Stats tell you what happened. SIQX scores what the hitter decided — what he swung at, what he let go, how he handled two strikes — and puts that next to his production so you can see where the two disagree, before the market catches up. Every pick comes from swing decisions, not box scores.
Built from Statcast pitch-by-pitch data. Ten measures for hitters, nine for pitchers, every qualified player, updated daily.
By Stephen Bridwell · Sr. Applied Scientist · M.I.D.S., UC Berkeley
A team can win on bad process, or coach a clean game and lose anyway. SIQX scores the decision itself — when a staff goes for it, how fast it snaps, whether it beats itself before the snap, how a quarterback handles pressure — and measures it for every team in the league. Then it shows you that next to what the market expects, so you can see where the two disagree.
Built from nflverse play-by-play and FTN charting. Fourteen team tendencies and four quarterback measures, every team, updated weekly. See this week’s board.
Where a quarterback’s decisions — interception-worthy throws, throw-aways, sacks he caused, leaving the pocket — and his production disagree, and by how much. Decision and production are league percentiles, 0–100, where 50 is the median quarterback and 80+ is the top fifth; the gap is production minus decision. A gap describes two lists today, not a forecast. Early in the season the samples are small.
SIQX Football measures what a staff chooses to do in defined situations, from nflverse play-by-play and FTN charting, each as a league percentile among the 32 teams. It does not assume which direction is smart: whether blitzing more or punting less is good depends on the roster and the opponent, so nothing here is graded and nothing is summed. Darker on the Analytics page means more of a decision, never better.
Excluded on purpose: completion percentage over expected, air yards, success rate and EPA. Those are execution or outcome. SIQX Football measures the choice, then puts the result in the next column.
The same outcome can result from excellent decisions or poor decisions that happened to work. SIQX identifies which is which — before the results correct.
Football SIQX asks one question: do decision metrics carry information the closing line does not already price? The answer is being settled the only way it can be — a live record, every pick posted before kickoff, graded in public against a checkpoint set in advance.
Full research paper with component definitions, behavioral-validity audit, bullpen separation methodology, validation framework, limitations, and the live pick record. Pick-model selection logic is proprietary and not published.
Working paper covering SIQX v12 (hitters), PSIQX v2 (pitchers), the bullpen model, the validation framework, multi-season validation (2023–2026), and threats to validity. The pick model is described at a high level only — selection logic and thresholds are held privately. Correlations in the paper predate the September 2026 FIQ correction; the current figures are +0.31 on 1,207 hitter-seasons and −0.71 on 1,334 pitcher-seasons (PA≥150, 2023–2025 closed). A revised edition is pending.
Criteria are locked before a result is seen and do not move after. Five backtests over sixteen closed seasons, one held out at a time, did not find the decision components adding to the closing line, so the live record is the test that counts. Its checkpoint is 250 picks: units positive, with the 90% lower bound above zero. The backtests are re-run, criteria unchanged, when the 2026 season closes.
The design document fixes the components, windows, controls and pass criteria. Component definitions and the test design are public; the pick models are not. Quarterback prop projections move 0.22 yards per unit of decision quality where about 4.6 are needed to beat a −110 line — measured before the feature was built, and stated here so the passing-yards market is not oversold.
The same discipline that powers SIQX — identifying signal in behavioral data at scale — is the foundation of my professional work in machine learning systems. SIQX applies that expertise to baseballfootball: measuring not what players and coaches do, but how they decide.
Daily pick cards, team analysis, profile-vs-results gaps, and weekly SIQX reports — all on Twitter.
The week’s posted picks, quarterback divergence, team tendencies, and the record as it stands — all on Twitter.
Follow @SIQXindex →SIQX is still in open validation — every pick is posted publicly, win or lose, and the record is building toward a pre-registered checkpoint. Leave your email and you’ll hear when picks go out by email. No picks are being emailed yet.
Football picks are posted publicly every week and graded win or lose against a checkpoint set in advance. Leave your email and you’ll hear when the record clears it. No picks are being emailed.
No spam, no selling your address, unsubscribe any time. Nothing here is betting advice.