Our Mission

We're building Powerlifting Trends to put real data analysis directly into the hands of lifters and coaches: cohort-aware analysis that answers the questions people actually have, rather than aggregated leaderboards or equipment-mixed top-N lists.

The goal is simple. When a lifter looks at their own progress, or a coach plans a programming block, they can ground their decisions in what tens of thousands of comparable lifters have actually done, rather than what a single social media champion claims worked for them.

In-depth comparisons

New lifters get a skewed picture of what "normal" looks like. Social media algorithms are tuned to surface outliers, so the 14-year-old hitting a 700kg total goes viral while the median 90kg-class progression (the path 95% of lifters will actually live through) gets buried.

Career Arc and Lifter Lookup are designed specifically to push back on this. You see the typical Dots curve from debut through meet 30+ for your exact slice (event, equipment, weight class, sex), with p25/p75 bands so the realistic spread is visible. When you compare your own debut, you're comparing it against other debuts, rather than the seasoned lifters whose meet videos dominate your feed.

Where the data comes from

Every entry on the site is sourced from the OpenPowerlifting project's bulk CSV release. We pull the dataset, run it through a multi-stage ingest pipeline (deduping lifters, normalizing meet metadata, computing career trajectories), and store it in our own database.

On top of OpenPowerlifting's raw data we expand with additional information we believe is helpful to lifters: career trajectories, percentile ranks within slice, debut-to-current deltas, and cohort comparisons keyed on the specifics of your event, equipment, weight class, and sex. The API sits behind CloudFront with query-aware caching, so the same cohort question hit twice in an hour comes back instantly.

OpenPowerlifting is open data and has been for years. We're built on top of their work and link back to their site throughout. None of this exists without them.

Statistical honesty

The Analysis tab runs Pearson correlation and pairwise linear regression and reports an effect-size-first verdict. With 3.9M rows, anything you regress will hit p < 0.001, so framing a regression as "very significant" because of the p-value alone is misleading. We gate the verdict on |r| first ("negligible" if |r| < 0.2 regardless of p), then on significance, then describe the result in plain English.

Where a relationship is genuinely weak we say so, even when it's "statistically detectable." The point is to give you a useful read, rather than to inflate findings.

The roadmap

We plan to train our own machine learning models on this dataset and use AI to deliver genuinely valuable insights to lifters and coaches: career-trajectory forecasts, peer-group comparisons that take more than a handful of variables into account, programming-decision signals grounded in what actually predicts longevity and peak strength.

The bar is statistical facts. Anything we ship will come with the same effect-size-first honesty we hold the existing regression cards to. If a model can't beat a sensible baseline, we won't dress it up as if it can.

Longer-term we also plan to open up the API to developers (regression endpoints, cohort- filtered trends, career-arc data) for anyone building tooling on top of the dataset. If that's of interest, drop us a note and we'll prioritize accordingly.