Statistical models trained on 470,000+ high school sprint & hurdle performances project where an athlete's times are headed – with an honest expected range, not a guess.
No new testing required. If you have season-best times, you have what's needed to start.
Share an athlete's (or team's) current-season bests – any format works.
Same-event and cross-event regressions estimate next-season performance, weighted by historical fit.
Projections, expected ranges, and fit labels are compiled into a coach-readable PDF.
Standard turnaround within 48 hours.
Coverage is built on the underlying historical dataset – these are the events and grade-to-grade transitions with enough athlete-year data behind them to support a reliable model.
Projections are not available for seniors entering college. The underlying dataset is built on high-school-to-high-school transitions, and we don't yet have the data needed to support a reliable projection across that jump. This may change as the dataset grows – check back if you're planning ahead.
Every event gets its own page: a same-event model, cross-event comparisons, an expected range, and a plain-language read of what it means. Reports are forward-looking – they project where a season is headed, not a comparison against results that haven't happened yet.
Illustrative example. Athlete and results are fictional.
Every projected event at a glance – current best, projection, expected range, and fit label.
Each event gets its own model comparison chart, showing every model considered – not just the one used.
For athletes already near the top of the field, a comparison built from similar-level peers – often more realistic than a population-wide model.
A written interpretation of what the models agree on and where they diverge.
This label summarizes how well the model has matched past athlete-season outcomes for that event. Higher values mean the model explained more of the historical variation, so those projections can generally be interpreted with more confidence; lower values mean the projection should be treated more cautiously.
Projection confidence depends on both the historical model and the athlete's available data. Models are usually stronger when they're built from well-populated events with many comparable athlete-seasons – common events like the 100m simply have more historical data behind them than rarer events. Individual projections are also more stable when an athlete has multiple performances across the season, rather than a single early-season mark. Even with a strong fit label, results should be interpreted as informed estimates, not guarantees.
R² thresholds above are used as plain-language interpretation bands. They describe historical model fit, not a guarantee of future performance.
Individual reports for a single athlete, or team packages with per-athlete pricing that scales down as your roster grows.
A focused look at one event – ideal for an athlete who wants a quick read on a single race.
The full picture across an athlete's main events – same-event and cross-event models, ranges, and fit labels.
A comprehensive multi-event report plus recruiting extras – built for athletes preparing to reach out to college programs.
Tailored for elite private training groups or small club rosters.
A predictable flat rate that's easy to get approved by an athletic director or booster club.
Need something custom? Get in touch for a quote.
Let us know which tier you'd like, and we'll email you back with exactly what to send – current times, meet history, or a team roster – followed by an invoice once we have everything we need.
Submit this short form with the report tier you're interested in.
We'll reply with exactly what to send – current times, meet-by-meet history, or a team roster – in whatever format is easiest for you.
Once we have what we need, we'll send a Square invoice. Your report is built and delivered within 48 hours of payment.
We'll email you to confirm before sending an invoice – nothing is charged automatically.
Submitting this form is not a payment – we'll reach out by email with next steps before any report work begins.
The plain-language version: we use thousands of athletes' year-over-year results to estimate what a similar athlete's next season tends to look like. Here's the fuller version.
The underlying dataset covers 470,000+ individual sprint and hurdle performances from western New York region high school meets, spanning many seasons and thousands of athletes. Each athlete-event-season is summarized into three derived features before modeling: season best (the fastest mark that season), season trend (the average direction and rate of change across the season's meets), and curvature (whether that change was steady, accelerating, or leveling off).
A same-event model uses an athlete's own history in that event to project the next season – e.g., this year's 100m informing next year's 100m. A cross-event model instead uses a related event – for example, the 55m and 100m together to project the 300m – on the logic that sprint and hurdle performances tend to move together. Cross-event projections are most useful when there's no same-event history to draw from, but they generally carry more uncertainty than a direct same-event model.
Next-season projections are estimated using multi-level (mixed-effects) regression models, which account for the fact that individual athletes develop at different rates – treating athlete as a random effect rather than assuming one population-wide growth curve. This applies to both same-event and cross-event models. k-means clustering is used separately to build peer comparison groups: athletes are grouped by current performance level so that an elite athlete is compared against athletes who were actually elite as juniors, not the full population.
Model fit is reported per event as R² against historical data, and translated into the plain-language Strong / Good / Fair / Limited fit label shown in the report. R² describes how well a model has matched past outcomes – it is a measure of historical fit, not a guarantee for any individual athlete's projection.
For athletes already near the top of the field, a population-wide model can understate their potential – the bulk of the data is athletes performing at a different level. The peer estimate instead draws from a smaller group of athletes who were within a narrow window of this athlete's current best, then reports what share of that peer group actually improved the following season and by how much. It's shown alongside the population model, not in place of it, so both reference points are visible.
These are statistical estimates based on historical patterns, not predictions or guarantees. They don't account for individual factors like injury, training load changes, coaching changes, weather, or race conditions. The expected range reflects the historical spread of outcomes for comparable athletes – a wider range means more year-to-year variability has historically existed for that event, not that the model is less carefully built. Projections are intended as a data-informed starting point for coaching conversations.
Current-season best times for the events you'd like projected, plus the athlete's transition (e.g., Junior → Senior). If you have meet-by-meet results rather than just a season best, the trend models can use those too – just send what you have.
For team packages, the "updated report" add-on re-runs the projection with an athlete's latest times, useful for checking how an early-season mark compares to what the models expected. For individual reports, get in touch and we can discuss re-running a projection with updated times.
No. These projections are a data-informed starting point for a coaching conversation, not a prediction or a promise. They don't know about injuries, training changes, or how an athlete is feeling this week – you do. Think of it as one more data point alongside what you already know.
Indoor: 55m, 300m, 55m Hurdles. Outdoor: 100m, 200m, 400m, 100m Hurdles, 110m Hurdles, 400m Hurdles. See the Events We Can Project section above for the full breakdown. Projections aren't currently available for seniors entering college – the dataset doesn't yet support a reliable projection across that transition.
Force plate assessments measure what an athlete can produce physically – power, asymmetry, force-velocity profile – right now. Projection reports use historical performance data to estimate where competition times are headed next season. They answer different questions and pair well together, but neither requires the other.
Each athlete receives their own individual PDF report. Both Small Squad and Full Roster licenses also include coach-facing summary notes covering the roster as a whole; add the team summary spreadsheet add-on for a sortable view across all athletes.
Send over current-season times and we'll confirm event coverage, turnaround, and the right report tier before anything is billed.
Get In Touch