Angles · look-through
Model Conviction Pickem
as of Oct 18, 5:25 PM ET
hit rate (all)
Ready0.0%
n=1
as of Oct 18
trailing 300
Ready0.0%
n=1
as of Oct 18
best market
Readyspread
0.0% · n=1
as of Oct 18
weeks 1-9
Ready0.0%
n=1
as of Oct 18
weeks 10-18
—
n=0
as of Oct 18
playoffs
—
n=0
as of Oct 18
Angle card · what every angle carries
“Model Conviction Pickem”
The tell
spread
Fires tonight on 0 games.
Blind spot
spread (0.0% · n=1).
Rolling hit rate needs 100 graded claims — not there yet.
- Graded claims
- 1
- Hit · Brier
- 0.0% · 0.250
- Lift vs break-even
- -52.4 pts
- CLV when followed
- +0.0
- By regime
- early 0.0% · late — · playoffs —
- Drift (trailing 300)
- — (needs 300 graded)
- Rating
- Not rated yet
- Used in other angles
- None yet
Counts differ by cohort: the tiles above use decided claims (wins and losses) in each window, while Graded claims counts every graded ledger row, pushes included. Lift here is measured against the −110 break-even; the card on /lenses measures it against the other lenses.
Record by part of the season
| Weeks | Hit rate | Games |
|---|---|---|
| 1-9 | 0.0% | 1 |
| 10-18 | — | 0 |
| playoffs | — | 0 |
Calibration
dot size = n; dashed = perfect calibration · n=1 · as of Oct 18, 5:25 PM ET
Tells it reads
- spread
Last graded calls
| Game | Call | Confidence | Result |
|---|---|---|---|
| NE @ BUF | spread | 50.0% | miss |
Scoreboard
Not enough graded games yet.
Last 300 games
Not enough graded games yet.
This season
Not enough graded games yet.
Reliability
Brier / ECE per window, read off the scoreboard's own rows — the confidence-vs-observed curve above (from the hourly update) is a separate reading.
Record by conditions
Not enough graded games to compare conditions yet (Last 300 games).
Recent calls
| Game | Market / side | Percentile | Details | Result |
|---|---|---|---|---|
| NE @ BUF | spread · home | 100% | modelPHome, ensembleDisagreement, modelCount, spreadClose | loss |