BLAST Bounty 2026 S2 — the model's view

Title race — model vs market

Model = Monte Carlo tournament win probability (calibrated pick model). Market = soft-book outright futures (2026-07-20 board), proportionally devigged (overround 35%). Differences reflect the model AND the sim's draft assumptions — this is an event study, not the graded ledger. The Spirit number is the model's one big call: it is driven by form (13–1 across PGL Astana + the Cologne Major, beating most of this field) and their fresh VRS #1 rank, while the Elo table still has Vitality ahead — and Spirit have not played an official since June 20. Judge accordingly.

model   market

Simulate the bracket yourself

One random tournament per click, using the model's series probabilities for every pairing (BO5 final composed separately; the current model generation prices online and LAN matches identically — the venue flag is not among its selected features). Re-draft picks are random here — the headline numbers above come from the full simulation with a calibrated pick model.

Round of 32 — the board

Win % = model probability the seeded team beats its picker (BO3, online). Veto columns come from the sequential veto model (event-aware ABABAB flow). Shared perma-ban matchups are flagged: in those, first bans deviate 63% of the time.

The draft report card

The bottom half picked their opponents from the top half. Rank #1 would mean "picked the top-half team you're most likely to beat". Availability approximated as the full top half — the broadcast only documented part of the pick order.

The Cache problem

Cache re-entered the pool on 2026-06-24. 21 of 32 teams have never played it in an official match since — including most of the top half of this field. Teams below are the ones who have at least met it in a veto; everyone else is at zero.

Never played: Nuclear TigeRES, Alliance, SINNERS, Nemesis, Wildcard, Falcons, Vitality, Spirit, FURIA, Aurora, G2, MOUZ, FUT, GamerLegion, The MongolZ, Astralis, Liquid, FOKUS, 100 Thieves, OG, Nemiga.

Player boards (last 90 days, min 15 maps)

Round swing is HLTV's signed per-map impact stat; opening +/- is opening kills minus opening deaths per map. Both come from scoreboard data this pipeline captures beyond the basics. Method: per-map scoreboard values captured verbatim, aggregated round-weighted — hltv.org's own player pages recompute from round-level data, so their aggregates can sit slightly lower (spot-checked 2026-07-21: donk 1.55 here vs 1.52 there, identical map counts).