This models a power-matched (Swiss) draw the way tab software like Calico / Tabbycat runs it, then reads off where the break falls.
The draw
- Each round teams are bracketed by record. AP / WSDC: two teams, winner takes the win (0–1 per round). BP: four teams ranked 1st–4th for
3/2/1/0 (0–3 per round).
- Odd brackets are completed by pulling up one team from the bracket below — one pull-up debate per odd boundary, as in the WUDC rules.
- The break is the top N on wins (BP: team points); the tie for the last seats is settled on speaker points.
The three numbers
- Secure the break — breaks no matter how the draw and pull-ups fall. The number to aim at.
- Where points decide — the record at the break boundary in the expected field; speaker points settle the last seats.
- Minimum to break — the lowest record that still breaks in a favourable draw.
Pull-up cases
In a same-bracket debate the counts are fixed — one team up, one stays. Only pull-up debates (and, in BP, mixed rooms) move teams between records. So the cases are every way those can resolve: they mostly change who clears the last seats, and sometimes move the line itself. The cases come from 12,000 simulated draws with a fixed seed, so the same inputs always give the same answer.
Outrounds
- AP / WSDC elimination is 1-v-1 knockout, seeded 1 v N, 2 v N−1. Final = 2, semis = 4, quarters = 8, octofinals = 16.
- BP elimination is rooms of four, top two advance. Final = 4, semis = 8, quarters = 16, octofinals = 32. Rooms use snake seeding.
Assumptions
- Teams are treated as evenly matched — every same-bracket debate a coin flip, every BP room an equally likely ranking. This is the neutral assumption and gives the widest honest range; real skill gaps narrow it.
- Expected bars use the exact closed form (binomial for AP/WSDC, the
{0,1,2,3} score-convolution for BP); the spread comes from simulation.
Engine cross-checked against a brute-force team-by-team simulation and the closed-form distributions; all three agree.