Loading
MBStatsLoading
Loading
MockingBoard Stats
Every weekly stat is part signal, part luck, and we measured which is which. Metrics built where the public ones fall short, 27 seasons of play by play to check our work, charts worth sharing, and plain English questions answered by executed queries, not model guesses.
Free. No account needed.
Sign in to save charts and ask Songbird.
Live from the workbench
Every chart below is a real chart spec, resolved against the same data the explorer gives you and drawn by the same renderer that draws every chart on this site. Open one in the builder and it is yours to change.
Lines with rolling means, league average annotations, minimum volume gates, and season facets are all part of the spec language. If you can describe the chart, the builder can draw it.
Built here
This is not just a window onto public data. The metrics with our name on them were built from split-half reliability studies, walked through absurdity tests, and shipped with their noise numbers printed on the label. When a stat is mostly luck, we say so.
The weekly score / QB, RB, WR, TE
For quarterbacks, Altitude is EPA per dropback with the luck priced out: one game of interceptions is about one percent signal, so interception plays are re-priced as incompletions and sack fumbles as clean sacks. For receivers and backs it is earned yards, opportunity priced at league rates, because weekly conversion is almost pure noise. Measured game-to-game, the QB score carries about half again the signal of passer rating, and the numbers behind that claim are published where you can check them.
The workbench
Ten seasons of player, team, and situational splits in a filterable, sortable table. Every column carries its definition. It all runs in your browser, which is why it is fast and why it is free.
Open the explorerScatters, lines, bars, distributions, grids, and facets, drawn in team colors with league average reference lines. Every chart is a typed spec you can save, fork, share, and export.
Open the builderTwelve curated procedures, resampling first, with effect sizes and intervals leading the answer. "This difference is consistent with noise" is a designed result here, not a failure state.
Try the test wizardThe deal
The workbench stays free because exploration is the input. Pro covers hosted analysis and production tools built for repeated use, including Songbird and live OBS overlays. That is the whole pricing model, stated plainly.
27 seasons, 1999 through 2025. All of it on your machine.
Learn the language
What EPA actually measures, how CPOE is computed, when a stat stabilizes, and how each one gets misused. Seventeen Stat Cards, written for football people, linked from the builder and cited by Songbird.
Read the glossaryWhere players will go, then where they should go. The mock blends insider finals into consensus at a weight validated on seven drafts; the board prices value against the pick curve with research-backed tilts.
Read the boardsEvery scored free agent carries two scenarios, re-sign and external buy, because the market prices them differently: external signings bust at twice the incumbent rate. Destination odds render only where the model earned them.
Open the classType a question. The model translates it into a query, the database executes it, and every number in the answer comes from the executed result. The query itself is one click away.
Ask SongbirdThe one MockingBoard subscription, shared across every MockingBoard app.
Exports carry a small MockingBoard mark on every tier. Your saved charts stay yours whether you ever pay or not.
Built on nflverse data, with FTN charting attribution where used. Charts carry team colors and abbreviations, never logos. Data freshness lives at /status.