FIRELOCKS / MODEL & METRICS

Know what the numbers mean.

FIRELOCKS is built to make the research understandable without pretending sports outcomes are certain. These are the major evidence layers behind rankings, CTH, grades and Watch surfaces.

OPEN EDGE RANKINGS →OPEN WATCHLISTS →

01 / CORE PROBABILITY●

Calibrated Chance to Hit

CTH is the final probability layer shown on FIRELOCKS cards. It blends sport-specific model evidence with the available sportsbook market, then applies evidence and uncertainty controls. The exact blend varies by sport and market because a pitcher strikeout prop should not be calibrated the same way as a WNBA assists prop.

Sport- and market-specific model inputs
Exact-line sportsbook evidence when available
Uncertainty controls prevent unsupported confidence

02 / EVIDENCE QUALITY●

FIRELOCKS Grade

Grade and CTH answer different questions. CTH estimates the side probability; FIRELOCKS Grade describes how complete and trustworthy the supporting evidence is. A high probability with weak verification should not receive the same presentation as a high probability backed by exact market, matchup and history evidence.

Probability is not the same as evidence quality
Exact market and research completeness matter
Incomplete cards remain pending instead of being promoted

03 / MODEL SEPARATION●

Projection Edge

Projection Edge measures the distance between the current line and the model projection on the selected side. MORE requires projection above the line; LESS requires projection below it. FIRELOCKS does not count a large raw projection difference as an edge when the recommended direction disagrees.

Direction-aware separation
No credit for model/side disagreement
Used with probability, not as a standalone pick signal

04 / MARKET CONFIRMATION●

Sportsbook Market Quality

FIRELOCKS evaluates exact-line sportsbook quotes, two-sided pricing, book count, sharp-book presence, hold and disagreement across books. Where possible, prices are converted into no-vig fair probabilities before being used as evidence.

Exact same-line support is strongest
Two-sided prices are more informative than one-sided quotes
Book disagreement lowers confidence rather than being hidden

05 / CONTEXT●

Matchup + Opportunity

The research layer evaluates the factors that determine whether a player is likely to get enough chances to beat a line: opponent, role, minutes or snaps, usage, lineup position, game environment and sport-specific matchup context. Missing context stays visibly unverified.

Role and opportunity before narrative
Opponent and event identity must resolve
Context is sport specific, not one generic formula

06 / HISTORICAL EVIDENCE●

Recent Form + History

Recent results provide supporting evidence, not a guarantee that a streak continues. FIRELOCKS uses verified game values and line-aware hit rates where they exist, while avoiding fabricated history when a source is missing.

Line-aware L5/L10 research
Verified source values only
Streaks can support a case but do not override stronger evidence

07 / POWER + TD RESEARCH●

Specialty Scoring Signals

When the underlying data is available, specialty Watch surfaces can add sport-specific context. MLB power research can surface barrel quality and home-run drought context; NFL touchdown research can surface red-zone share, goal-line opportunity, expected-TD gaps and role-based splits.

Specialty metrics appear only when sourced
Drought alone is never treated as a reason something is due
Opportunity and price remain part of the decision

08 / FEEDBACK LOOP●

Calibration + Learning

Published recommendations and projections are evaluated after games complete. Wins, losses, pushes and unresolved rows remain separate so model evolution can be based on complete outcomes rather than winner-only screenshots.

Timestamp before outcome
Losses stay in the ledger
Model changes should be tested against completed results

TRUST STANDARD

No forced certainty.

Rankings, CTH, grades, projections and Watch lists are research outputs. If exact evidence is unavailable, FIRELOCKS should show pending, unavailable or no qualifying plays rather than inventing support. Users remain responsible for their own wagering decisions.

VIEW VERIFIED RESULTS →