Two Long Island towns fourteen miles apart look like opposites in Division I. Massapequa's players start; Islip's sit. It is exactly the kind of pattern a recruiter remembers — and most of it is chance.
Two towns
Since 2022, 25 men from Massapequa, New York have played Division I lacrosse, at 17 programs. They started in 40% of their seasons and spent only 10% on the bench. Across D1, players start in 18% of seasons.
Over the same five years, 19 men from Islip played at 16 programs. They started in none of their 39 seasons, and sat on the bench in almost half.
Taken at face value, that is a recruiting rule: go to Massapequa, skip Islip.
Shuffle the players
Here is the test. Take every Division I player from 2022 to 2026 — 7,762 of them, from 1,880 towns — and deal them back out to towns at random, each town keeping its number of players. Now a hometown means nothing at all. Then compare the towns with at least 25 player-seasons, just as before.
Towns by share of seasons their players started — real, and shuffled
We ran the shuffle 2,000 times. In 80% of them, luck alone produced a town whose players started at least 40% of the time — a Massapequa. Luck also produced towns where nobody started: 1.5 per shuffle on average, and three or more in one shuffle in six. The real data has three: Islip, Miller Place and Sparta, New Jersey.
The real towns do spread out a little further than the shuffled ones, by more than luck explains, so some towns genuinely produce more starters than others. You just can't tell which from five seasons of one town.
How many players does a town need?
Split each town's record into signal and luck, and a town needs about 100 Division I players before its starting rate is even half signal. No town comes close. The largest had 64 players over the five seasons, and only five had 50 or more.
How much of a town's starting rate is signal, not luck
Adjusted for that, the best estimate is that about 23% of Massapequa's players start, and about 15% of Islip's. There is a gap. It is a small fraction of the 40-to-0 on paper.
What does hold up
Bigger groups are a different story, because hundreds of players wash the luck out. Canadian players start more and sit less than a random draw of the same number of D1 players; for starting, luck did not match them once in 4,000 draws. Measured against their own teammates, which rules out Canadians simply landing at weaker programs, they still start 7 points more often. That is the finding behind our piece on Canada.
Other state patterns that look just as striking do not. Florida's 43% bench rate comes close but falls short; so do Georgia's and Colorado's high bench rates, and the low starting rates of players from Texas, Ohio and Virginia, fall within what luck produces once you allow for testing 21 states at once.
What to do with it
Coaches: judge a town by its region, not its record. A town's run of starters, or of players who never see the field, is mostly the particular kids who happened to come through it. The region, with hundreds of players, is where the signal is.
Families: your town's record is not your ceiling. Islip's zero is what chance does to 19 players.
How we counted
- Players: Division I men, 2022–2026, counted once per school. Seasons where the team played fewer than 10 games are left out.
- Starter: started at least 60% of the team's games that season. Bench: played in fewer than 25%, or had no stats line.
- The shuffle: every player keeps his own seasons and results; only his hometown is reassigned, and every town keeps its number of players. The comparison uses towns with at least 25 player-seasons, the same floor as the town tables.
- Signal share: the spread between real towns (at least 10 players) minus the spread luck produces at the same size. The true difference between towns is about 3.6 percentage points either side of the 18% average.
- States and Canada: each group against 4,000 random draws of the same number of players. A result counts only if it would survive testing all 21 groups with at least 150 player-seasons, in either direction. Corrected 25 September 2026: an earlier version tested one direction only and said Florida's bench rate held up.