v1 · "Line Agreement + Talent Gap"
2026-05-14 · superseded
Rules:
- Pick favorites in talent-mismatch matchups
- Use narrative factors (home court, momentum, must-win)
- $5 / $8 stake by confidence tier
- No skip discipline
BT1 (NBA Playoffs): +39.4% ROI · 16-4 record ·
missed blind benchmark by 4pp. full breakdown →
Lesson: same strategy fails across sports. Narrative factors hurt more than help — books already price them in.
v2 · "Selectivity + Sport-Specific"
2026-05-15 AM · superseded by v3
Added:
- HARD SKIP coin flips (probability < 56%)
- SKIP heavy juice (-200 or worse) unless clear talent gap
- FLAT $5 stakes (no confidence tiers — calibration unproven)
- No narrative-only picks
- Sport-specific calibration framework documented
BT3 (NBA Playoffs, retroactive): +59% ROI · 12-1 record on 13 bet / 7 skip · beat blind by +15.5pp.
BT4 (NBA Reg Season): +46% ROI · 7-1 record · 0% skip rate (sampling artifact).
full breakdown →
Lesson: selectivity > analysis quality. Same picks, fewer bets, much higher hit rate. Confidence calibration confirmed broken (STRONG bets actually underperformed MEDIUM).
v3 · "3-Tier Portfolio + BCE"
2026-05-15 PM · current
Added:
- 3-tier per-game portfolio (Tier B safe / Tier C correlated / Tier A longshot) — see /bets/portfolio/
- Blowout Correlation Edge methodology with 4 precise trigger conditions — see /bets/edges/blowout-correlation/
- Paper-bet validation mode — track everything, score outcomes, mark which were placed
- KV-backed prediction log with per-game outcome modal — see /bets/prediction-log/
- Methodology tagging on every bet (BCE_full_trigger / BCE_partial / v2_selectivity / correlated_2leg / etc.)
- Phase 1 validation plan: 30 BCE bets at $10 paper-stake, decision points at 10 and 30
Status: Phase 1 in progress. Bet #1 of 30 (SAS @ MIN G6) played Fri 5/15 — outcome TBD. Bet #2 (Pistons G7) is tonight Sun 5/17 8:00 PM ET — 4/4 BCE trigger conditions met. Live tracker at
2026-05-15-weekend.
Hypothesis being tested: correlated parlay edge of 5-15% over book line when 4 BCE conditions align. Target: 10-15% hit rate on $10 → ~$1,000 parlays = +$5,000-15,000/yr if pattern holds, -$1,500/yr worst case.
v4 · TBD (after Phase 1 completes)
~July 2026 · planned
Trigger to update:
- 30+ BCE bets scored with outcome data
- Any specific bet type at ≥55% hit rate over 30+ samples
- Any factor at ≥65% accuracy over 20+ scored outcomes
- MLB starting-pitcher data pipeline integrated (if budget for API)
Likely refinements:
- Refined BCE trigger conditions based on hit/miss patterns
- Per-sport stake calibration (NBA higher, MLB lower, Golf SameSHOT-only)
- CLV tracking as primary edge metric (instead of ROI)
- Possibly drop methods that haven't shown edge after 30 attempts
Reality check: over 40 games tested across NBA + MLB, my analysis hasn't beaten blind favorite-betting. Best case at year 1 is +$1,000. Most likely first year is break-even or slight loss. Real edge takes 3-5 years to confirm. v4 will reflect what the data actually says.
Strategy version bumps when:
- A major methodology component changes (e.g., adding skip discipline, new tier system)
- Backtest validates or invalidates a hypothesis at sufficient sample size (30+ bets minimum)
- A new sport-specific framework gets added (NHL / NFL when ready)
Append-only log — never edit prior versions. Old strategies stay documented for retrospective learning.