Beach Sprint Race Analyser

1. Choose what you timed. Time all seven phases if you can. With one stopwatch, split as the boat completes the turn and stop at the line: that recovers the two rowing legs well and is much better than splitting a total by fixed percentages. A finish time on its own cannot recover a breakdown, so that mode shows a typical profile and gives no training priorities.
2. Time your splits. Run 1 ends when a hand first touches the boat. T1 ends on the first stroke. Row 1 ends as the bow reaches the turn buoy. The turn ends when the boat is straight and driving for home. Row 2 ends at shore contact. T2 ends as the athlete leaves the boat. Run 2 ends at the line.
3. Pick a benchmark. The target is the median race profile of the fastest slice of real racing, not an average of winners. Competitive is the fastest quarter of all recorded performances; Championship is the fastest tenth.
4. Pick a comparison pool. All events is the most stable. A single championship is more specific but rests on fewer races.
5. Read the verdict column last. It combines how far you are off with whether that phase actually separates winners from losers, so a small gap on a decisive phase outranks a large gap on one that does not decide races.

Three things worth knowing before you act on this.

The seven phase targets come from the same group of real races, so they add up to roughly the target race time and are achievable together. That is deliberate. Taking the fastest recorded time for each phase separately would produce a race nobody has ever rowed, which is what the perfect race panel at the bottom is for.

Faster is not always better in the model. For women, quicker run 1 and row 2 times are associated with losing once the other phases are held constant, because the winning pattern is a controlled beach sprint that feeds a dominant outbound row. The tool will tell you when a phase behaves this way rather than pushing you to chase it.

For mixed pairs no phase predicts the result at all. That is a real finding, not missing data, and the tool says so rather than inventing a priority order.

Anything you did not time is marked est and carries an uncertainty range. A phase is only given a training priority if you measured it, or if the mode you chose estimates it to within half a standard deviation. Everything else is left unranked rather than guessed at.

Wind is deliberately not an input. Wind does slow female and mixed pair racing, but in this dataset it cannot be separated from venue and year: the 2022 championship was raced in strong wind for 97% of races. Conditioning the estimate on wind changes it by under a second, against an uncertainty range of twenty to thirty seconds, so it would add a control that looks meaningful and is not. Your own splits already carry the conditions you raced in.

Background that holds for everyone, independent of your own times. Findings from analysed races across five World Rowing championships.

Where the race is decided

Racing from behind

How much of the race is each phase