Welcome. If this is your first visit, the homepage is the Starting 10, but it isn't the whole 1500 meters. The site has a few dozen pages, all built on the same model — a hierarchical Bayesian fit to every result we have on both sides of the Atlantic. This is a tour.
If you have not read the methodology piece yet, read that first. The short version: every crew gets an RPI rating and an honest measure of how confident the model is in it.
The rankings — your starting point
Open the boat-class menu and you'll find four rankings: men's V8+, women's V8+, men's 2V8+, and women's 2V8+. Pick the one you care about.
Each ranking shows the top crews by RPI, with the 90% credible interval next to it as a small bar chart. Crews with fewer than three rated races this season are excluded from the public list — the model has an opinion on them, but not yet one it considers defensible at this level. You can filter by league, region, or country, which is the most useful thing on the page if you're trying to compare an Atlanta crew to a New England crew without leaving the table.
Click any crew name to open their team page.
The team page — where the model shows its work
Every team page lays out the same things: the crew's RPI distribution as a curve, a season-long trajectory chart, every result the model has scored, the head-to-head record against any opponent it has shared a course with, and the championship win-probability heatmaps if the crew is eligible.
If you spot a crew somewhere on the site labelled "provisional", the team page is where you'll see the model's reasoning: not enough races yet, wide credible interval, hold this rating loosely. The model will rerate them as more results come in.
The matchup tool — settling specific arguments
The matchup tool is the page you go to when you've already had the argument. It compares two crews directly: RPI gap, distribution overlap, common-opponent record, win probability, and a chart of how each crew's rating has moved over the season.
Head-to-Head Record
Most-raced opponents — win/loss based on finishing position in shared races
| Opponent | W | L | Record |
|---|---|---|---|
| Row America Rye | 6 | 1 | 86% |
| Community Rowing | 7 | 0 | 100% |
| St. Joseph's Prep | 4 | 3 | 57% |
| Saugatuck | 6 | 0 | 100% |
| PNRA/Mercer | 6 | 0 | 100% |
| Central Catholic | 6 | 0 | 100% |
| D.C. National | 4 | 0 | 100% |
| Pelham Community Rowing Association (PCRA) | 4 | 0 | 100% |
| Duxbury | 4 | 0 | 100% |
| Gonzaga | 4 | 0 | 100% |
Greenwich vs most-faced opponents across all recorded races
The win probability is calibrated. When the page says 70%, the model means it: across our backtested matchups, crews we gave 70% odds to won roughly 70% of the time. The methodology page publishes the calibration plot, including the model's known overconfidence in the 55–70% band.
The championship simulators — what 10,000 race-throughs say
The championship predictions page gets the most use in the last week before each major regatta. We run Monte Carlo simulations of the five tier-one championships — Stotesbury Cup, SRAA Nationals, National Schools, Youth Nationals, and Henley Royal Regatta — sampling each crew's RPI distribution, advancing through the actual bracket structure, and reporting the percentage of simulations in which each crew makes the semifinal, the final, and wins the title.
The model runs ten thousand simulations per boat class. The output is a table of contenders with three numbers each. If you have ever wanted to know what the favourite's actual win probability is — usually less than the seeding implies, sometimes a lot less — this is the page that will tell you.
The regatta simulator — build your own race
The regatta simulator is the more freeform sibling of the championship page. Pick any crews from the rankings, set a distance, and run the same Monte Carlo against your hand-picked field. The page returns podium percentages and a heat-map of finish-order distributions. The gap between "a crew the model thinks is the favourite" and "a crew the model thinks wins more than half of the time" is real, and the simulator is honest about it.
Predictions — three flavours
The site has three different ways for a logged-in user to make predictions and have them scored.
Pick'em scores you on individual race outcomes. The leaderboard runs all-time and rolling-30-day, sliced by boat class. If you can consistently beat the model on, say, women's V8+, that fact will become extremely visible.
Prop bets are season-long propositions ("Crew X finishes top three at SRAA"). The crew submitting them is mostly us, with some user-proposed props going through review.
Crew coins are a parimutuel pool — a play-money market in which users buy and sell positions on race outcomes. The most interactive part of the site, and the part most likely to teach you something about your own confidence calibration. The model is in the market too, posting prices it considers fair, which means a poorly-calibrated user will lose to it over time and a well-calibrated one will beat it.
Bracket pools — fill out a championship
The bracket pools page opens up at the start of each tier-one championship week. Pick the field heat-by-heat or matchup-by-matchup, and your accuracy is scored against the actual bracket result. The leaderboard runs per bracket.
If you have ever wanted to put your "Greenwich is going to take the SRAA" instinct on the record, this is where you do it.
Power 10 — the editorial archive
The page you're reading this on. Power 10 is the writing layer on top of the numbers — ranking pieces, championship previews, season reviews, and methodology posts. Pieces are tagged so you can read your way through the parts of the site you care about. Expect a ranking piece every couple of weeks, previews in the days before each tier-one event, and a quieter run of methodology posts when something about the model has changed.
Teams, leagues, events — the directory pages
/teams is the alphabetical directory of every rated crew. /leagues organises the same set by conference and region, with championship odds laid out as gradient-coloured probability heatmaps — useful for spotting which leagues are over- or under-represented at the top of a boat class. /events is the searchable regatta catalogue.
Contributing — the back of the house
The rankings are only as good as the results behind them. If your crew raced a regatta that isn't yet in the database, the contribute page is where to start. Submitted results go through admin review before entering the system. If you're a coach or programme manager, the team-manager flow lets you claim your crew's page; that also goes through review.
How to actually engage
A few suggestions, without overthinking it.
Bookmark your boat class. The rankings update daily; check in once a week and you'll have a better view of the season than most people in the sport.
Use the matchup tool when an argument starts. It's faster than scrolling and more honest than either side's recall.
If a tier-one championship is within ten days, run the simulator before reading the seeding.
If you make predictions, log in — pick'em, prop bets, and bracket pools all score against the model and against other users.
If something looks wrong, tell us. The methodology page lists the model's known calibration weaknesses; the contact page is for the unknown ones.
Welcome to the Power 10. Let's row.
