Methodology
Process over noise
Built on Data. Driven by Edge.
We don't sell locks. We run a desk.
KosEdge prices games independently, reprices with real information, then measures our number against the market. Edge only exists when the separation clears our thresholds.
The core contract
Every serious number on this platform follows the same three-layer system:
- Model — independent research fair from the simulation / efficiency engine
- KEI — final handicap after information and situation (injuries, rest, weather, confirmation, market structure)
- Edge — KEI versus the best available market price
- Model is research.
- KEI is the board number.
- Edge is only vs the market — never model vs market for PLAY/LEAN.
If information is thin, KEI stays closer to Model.
If information is strong, KEI moves.
If price isn't there, we pass.
How a number is built
1. Inputs
Sport-specific data stacks, not one generic formula:
- team strength / efficiency
- roster and availability
- usage and role
- schedule / rest / travel
- environment (park, weather, venue)
- market context where useful
2. Projection
The model produces a research fair — spread, total, win probability, and where available player-level output.
3. Handicap (KEI)
We reprice the research fair with current information. This is the desk judgment layer. It is explicit, not hidden inside the sim.
4. Market compare
KEI is compared to live books. Separation becomes candidate edge.
5. Thresholds
Minimum edge, price quality, volatility, and confidence filters decide what is actionable.
Close is not enough. Passing is a position.
What appears on the Edge Board
- KEI lines (the handicap)
- Market prices
- Edge / tags only when thresholds clear
- Honest empty or markets-only states when a sport is not fully modeled yet
We do not invent edges to fill a board.
Sport stacks, same standard
Same professional energy. Different factor emphasis.
- NFL — EPA/script, depth, injuries, coaching tendencies, survivor/fantasy paths
- CFB — roster construction, QB situation, efficiency backbone, HFA, early-season uncertainty
- CBB / NBA / MLB / NHL / WNBA — sport-native efficiency, availability, and market structure
Every sport is held to the same honesty bar: real model when ready, identity/fallback when not, no fake precision.
Tracking and accountability
We care about more than last night's result:
- projection vs open and close
- CLV as a diagnostic
- process grades (good bet / bad bet independent of outcome)
- continuous calibration where evidence supports it
CLV is useful. It is not the whole religion.
Long-term edge is the goal. Short-term noise is expected.
Discipline
- No forced action
- No chasing
- No parlay culture as process
- Bankroll and variance are part of the system
- Information has tiers — official and reliable sources outrank noise
If the number isn't there, the correct play is nothing.
What this is / isn't
This is
A modeling + handicapping desk with transparent process language and live tools.
This isn't
A lock service, a tipster feed, or a promise that every night pays.
We promise process.
We promise independent numbers.
We promise that betting with a real framework beats betting without one.
Go deeper
- Insights → Doctrine — house rules of the desk
- Edge Board — live numbers
- About — what KosEdge is building
Built on Data. Driven by Edge.