single game. On 60 games, the pregame market went 50/50 — a literal coin flip. This is the hard proof behind "the league is wide open." SOLID
  Say "A room full of people betting real stakes couldn't beat a coin flip on individual games. That's not a knock on them — that's how random this league is."
• But the sharp money is real: the top forecasters, on bets placed 6+ hours before tip, beat the market line 69% vs 50%. Cleanest example — they loaded dan.k.memes when the market favored anurag; dan.k won. COLOR
• Biggest mover: the Theenash S all-star market ran from 13% → 95%. SOLID

Throws, runs & the ban wars

• ⭐ The worst throw of the season. Connor Chang peaked at 86% win probability — and lost 86–92 to czz. The official Heartbreak Index champion. (Runners-up: jonathan and anurag both hit 80% and lost.) SOLID
  Say "We can now settle the 'worst throw' argument with math. Connor Chang was at 86% — and lost. Nobody blew a bigger lead all season."
• ⭐ The Dagger. lysine's 14-point math swing (tossup + bonus) with one question to play flipped czz from 37% → 80% — a +44% single-buzz swing, the biggest of the year. Connor got the final CS tossup but couldn't erase it. SOLID
• Biggest run: cryo ripped off 72 unanswered points on Connor Chang. SOLID
• Game of the Year: cryo vs ChessFun, 7 lead changes, the tightest game by average win-probability. SOLID
• ⭐ The ban wars. Home teams remove a category pair each game — and their choices reveal what they fear: CS banned 8 times, Bio 7; Earth/Space almost never (2). SOLID
  Say "Teams are allowed to ban a category, and they overwhelmingly ban CS and Bio — the two subjects with the scariest lockdown specialists. Nobody's afraid of Earth & Space."

Close

Wrap: the league is a coin flip by design, Akhil's the best player, Edward's the breakout, and the chip is anyone's. Predictions for the back half.