From Metric to Platform

Building the behavioral
intelligence layer for baseball.

Two destinations, one foundation — a public, validated track record. Everything below is gated on it.

AMBITION 01

An industry-recognized metric

SIQX & PSIQX cited alongside DRC+, Stuff+, and OAA — open methodology, published validation, a predictive public record.

AMBITION 02

The AI insights platform

One behavioral engine across sports betting, fantasy, and player & team performance — delivered by push and a paid interactive agent.

One gate governs both. Nothing is sold until the model proves a real, positive-EV edge in public. The early stages buy credibility, not revenue — and credibility is the asset competitors can’t fake.

Stage 0
Validate the core
JUN – AUG 2026
GATE: win rate > implied AND positive EV at opening odds, over a pre-registered sample — Cleared at the 30-pick checkpoint (19–13, 59% vs 45% implied, +EV). That cohort closed at 19–16, 54.3% vs 45.2% implied over 35 resolved picks — the interim figure is shown because it was published at the time; the closing figure is the one that stands. A defect in team aggregation was found on Aug 24 and corrected; the pick log reset and validation restarts from zero on the corrected model.
  • ✓ DONE — count-split correction + refreshed correlations (PA≥150): r = +0.444 / −0.708
  • ✓ DONE — Sept 9: historical FIQ was a hard-hit proxy (an outcome measure); replaced with the fastball-specific FIQ the live model uses. Hitter correlation re-derived +0.444 → +0.31 (1,207 closed-season hitter-seasons); pitchers unchanged at −0.71. Same rule as June: re-measure after a score-side fix, never tune on outcomes.
  • ✓ DONE — v11→v12 decision; pick log reset, tracking live from June 10
  • Weighting revalidation at full sample — validated against xwOBA correlation only, never the pick record; any change is a new version on a fresh clock
  • Log every pick publicly — wins and losses, including the closed cohorts
  • Team aggregates weighted by playing time — a flat average counted a bench bat and an everyday regular equally, and an injured starter the same as one making every turn. Found through a public question, not the pick record, and corrected in the direction the evidence pointed. Pick log reset on a fresh clock.
  • ✓ DONE — 30-pick gate passed. The 50-pick checkpoint pre-registered against that cohort did not complete: the cohort closed at 35 when the aggregation defect was corrected. Not carried forward — a checkpoint registered against one model is not evidence about another.
Stage 1
Credibility & audience
2026 H2 (PARALLEL)
GATE: ~1,000 engaged followers + public track record live
  • Track-record page + reproducible methodology repo
  • Reply-driven X growth; enter SABR / FanGraphs community
  • Methodology white-paper draft
Stage 2
First paid product: push
POST-VALIDATION
GATE: sustained +EV over a second independent block
  • Daily card + reasoning via Discord / Substack tiers
  • Free tier keeps yesterday’s card; paid unlocks today’s edge
Stage 3
Expand bet types
2026 H2 → 2027
GATE: each product validated independently (30+ sample, +EV)
  • Props and run totals — under research; each gated on its own independent Step-0 + live validation
  • Parlays last — only after the legs are proven
Stage 4
The paid agent
BUILD 2026/27 · LAUNCH 2027
GATE: retention / usage that justifies cost
  • Interactive SIQX/PSIQX agent on siqx.net — ask it anything
  • Freemium → premium: live edges, props, unlimited queries
Stage 5
Durable B2B revenue
2027
After the metric earns recognition
  • Data/API licensing to media & tools
  • B2B player-evaluation & scouting reports; team/org consulting; DFS
Stage 6
Industry-metric status
2027 → 2028+
THE NORTH STAR
  • White paper, conference talks (SABR / Saberseminar), media citations
  • Team adoption; license the metric; pursue Statcast/MLB partnership