OSD / Studio / QK

QK — quality-of-kill scoring

How QK works

QK (quality-of-kill) is a composite score that ranks candidates by signal strength. It blends EV, sharp-consensus quality, line-stability, and other factors into a single number. Candidates with low QK get blocked even if their EV passes; candidates with very high QK can also be blocked (high QK often correlates with stale-line outliers).

The three threshold knobs

filters.minQk — global floor. Default 1.75.
filters.scopedMinQk — per-(mode, market, league, book, flow, sharpCount) floor overrides. Can go negative in scoped rules (e.g. -1.25) — that's intentional, not a bug. Negative QK floors mean "any quality, even very low, is acceptable for this scope."
filters.scopedMaxQk — per-scope ceiling. Caps suspiciously-high QK that often indicates stale-line traps or pricing anomalies.

Resolution order

Match scoped rules by mode + market + league + book + flow + sharpCount.
Most-specific match wins; ties → more restrictive value.
If no scoped match, fall back to global filters.minQk.

What this Studio page does

Replays settled positions and sweeps QK floor + ceiling across scopes. The floor sweep finds where below-threshold candidates underperform; the ceiling sweep finds where above-threshold candidates flip from edge to noise. Use the per-(book, league, flow) detail tables for scope-specific inflections.

Canonical contract for QK thresholds is in docs/gates.md#qk-thresholds. Negative scoped values are documented there as intentional.

Data

Remote API Base URL

Remote User ID

Remote API Key

Load remote positions to begin.

Controls

Run Settings

Objective Metric This is the score the optimizer maximizes when ranking candidates.

Minimum Bets for Recommendation Requires at least this many bets for a recommendation to be eligible.

Aggressiveness Profile Applies a preset bundle for min bets, guardrails, and keep-rate strictness.

Guardrails

Min Distinct Days Requires recommendations to span at least this many settled days. Set 0 to disable.

Max Worst-Day Degradation ($) Rejects recommendations when worst-day profit drops by more than this amount. Set 0 to disable.

Min Green Day Rate (0-1) Requires this minimum share of profitable days for recommendations. Set 0 to disable.

Global Defaults Min Keep Rate (0-1) Stops ultra-strict global defaults: pre/live defaults must keep at least this share of baseline bets. 0 disables.

Exclude Unknown League

Skip unknown/blank league rows

Reject Still-Negative Winners

Reject rules with negative ROI/profit

When enabled, winners must be non-negative for the selected objective metric.

Guardrail Preset

Sets a practical default bundle: min bets, day stability, degradation cap, and keep-rate floor.

Guardrail order: (1) pass Minimum Bets, (2) pass enabled guardrails, (3) rank remaining candidates by Objective Metric. If all fail guardrails, the run falls back to score-only with a warning.

Sweep Setup

Min Distinct Settled Days (0 = Auto) Auto uses 25% of available distinct days in the sample window, clamped to a sane range.

Sample Window (Live days) 0 disables window filtering.

Sample Window (Pregame days) 0 disables window filtering.

Winning-Day Threshold ($) Day is "winning" when daily profit is greater than this threshold.

Data Quality & Output

Missing Data Policy Only affects missing gate fields. Other pre-filters (settled rows, known leagues) still apply.

Rows to Show Limits how many rows are rendered in the Distribution table.

Load remote positions to start.

Summary

Rows Loaded

Settled Matched

Baseline (ROI)

Best Delta

Baseline snapshot: ROI -, Profit -, Stake -, Win Rate -, Bets -

This is historical replay on bets you already took. It cannot estimate bets that were never placed.

Summary table shows optimized global defaults and selected scoped min-QK rules.

Param	Baseline	Best	Bets Kept (Best / Baseline)	Baseline Score	Best Score	Delta

Chart Scope Default sweep charts are global; selected segments appear here after a run.

Chart source: Global minQk sweep.

Shaded band = top-performing value zone (top 20% by selected objective metric).

Gold dot = selected value.

Line + dots = metric result for each tested value.

X-axis = tested minQk value.

ROI Curve

Run the optimizer to render ROI curve.

Profit Curve

Run the optimizer to render profit curve.

Win Rate Curve

Run the optimizer to render win-rate curve.

Bet Count Curve

Run the optimizer to render bet-count curve.

Distribution

Rows are tested parameter values. Each row shows aggregate results for that value.

Parameter Value	Winner Combo	Rowset	Proxy?	Tightness Avg (gap)	Tightness Fail %	Bets	Pass Rate	Distinct Days	Green Day %	Worst Day	Win Rate	ROI	Profit	Stake	Avg EV