Predicting each of Fall 2024, Spring 2025, Fall 2025 and Spring 2026 using only races sailed before that season began, the site's rating model put 69.5% of boat pairs in the right order across 1,356,317 pairs in 15,965 races. The strongest baseline, Elo (multi-player, tuned K), managed 68.0%; a coin flip gets 50%.
A rating is only useful if it predicts results it has not seen. The ratings on this site come from a Plackett–Luce model: every skipper has a strength, and a race finish is modelled as repeatedly choosing the next boat across the line with probability proportional to strength. The site's version lets each sailor's strength drift week to week. This page tests whether that model forecasts future races better than simpler ways of rating sailors.
| Model | Pairwise accuracy | Log-loss | Winner picked | Rank correlation | ICSA accuracy | ISSA accuracy | Returning sailors | With a newcomer |
|---|---|---|---|---|---|---|---|---|
| No information (coin flip) | 50.0% | 0.6932 | 10.8% | 0.000 | 50.0% | 50.0% | 50.0% | 50.0% |
| Average finish percentile | 64.7% | 0.6212 | 28.5% | 0.385 | 65.8% | 63.8% | 66.5% | 59.7% |
| Elo (multi-player, tuned K) | 68.0% | 0.5885 | 32.8% | 0.461 | 69.4% | 66.8% | 70.3% | 61.6% |
| Plackett–Luce, static rating | 68.0% | 0.5889 | 32.9% | 0.458 | 69.6% | 66.5% | 70.6% | 60.5% |
| Plackett–Luce, weekly, before the rating fix (previous site model) | 68.8% | 0.5788 | 34.0% | 0.481 | 70.1% | 67.7% | 71.3% | 61.9% |
| Plackett–Luce, weekly (site model) | 69.5% | 0.5818 | 34.6% | 0.498 | 70.7% | 68.5% | 72.0% | 62.4% |
The rating fix. Until September 2026 the model stopped fitting after 2,000 steps, long before the ratings had settled, which squeezed them together. Ratings are now fitted until they settle, with a pull toward the average on each sailor's rating when they start racing, which matters most for sailors with few races; its strength and the week-to-week smoothing were tuned on week-ahead forecasts from 2024 and spring 2025. On these season-ahead forecasts, pairwise accuracy went from 68.8% before the fix to 69.5%, and log-loss from 0.579 to 0.582. Sharper ratings age, though: a whole season ahead they order boats better but state their confidence a little too strongly, which is why log-loss here went the other way. With ratings from the week before an event, which is how the site uses them, they are better on both counts: log-loss 0.547 before the fix against 0.538 now. At championships, with ratings from the week before, the fix was worth +0.3 to +0.7 points of pairwise accuracy (95% interval).
Pairwise accuracy for each forecast season. Fall seasons bring a new freshman class, so a larger share of pairs involve sailors no model has seen race before.
| Forecast season | Ratings frozen at | Races | New sailors | Coin flip | Average finish | Elo | Static PL | Previous site model | Weekly PL (site) |
|---|---|---|---|---|---|---|---|---|---|
| Fall 2024 | 2024-W36 | 4,459 | 955 | 50.0% | 64.1% | 66.6% | 67.0% | 67.3% | 67.9% |
| Spring 2025 | 2025-W03 | 3,563 | 522 | 50.0% | 64.9% | 68.9% | 68.6% | 69.8% | 70.7% |
| Fall 2025 | 2025-W36 | 4,211 | 909 | 50.0% | 64.2% | 67.3% | 67.5% | 68.0% | 68.6% |
| Spring 2026 | 2026-W03 | 3,732 | 719 | 50.0% | 65.9% | 69.8% | 69.1% | 70.7% | 71.4% |
Championships are where ratings get the most attention, so this test scores only championship-level fleet regattas: ICSA national finals and semifinals, conference championships and Atlantic Coast Championship rounds, and ISSA national championships, district championships and district qualifiers. For each one, every model is fit on all races through the week before the event and then predicts it. Baseline settings come from that season's forecast tuning.
The site model ordered 72.3% of boat pairs correctly (95% interval 71.5–73.2%) across 166 regattas and 3,114 races; Plackett–Luce, static rating managed 71.2%. Resampling whole regattas, the site model came out ahead of Static PL in 100% of resamples, and the interval for its margin (+0.8 to +1.5 points) stays above zero. With ratings frozen at the start of the season instead, the same model scored 70.8% on these regattas, so results from earlier in the season add real information.
| Model | Pairwise accuracy | 95% interval | Log-loss | Winner picked | ICSA | ISSA | Site model better in |
|---|---|---|---|---|---|---|---|
| No information (coin flip) | 50.0% | 50.0–50.0% | 0.6932 | 8.5% | 50.0% | 50.0% | 100% |
| Average finish percentile | 67.0% | 66.2–67.8% | 0.6056 | 26.8% | 66.9% | 67.0% | 100% |
| Elo (multi-player, tuned K) | 70.9% | 70.0–71.8% | 0.5592 | 30.9% | 71.5% | 70.3% | 100% |
| Plackett–Luce, static rating | 71.2% | 70.3–72.2% | 0.5581 | 32.8% | 71.8% | 70.7% | 100% |
| Plackett–Luce, weekly, before the rating fix (previous site model) | 71.9% | 71.0–72.8% | 0.5466 | 32.7% | 72.4% | 71.4% | 100% |
| Plackett–Luce, weekly (site model) | 72.3% | 71.5–73.2% | 0.5382 | 34.0% | 72.6% | 72.0% | — |
Predicted versus actual team totals at every championship
Not every championship is scored yet. Some championship regattas appear on the official scoring sites but their results are not in this site's data, so they are left out: Fall 2024 ICSA 1 of 19.
| Season | League | Listed | Scored | Missing |
|---|---|---|---|---|
| Fall 2024 | ICSA | 19 | 18 | Women's Urn Trophy/NEISA Women’s Fall Champs |
Nationals are the hardest test for the ratings. Each table uses only results from before the week of the regatta: every skipper's rating going in, the standings those ratings predict, and how teams actually finished. Ratings are on the rankings scale, and “new” marks a skipper with no earlier races, who starts at the typical newcomer rating. Predicted points also include each skipper's penalty risk. The chance columns come from simulating the whole regatta, every race and every penalty, thousands of times with those ratings: how often each team won, and how often it finished where it did or better.
The ratings picked the winner. Predicted totals were off by 27.2 points per team and finishes by 2.7 places, with a rank correlation of 0.80 between predicted and actual standings. Over 14 races per division, the ratings gave Brown University a 31% chance to win, and standings at least this far from the prediction came up in 35% of 2,000 simulated regattas.
| Actual finish | Team | Predicted finish | Beat prediction by | Chance of this finish or better | Win chance | Points | Predicted points | Skippers and ratings going in |
|---|---|---|---|---|---|---|---|---|
| 1 | Brown University Bears | 1 | +0 | 31% | 31% | 163 | 179.3 | A Guthrie Braun 492 B Blake Behrens 449 |
| 2 | Stanford University Cardinal | 3 | +1 | 44% | 23% | 174 | 186.7 | A Thomas Sitzmann 486 (14 races) B Reade Decker 414 (10 races) B Vanessa Lahrkamp 505 (4 races) |
| 3 | Georgetown University Hoyas | 8 | +5 | 12% | 2% | 211 | 248.0 | A Enzo Menditto 392 B Peter Herlihy 406 |
| 3 | Harvard University Crimson | 2 | -1 | 63% | 26% | 211 | 184.6 | A Justin Callahan 490 B Mitchell Callahan 426 |
| 3 | Yale University Bulldogs | 11 | +8 | 6% | 0% | 211 | 263.7 | A Morgan Pinckney 386 B Dorothy Mendelblatt 401 |
| 6 | U. S. Naval Academy Midshipmen | 7 | +1 | 38% | 2% | 223 | 242.3 | A Nathan Smith 422 B Henry Allgeier 403 |
| 7 | Dartmouth College Big Green | 9 | +2 | 43% | 2% | 225 | 248.2 | A Ryan Satterberg 406 B Chase Decker 408 |
| 7 | University of Pennsylvania Quakers | 10 | +3 | 31% | 1% | 225 | 261.5 | A Cole Woodworth 396 B Jackson Mcaliley 394 |
| 9 | Tulane University Green Wave | 4 | -5 | 84% | 6% | 242 | 219.8 | A Hamilton Barclay 460 (14 races) B Christian Ebbin 409 (10 races) B Kelly Holthus 393 (4 races) |
| 10 | College of Charleston Cougars | 5 | -5 | 88% | 5% | 243 | 223.3 | A Noah Zittrer 424 (12 races) A Pierce Olsen 403 (2 races) B Benjamin Dufour 437 (14 races) |
| 11 | Roger Williams University Hawks | 6 | -5 | 85% | 3% | 253 | 236.8 | A Carlos De Castro 458 (14 races) B Kyle Pfrang 382 (12 races) B Oliver Stokke 352 (2 races) |
| 12 | St. Mary's College of Maryland Seahawks | 13 | +1 | 59% | <1% | 302 | 284.9 | A Nathan Jensen 363 (12 races) A Raam Fox 344 (2 races) B Landon Cormie 389 (14 races) |
| 13 | Connecticut College Camels | 18 | +5 | 7% | <1% | 318 | 365.3 | A Henry Scholz 327 (10 races) A Rory Murray 270 (4 races) B William Hurd 289 (14 races) |
| 14 | Boston College Eagles | 12 | -2 | 90% | <1% | 321 | 275.8 | A Tanner Krygsveld 400 (6 races) A Alex Lech 374 (4 races) A Peter Busch 381 (2 races) A Peter Joslin 384 (2 races) B Caroline Sibilly 382 (6 races) B Cody Roe 360 (4 races) B Jack Redmond 395 (4 races) |
| 15 | Jacksonville University Fins | 14 | -1 | 79% | <1% | 337 | 305.4 | A Owen Bannasch 404 (14 races) B Patrick Igoe 299 (8 races) B Hank Seum 323 (4 races) B Cole Schweda 317 (2 races) |
| 16 | Bowdoin College Polar Bears | 16 | +0 | 63% | <1% | 343 | 345.0 | A Michelangelo Vecchio 334 (12 races) A Ryan Keenan 304 (2 races) B Kyra Phelan 317 (11 races) B Lucca Antonietti 288 (3 races) |
| 17 | Cornell University Big Red | 15 | -2 | 89% | <1% | 370 | 330.7 | A Winborne Majette 356 (14 races) B Gilda Dondona 321 (12 races) B Marcus Greco 245 (2 races) |
| 18 | Fordham University Rams | 17 | -1 | >99% | <1% | 380 | 355.1 | A Jacob Zils 327 (14 races) B Lucas Thress 312 (8 races) B Patrick Shachoy 280 (4 races) B Erickson Rankin 234 (2 races) |
The ratings picked the winner. Predicted totals were off by 28.1 points per team and finishes by 2.0 places, with a rank correlation of 0.87 between predicted and actual standings. Over 12 races per division, the ratings gave Stanford University a 76% chance to win, and standings at least this far from the prediction came up in 46% of 2,000 simulated regattas.
| Actual finish | Team | Predicted finish | Beat prediction by | Chance of this finish or better | Win chance | Points | Predicted points | Skippers and ratings going in |
|---|---|---|---|---|---|---|---|---|
| 1 | Stanford University Cardinal | 1 | +0 | 76% | 76% | 94 | 112.6 | A Vanessa Lahrkamp 473 B Sophie Fisher 393 |
| 2 | Yale University Bulldogs | 2 | +0 | 31% | 7% | 128 | 165.1 | A Dorothy Mendelblatt 385 B Carly Kieding 363 |
| 3 | Harvard University Crimson | 4 | +1 | 27% | 3% | 161 | 181.7 | A Zoey Ziskind 374 B Kate Danielson 338 |
| 4 | Bowdoin College Polar Bears | 9 | +5 | 11% | 1% | 164 | 215.9 | A Lauren Russler 346 B Kyra Phelan 291 |
| 5 | College of Charleston Cougars | 6 | +1 | 44% | 2% | 171 | 190.2 | A Bella Shakespeare 352 B Ashley Alfortish 324 |
| 5 | Cornell University Big Red | 7 | +2 | 41% | 1% | 171 | 190.7 | A Winborne Majette 339 (12 races) B Sophia Devling 362 (10 races) B Gilda Dondona 308 (2 races) |
| 7 | Tulane University Green Wave | 11 | +4 | 30% | 1% | 208 | 220.8 | A Ava Anderson 365 (12 races) B Gabriela Vassel 269 (8 races) B Lola Kohl 244 (4 races) |
| 8 | Brown University Bears | 3 | -5 | 87% | 5% | 214 | 174.0 | A Katharine Doble 378 (12 races) B Katherine Mcnamara 320 (6 races) B Laura Hamilton 381 (6 races) |
| 9 | Georgetown University Hoyas | 5 | -4 | 81% | 3% | 226 | 185.9 | A Emily Doble 341 B Kelly Bates 362 |
| 10 | Roger Williams University Hawks | 10 | +0 | 59% | <1% | 231 | 220.2 | A Lucy Meagher 356 B Tavia Smith 270 |
| 11 | Tufts University Jumbos | 12 | +1 | 57% | <1% | 236 | 234.4 | A Ella Hubbard 284 (8 races) A Sophia Hubbard 282 (4 races) B Maddie Janzen 315 (12 races) |
| 12 | Dartmouth College Big Green | 8 | -4 | 95% | 2% | 247 | 192.0 | A Bella Casaretto 366 (11 races) A Alders Kulynych-Irvin 240 (1 races) B Olivia Drulard 333 (12 races) |
| 12 | George Washington University Revolutionaries | 16 | +4 | 4% | <1% | 247 | 312.8 | A Arrieta Angueira Salbidegoitia 257 B Hayden Clary 157 |
| 14 | Massachusetts Institute of Technology Engineers | 14 | +0 | 68% | <1% | 258 | 267.6 | A Brooke Barry 255 (12 races) B Karya Basaraner 281 (9 races) B Emma Wang 234 (3 races) |
| 15 | Boston College Eagles | 13 | -2 | 96% | <1% | 276 | 244.9 | A Caroline Sibilly 394 B Kate Joslin 170 |
| 16 | U. S. Coast Guard Academy Bears | 15 | -1 | 86% | <1% | 316 | 288.7 | A Madeline Murphy 267 (12 races) B Ella Demand 213 (10 races) B Meara Conley 188 (2 races) |
| 17 | Jacksonville University Fins | 18 | +1 | 57% | <1% | 350.1 | 339.1 | A Kaitlyn Liebel 188 (12 races) B Fiona Froelich 177 (10 races) B Kaitlyn Anderson 9 (2 races) |
| 18 | University of Rhode Island Rams | 17 | -1 | >99% | <1% | 368 | 335.0 | A Ariana Schwartz 170 B Emaline Ouellette 187 |
The ratings picked Point Loma High School; Severn School won. Predicted totals were off by 47.1 points per team and finishes by 2.3 places, with a rank correlation of 0.89 between predicted and actual standings. Over 16 races per division, the ratings gave Severn School a 23% chance to win, and standings at least this far from the prediction came up in 52% of 2,000 simulated regattas.
| Actual finish | Team | Predicted finish | Beat prediction by | Chance of this finish or better | Win chance | Points | Predicted points | Skippers and ratings going in |
|---|---|---|---|---|---|---|---|---|
| 1 | Severn School Admirals | 2 | +1 | 23% | 23% | 167 | 216.7 | A Annie Sitzmann 354 B Harrison Szot 364 |
| 2 | Lucy Beckham High School Bengals | 6 | +4 | 14% | 6% | 192 | 262.9 | A James Pine 382 B Nathan Pine 252 |
| 3 | St. George's School Dragons | 4 | +1 | 28% | 7% | 196 | 254.3 | A Gil Hackel 379 (16 races) B Miles Cundey 264 (12 races) B Amelon Rule 318 (4 races) |
| 4 | Point Loma High School Pointers | 1 | -3 | 85% | 43% | 200 | 197.1 | A Wyatt Kelly 369 (13 races) A Kevin Cason 353 (2 races) B Anton Schmid 382 (16 races) |
| 5 | Mater Dei High School Monarchs | 3 | -2 | 61% | 10% | 203 | 241.7 | A Nickolas Lech 350 (16 races) B Kingston Keyoung 312 (12 races) B Colin Kennedy 378 (2 races) B Gage Christopher 373 (2 races) |
| 6 | Ransom Everglades School Raiders | 11 | +5 | 17% | <1% | 229 | 319.3 | A Ava Mc Aliley 292 (10 races) A Sander Block 287 (6 races) B Max Wolfensberger 270 (16 races) |
| 7 | Barrington High School Eagles | 7 | +0 | 48% | 2% | 304 | 284.6 | A Duffy Macaulay 355 B Ben Reuter 254 |
| 8 | Christchurch School Seahorses One | 5 | -3 | 77% | 5% | 313 | 260.7 | A Wylder Smith 352 (16 races) B Sam De Los Reyes 291 (14 races) B Elliott Lipp 301 (2 races) |
| 9 | Gulliver Preparatory School Raiders | 12 | +3 | 42% | 1% | 318 | 323.8 | A Connor Karr 326 (16 races) B Arturo Zizold 242 (14 races) B Danika Torres 121 (2 races) |
| 10 | Southern Regional High School Rams | 9 | -1 | 68% | 1% | 321 | 299.2 | A Jude Ryon 317 B Gannon Botwinick 270 |
| 11 | Key School Zags | 15 | +4 | 12% | <1% | 335 | 412.1 | A Trey Waters 252 (16 races) B Casey Burman 174 (14 races) B Ethan Purdon 104 (2 races) |
| 12 | Brunswick School Bruins | 10 | -2 | 75% | 4% | 360 | 305.5 | A Harrison Gandy 299 (14 races) A Sebastian Sheppard 354 (2 races) B William Whidden 276 (16 races) |
| 13 | San Marcos High School Royals | 8 | -5 | 95% | 1% | 365 | 289.9 | A Dylan Seawards 322 (16 races) B Sam Wells 303 (12 races) B Taylor Escola 218 (4 races) |
| 14 | Arrowhead High School Warhawks | 13 | -1 | 87% | <1% | 402 | 339.5 | A John Lieber 248 B Nicholas Berkowitz 281 |
| 15 | Bainbridge High School Spartans | 14 | -1 | 58% | <1% | 429 | 403.3 | A Cyrus Yan 187 (10 races) A Stone Dewey 252 (6 races) B Nelson Dorsey 219 (16 races) |
| 16 | Jesuit High School, NOLA Blue Jays | 20 | +4 | 8% | <1% | 429.7 | 518.5 | A Jack Meade 199 (14 races) A Liam Moore 45 (2 races) B Reed Gibbs new (12 races) B David Karcher 178 (4 races) |
| 17 | New Trier HS Trevian | 17 | +0 | 77% | <1% | 450 | 424.9 | A Nathan Finkelstein 230 (16 races) B Aiala Angueira Salbidegoitia 201 (12 races) B Ralph Lipford 30 (4 races) |
| 18 | Minnetonka High School Skippers | 19 | +1 | 62% | <1% | 459 | 461.3 | A Connor Jewett 216 (12 races) A Mark Yakovlev new (4 races) B Reese Kottke 191 (10 races) B Maggie Mcgary 105 (4 races) B Maxwell Kelley 89 (2 races) |
| 19 | Jones College Prep Eagles | 16 | -3 | 97% | <1% | 460 | 421.3 | A Nissa Berman 252 (14 races) A Jack Eskilson 97 (2 races) B Duke Diep 187 (12 races) B Quinn Frakt 92 (4 races) |
| 20 | Olympia High School Bears | 18 | -2 | >99% | <1% | 539 | 457.0 | A Alan Timms 219 (16 races) B Simone Reck 150 (10 races) B Kaden Kim 67 (6 races) |
The ratings picked the winner. Predicted totals were off by 24.1 points per team and finishes by 3.9 places, with a rank correlation of 0.56 between predicted and actual standings. Over 7 races per division, the ratings gave Stanford University a 36% chance to win, and standings at least this far from the prediction came up in 20% of 2,000 simulated regattas.
| Actual finish | Team | Predicted finish | Beat prediction by | Chance of this finish or better | Win chance | Points | Predicted points | Skippers and ratings going in |
|---|---|---|---|---|---|---|---|---|
| 1 | Stanford University Cardinal | 1 | +0 | 36% | 36% | 58 | 86.3 | A Thomas Sitzmann 469 B Vanessa Lahrkamp 484 |
| 2 | U. S. Naval Academy Midshipmen | 8 | +6 | 6% | 2% | 74 | 128.9 | A Nathan Smith 410 B Henry Allgeier 391 |
| 3 | Dartmouth College Big Green | 12 | +9 | 7% | 2% | 91 | 136.5 | A William Michels 386 B Chase Decker 388 |
| 4 | Yale University Bulldogs | 3 | -1 | 45% | 10% | 98 | 107.3 | A Jack Egan 430 B Stephan Baker 445 |
| 5 | Harvard University Crimson | 2 | -3 | 71% | 21% | 107 | 97.0 | A Justin Callahan 498 B Mitchell Callahan 417 |
| 6 | Brown University Bears | 6 | +0 | 51% | 6% | 113 | 114.9 | A Guthrie Braun 450 B Blake Behrens 401 |
| 7 | Tufts University Jumbos | 14 | +7 | 22% | 2% | 119 | 145.6 | A Ben Mueller 385 B Kurt Stuebe 359 |
| 8 | George Washington University Revolutionaries | 18 | +10 | 3% | <1% | 132 | 178.8 | A Tyler Wood 323 B Jedidiah Bechtel 297 |
| 9 | College of Charleston Cougars | 4 | -5 | 75% | 7% | 135 | 113.3 | A Noah Zittrer 440 B Benjamin Dufour 413 |
| 10 | Boston College Eagles | 5 | -5 | 79% | 7% | 136 | 114.1 | A Peter Busch 426 (7 races) B Jack Redmond 429 (6 races) B Michael Kirkman 416 (1 races) |
| 11 | University of Rhode Island Rams | 9 | -2 | 61% | 1% | 140 | 134.2 | A Kerem Erkmen 455 (6 races) B Tyler Nash 310 (4 races) B Christopher Chwalk 285 (3 races) |
| 12 | Tulane University Green Wave | 7 | -5 | 89% | 6% | 142 | 115.9 | A Kelly Holthus 407 (6 races) B Hamilton Barclay 403 (4 races) B Christian Ebbin 414 (3 races) |
| 13 | University of Miami Hurricanes | 13 | +0 | 61% | 1% | 148 | 143.7 | A Atlee Kohl 408 B Aidan Dennis 340 |
| 14 | Bowdoin College Polar Bears | 17 | +3 | 54% | <1% | 149 | 153.8 | A Thibault Antonietti 345 B Sam Bonauto 367 |
| 15 | Georgetown University Hoyas | 11 | -4 | 88% | 1% | 151 | 134.9 | A Piper Holthus 379 (3 races) A Enzo Menditto 377 (2 races) A Mateo Di Blasi 429 (2 races) B Peter Barnard 383 (6 races) B Diego Escobar 395 (1 races) |
| 16 | St. Mary's College of Maryland Seahawks | 10 | -6 | 93% | 1% | 168 | 134.7 | A Owen Hennessey 429 (7 races) B Landon Cormie 343 (4 races) B Charlie Anderson 363 (3 races) |
| 17 | Hobart and William Smith Colleges Statesmen | 15 | -2 | 93% | <1% | 174 | 150.4 | A James Kopack 285 (3 races) A Juan Carlos Lacerda Jones 336 (3 races) B Jj Klempen 365 (7 races) |
| 18 | Massachusetts Institute of Technology Engineers | 16 | -2 | >99% | <1% | 206 | 152.6 | A Sam Bruce 403 (7 races) B Julius Heitkoetter 307 (5 races) B William Kulas 330 (2 races) |
The ratings picked Yale University; Stanford University won. Predicted totals were off by 37.1 points per team and finishes by 2.4 places, with a rank correlation of 0.85 between predicted and actual standings. Over 16 races per division, the ratings gave Stanford University a 30% chance to win, and standings at least this far from the prediction came up in 57% of 2,000 simulated regattas.
| Actual finish | Team | Predicted finish | Beat prediction by | Chance of this finish or better | Win chance | Points | Predicted points | Skippers and ratings going in |
|---|---|---|---|---|---|---|---|---|
| 1 | Stanford University Cardinal | 2 | +1 | 30% | 30% | 198 | 183.8 | A Vanessa Lahrkamp 476 (16 races) B Ellie Harned 350 (8 races) B Sophie Fisher 370 (8 races) |
| 2 | Tulane University Green Wave | 5 | +3 | 13% | 4% | 209 | 247.0 | A Samantha Gardner 392 B Ava Anderson 333 |
| 3 | Harvard University Crimson | 6 | +3 | 12% | 1% | 223 | 271.4 | A Cordelia Burn 341 B Zoey Ziskind 345 |
| 4 | Yale University Bulldogs | 1 | -3 | 98% | 58% | 250 | 165.2 | A Emma Cowles 430 (8 races) A Mia Nicolosi 463 (8 races) B Carmen Cowles 422 (16 races) |
| 5 | Cornell University Big Red | 3 | -2 | 64% | 4% | 255 | 240.6 | A Bridget Green 424 (12 races) A Winborne Majette 353 (4 races) B Sophia Devling 330 (16 races) |
| 6 | Boston College Eagles | 10 | +4 | 14% | <1% | 277 | 323.8 | A Caroline Sibilly 363 B Sara Schumann 244 |
| 7 | Georgetown University Hoyas | 4 | -3 | 85% | 4% | 284 | 241.0 | A Piper Holthus 379 (16 races) B Emily Doble 357 (8 races) B Kelly Bates 352 (8 races) |
| 8 | Bowdoin College Polar Bears | 12 | +4 | 27% | <1% | 287 | 332.2 | A Kyra Phelan 305 B Lauren Russler 289 |
| 9 | Brown University Bears | 7 | -2 | 61% | <1% | 294 | 295.7 | A Katharine Doble 327 (16 races) B Katherine Mcnamara 324 (10 races) B Laura Hamilton 319 (6 races) |
| 10 | Dartmouth College Big Green | 9 | -1 | 60% | <1% | 297 | 310.3 | A Sarah Young 348 (14 races) A Bella Casaretto 341 (2 races) B Olivia Drulard 281 (16 races) |
| 11 | College of Charleston Cougars | 11 | +0 | 56% | <1% | 307 | 324.4 | A Emma Tallman 331 B Emily Alfortish 278 |
| 12 | George Washington University Revolutionaries | 17 | +5 | 12% | <1% | 314 | 401.5 | A Avery Canavan 239 B Arrieta Angueira Salbidegoitia 241 |
| 13 | Massachusetts Institute of Technology Engineers | 8 | -5 | 87% | <1% | 320 | 303.9 | A Brooke Schmelz 323 B Lucy Brock 313 |
| 13 | Northeastern University Huskies | 15 | +2 | 50% | <1% | 320 | 357.2 | A Eva Ermlich 276 B Lucia Loosbrock 279 |
| 15 | Roger Williams University Hawks | 14 | -1 | 81% | <1% | 371 | 345.6 | A Lucy Meagher 334 (16 races) B Tavia Smith 256 (10 races) B Katherine Mcgagh 211 (6 races) |
| 16 | University of Pennsylvania Quakers | 13 | -3 | 94% | <1% | 388 | 334.2 | A Sofia Segalla 333 B Adra Ivancich 259 |
| 17 | University of South Florida Bulls | 16 | -1 | 84% | <1% | 426 | 384.7 | A Kay Brunsvold 302 (15 races) A Kailey Warrior 211 (1 races) B Kalea Woodard 218 (14 races) B Heidi Hicks 179 (2 races) |
| 18 | Tufts University Jumbos | 18 | +0 | >99% | <1% | 453 | 414.0 | A Maisie Macgillivray 199 (6 races) A Meredith Broadus 207 (6 races) A Kiana Beachy 196 (4 races) B Sophia Hubbard 251 (16 races) |
The ratings picked the winner. Predicted totals were off by 36.9 points per team and finishes by 1.9 places, with a rank correlation of 0.91 between predicted and actual standings. Over 20 races per division, the ratings gave Point Loma High School a 43% chance to win, and standings at least this far from the prediction came up in 70% of 2,000 simulated regattas.
| Actual finish | Team | Predicted finish | Beat prediction by | Chance of this finish or better | Win chance | Points | Predicted points | Skippers and ratings going in |
|---|---|---|---|---|---|---|---|---|
| 1 | Point Loma High School Pointers | 1 | +0 | 43% | 43% | 266.6 | 241.4 | A Ian Nyenhuis 412 B Anton Schmid 417 |
| 2 | Antilles School Hurricanes | 3 | +1 | 27% | 15% | 270 | 299.8 | A Tanner Krygsveld 427 B Cobia Fagan 280 |
| 3 | Ransom Everglades School Raiders | 6 | +3 | 22% | 3% | 294 | 326.0 | A Griggs Diemar 389 (20 races) B Sebastian Van De Kreeke 288 (16 races) B Ava Mc Aliley 234 (4 races) |
| 4 | Severn School Admirals | 4 | +0 | 50% | 7% | 301 | 301.3 | A Harrison Szot 339 (12 races) A Alex Baker 330 (8 races) B Annie Sitzmann 365 (20 races) |
| 5 | Christchurch School Seahorses | 5 | +0 | 60% | 7% | 307.8 | 304.7 | A Bo Angus 367 (16 races) A Madeline Janzen 254 (4 races) B Wylder Smith 352 (20 races) |
| 6 | Lucy Beckham High School Bengals | 9 | +3 | 23% | 1% | 326 | 380.7 | A James Pine 359 B Nathan Pine 241 |
| 7 | Mater Dei High School Monarchs | 2 | -5 | 96% | 22% | 327 | 264.1 | A Tate Christopher 403 (20 races) B Brady Kennedy 343 (10 races) B Noah Stapleton 349 (10 races) |
| 7 | St. George's School Dragons | 11 | +4 | 28% | <1% | 327 | 392.9 | A Gil Hackel 310 (20 races) B Amelon Rule 287 (18 races) B Kai Watters 160 (2 races) |
| 9 | Tabor Academy Seawolves | 10 | +1 | 48% | 1% | 345 | 392.4 | A Peter Herlihy 339 (20 races) B Jack Spillane 257 (16 races) B Perrin Mueller 204 (2 races) B Sander Skaane 197 (2 races) |
| 10 | The Hotchkiss School Bearcats | 7 | -3 | 71% | 1% | 351 | 367.8 | A Pierce Olsen 337 (20 races) B Thomas O'Grady 265 (11 races) B Fynn Olsen 298 (9 races) |
| 11 | Southern Regional High School Rams | 12 | +1 | 54% | <1% | 407 | 417.5 | A Turner Ryon 271 (16 races) A Gannon Botwinick 220 (4 races) B Jude Ryon 293 (20 races) |
| 12 | Corona del Mar High School Sea Kings | 13 | +1 | 61% | <1% | 443 | 427.8 | A Michael Sentovich 275 (20 races) B Maddie Nichols 289 (12 races) B Siena Nichols 269 (6 races) B Jonah Moore 123 (2 races) |
| 13 | Christian Brothers Academy Colts | 8 | -5 | 95% | <1% | 444 | 374.0 | A Christopher Small 319 B Cole Buczkowski 287 |
| 14 | Jones College Prep Eagles | 16 | +2 | 24% | <1% | 445 | 526.0 | A Nissa Berman 216 B Grace Renz 203 |
| 15 | Wayzata High School Navy | 18 | +3 | 7% | <1% | 544 | 587.7 | A Dominik Moncur 227 B Stonewall Anderson 69 |
| 16 | Arrowhead High School Warhawks | 14 | -2 | 91% | <1% | 547 | 469.6 | A John Lieber 246 B Nicholas Berkowitz 246 |
| 17 | Lake Forest High School Scouts | 15 | -2 | 85% | <1% | 559 | 515.4 | A Mason Keane 230 (13 races) A Grady Strothman 261 (4 races) A Owen Kohut 331 (3 races) B Keegan Chatburn 218 (14 races) B Maddie Rode 98 (4 races) B Jackson Schwartz 61 (2 races) |
| 18 | Olympia High School Bears | 19 | +1 | 52% | <1% | 603 | 600.6 | A Alan Timms 176 B Liam Taylor 141 |
| 19 | Clear Lake High School Falcons Green | 17 | -2 | 83% | <1% | 621 | 578.2 | A Sydney Small 234 B Casey Small 111 |
| 20 | Roosevelt High School Rough Riders | 20 | +0 | >99% | <1% | 643 | 629.6 | A Roan Olson 142 B Ethan Lee 128 |
Predicted versus actual standings at every championship
The model predicts each race from ratings alone, as if every race were a fresh draw. Real regattas are not like that: a sailor who beats their rating in one race tends to beat it again. To measure it, each championship skipper's result in every race was scored against their rating going in (places better or worse than expected, as a share of the fleet), then compared race by race, with up to 4,440 sailors behind each pair of races. If races were independent every cell below would be about zero.
Every pair of races is positively correlated, 0.16 on average. Back-to-back races are the most alike (0.24) and the link fades with distance, to 0.09 for races 11 apart, so part of it is form or conditions that change through the regatta. Races also come in pairs: back-to-back races within the same pair (1–2, 3–4, …) average 0.27, while back-to-back races across a pair (2–3, 4–5, …) average 0.19. Rotations show why. Back-to-back races sailed in the same boat correlate at 0.27, back-to-back races in different boats at 0.18. The boat matters even when races are far apart: 2 races apart, the same boat gives 0.35 and a different boat 0.18.
So yes, winning race 1 means something. Across 265 championship divisions, the race-1 winner went on to average 4.4 in later races. Their rating alone predicted 5.2; updating on race 1 predicts 4.7.
Using it. If each sailor's strength is nudged by how they have sailed so far in the regatta, the order of later races is easier to predict. Tuned on 2024–25 championships and tested on 2025–26, pairs ordered correctly rose from 72.9% to 74.5% and log-loss fell from 0.534 to 0.514. That is a bigger gain than the site model's whole edge over Elo.
In the simulations. The championship simulations and the regatta analysis pages now build these correlations into every simulated regatta instead of treating each race as a fresh draw. Each sailor gets a good or bad regatta, form that drifts from race to race, and the speed of whichever boat the rotation puts them in, shared with every team that sails that boat. The three sizes were fitted to the correlations above. How big the swings should be depends on how well a sailor's rating is known, because they also cover the rating being wrong. For skippers with more than 200 races before the event they are 0.6 times the fitted size; for 61–200 races 0.7; for 1–60 races 1.3; for newcomers 2.0. Those sizes were chosen on the 2024–25 championships and checked on 2025–26 (the newcomers still finish worse than their starting rating suggests, which bigger swings cannot fix). Before both changes, team totals needed much smaller swings than individual sailors did: championship winners are mostly experienced skippers with well-known ratings, and extra swings were also flattening favourites whose ratings were already too bunched. With both, team totals chose 100% of the fitted size. See how likely was each result.
When the model says a sailor has a 70% chance of finishing ahead of another, that should happen about 70% of the time. Points on the diagonal mean the stated probabilities can be taken at face value. Up to about 85% confidence the points sit within 4.2 percentage points of the diagonal. Above that the model is slightly overconfident: when it says 97%, the favourite finishes ahead 90% of the time.
| Confidence bin | Mean predicted | Observed | Pairs |
|---|---|---|---|
| 50–55% | 52.5% | 52.3% | 144,554 |
| 55–60% | 57.5% | 56.1% | 145,299 |
| 60–65% | 62.5% | 60.5% | 141,952 |
| 65–70% | 67.5% | 65.0% | 136,003 |
| 70–75% | 72.5% | 69.8% | 134,944 |
| 75–80% | 77.5% | 73.8% | 131,363 |
| 80–85% | 82.5% | 78.3% | 130,268 |
| 85–90% | 87.5% | 82.9% | 121,292 |
| 90–95% | 92.5% | 87.6% | 120,611 |
| 95–100% | 97.0% | 90.1% | 92,542 |
An OCS or a DNF says little about boat speed, so boats that did not finish are left out of each race's finishing order. They never count as losing to every boat that finished, and they are left out of the accuracy numbers above. Penalties are predicted separately: each skipper gets a per-race chance of a start penalty, a DNF and a DNS, pulled toward the league rate by an amount tuned on the previous season, so a few penalties in a handful of races do not brand a sailor. The tenth of race entries it rated most penalty-prone had a penalty 2.9% of the time, against 2.0% for everyone else.
| Outcome | Share of race entries | Log-loss vs league rate | Championships: share | Championships: vs league rate |
|---|---|---|---|---|
| Start penalty (OCS) | 0.26% | +0.1% | 0.32% | +0.3% |
| Did not finish (DNF, RAF) | 0.66% | +1.0% | 0.41% | +3.3% |
| Did not start (DNS) | 1.15% | +1.6% | 0.70% | +8.0% |
| Any of these | 2.07% | +1.6% | 1.43% | +5.2% |
“Log-loss vs league rate” is how much better the penalty scores predict than giving every skipper the league's average rate, season-ahead and at championships. Penalties are rare and only loosely a habit, so these gains are small; did not start is the most predictable.
The trainer holds out every 20th regatta (334 regattas, 3,588 races) and prints its accuracy on them. Re-scoring that split with this evaluator should reproduce the trainer's numbers, which confirms the metric code. These figures are optimistic for the two Plackett–Luce models: the weekly model smooths ratings across time, so a held-out race's rating is informed by races that came after it. Elo and average finish only use earlier weeks here.
| Model | ICSA accuracy | ICSA log-loss | ISSA accuracy | ISSA log-loss |
|---|---|---|---|---|
| No information (coin flip) | 50.0% | 0.6932 | 50.0% | 0.6932 |
| Average finish percentile | 63.3% | 0.6333 | 67.4% | 0.5981 |
| Elo (multi-player, tuned K) | 68.1% | 0.5895 | 70.9% | 0.5644 |
| Plackett–Luce, static rating | 71.1% | 0.5486 | 73.5% | 0.5278 |
| Plackett–Luce, weekly, before the rating fix (previous site model) | 70.9% | 0.5585 | 73.7% | 0.5317 |
| Plackett–Luce, weekly (site model) | 71.7% | 0.5425 | 74.6% | 0.5122 |
| Trainer's own printout (weekly PL) | 71.68% | 0.5425 | 74.61% | 0.5122 |
| Forecast season | Tuned on | Elo K | Avg-finish shrinkage | Static PL L2 λ |
|---|---|---|---|---|
| Fall 2024 | Spring 2024 | 64 | 3 | 3 |
| Spring 2025 | Fall 2024 | 64 | 20 | 6 |
| Fall 2025 | Spring 2025 | 128 | 10 | 3 |
| Spring 2026 | Fall 2025 | 64 | 20 | 6 |
python3 analysis/plackett_luce/evaluate_models.py
Generated 2026-09-15 from 102,091 races and 24,501 sailors (2008-W38 to 2026-W37). Full run 91 minutes.
