What the data actually says about sport.
Long-form analysis from inside the MyAthlete database — plus the technical whitepapers behind our training and prediction engines. We start with Irish competitive swimming.
From the database
Long-form breakdowns of real race results — starts, finals, and pacing, with real numbers and no anecdotes.
Reaction times: what 110,000 starts really decide
When do the blocks decide races, and what actually makes a start quick? 13,000+ heats and same-swimmer tests take on lanes, pressure, and growing up.
Read the full analysis →Finals: what 36,777 second swims really reveal
When do finals go faster than prelims, and does anyone learn to lift? 36,000+ paired swims take on age curves, headroom, and the myth of clutch.
Read the full analysis →The pacing paradox: what 115,000 splits really decide
Does going out fast ruin your race, or does negative splitting surrender it? 115,000+ splits take on personal rhythm corridors, pacing discipline, course purity, and how swimmers actually evolve over time.
Read the full analysis →Under the hood
The science behind MyAthlete — how we model training load and predict performance across events. Essential reading for coaches and sport-science nerds.
Sesh 3.0: the true cost of every swim set
How MyAthlete scores every rep — computer-vision set parsing, multi-exponential recovery kinetics, and phenotypic stroke decoupling. Distance is an illusion; this is what replaces it.
Read the whitepaper →Project ATLAS: unified longitudinal & cross-event synthesis
Historical event affinity, multi-event progression bridges, and relative-freshness gating — how MyAthlete solves the dormant event paradox and uncovers major personal best breakthroughs.
Read the whitepaper →Project M11: cross-event transfer & sparse imputation
Allometric power-law scaling, latent factorization, and specialist routing — the original research that established zero-shot imputation across missing strokes and courses.
Read archived paper →