Bracket Pool Simulation

The pure coaching strategy averages 306 points per tournament versus 387 for chalk — an 81-point deficit. Even the hybrid approach, which only flips a handful of games, trails chalk by 18 points per season.

This might seem contradictory — we proved coaching experience predicts upsets, so why doesn't it win brackets? Three reasons:

1. Bracket scoring is exponential, not linear. Championship = 320 points. Round of 64 = 10. A correct championship pick is worth 32 first-round games. The coaching strategy picks more first-round upsets correctly, but the chalk strategy picks more later-round games correctly because favorites who advance keep earning points. One correct Final Four pick is worth more than getting 16 first-round upset calls right.

Notice: chalk's edge grows in later rounds where the points are worth exponentially more. The coaching strategy bleeds points in the Sweet 16 and beyond.

2. Coaching experience picks cascade incorrectly. When you pick an underdog in round 1, that pick carries forward. If the coaching strategy picks a 10-seed over a 7-seed and gets it right, great — 10 points. But now that 10-seed faces a 2-seed in round 2. The coaching strategy might still pick the 10-seed (their coach is more experienced), but they'll probably lose. So the correct round-1 upset call actually hurts you in round 2 because your bracket now has the wrong team advancing.

3. The coaching edge is real but narrow. We showed coaching experience bumps upset probability from 26% to 38%. That's meaningful for analysis, but 38% is still a losing bet in any individual game. A strategy built entirely on 38% calls will be wrong more often than right, and bracket scoring punishes cascading errors.

The hybrid strategy (default to seeds, flip only when the underdog has 5+ more prior appearances and the seed gap is 5 or fewer) averaged 369 points — closer to chalk's 387 but still 18 points behind on average. It won 6 out of 21 seasons and tied 3.

The hybrid tells us something important: the coaching insight has value as a modifier to a seed-based strategy, but the specific threshold we tested (5+ apps, seed gap ≤ 5) is too aggressive. A more conservative version — perhaps flipping only 1-2 games per bracket where the coaching edge is extreme and the seed gap is small (like a 7-vs-10) — might actually outperform chalk in certain pool formats.

The pattern: coaching wins in chaotic years when top seeds flame out. In orderly years (2007, 2009) where favorites advance, chalk dominates. This makes coaching experience a contrarian signal — it's most valuable precisely when the conventional wisdom (seeds) is wrong.

This simulation reveals a fundamental tension that shows up across economics and investing:

The coaching effect is real. We proved it with 40 years of data across multiple analyses. Experienced coaches genuinely outperform their seeds. The upset rate genuinely jumps 11.5 points when the underdog has the coaching edge.

But "real" doesn't mean "sufficient." A 38% upset probability is better than 26%, but it's still less than 50%. Bracket scoring rewards correct favorites far more than correct upsets because of compounding. And a strategy needs to survive the structure of the competition (cascading brackets), not just be right on individual games.

The right way to use coaching data isn't to build a bracket around it. It's to use it as a tiebreaker within a seed-based framework — and to recognize the handful of games each year where the coaching mismatch is extreme enough to justify overriding the committee's seeding. That's the hybrid approach, refined with better thresholds.