Mithrandir Metrics
34 aligned 18 lane-health rows Refreshed Sep 10 06:00 AM ET
Betting desk · Latest available models

Game reads and model-market comparisons.

Start with a game. Historical model boards are labeled clearly when today’s slate is not ready.

Slate
15
Aligned
34
Split watch
1
Historical model boards The latest available betting-model slate is 2026-09-10. Current-game reads may still be available below, but incomplete player comparisons are withheld.

Start with the game

Choose a matchup to answer the side question first, then see totals, first inning, player markets, model availability, and Kalshi in one place.

Open full slate
2:20 PM ET · Kalshi market context PIT at CHC No Mithrandir side probability is published yet. Read Model pending Total -- NRFI -- 6:40 PM ET · Kalshi market context COL at DET No Mithrandir side probability is published yet. Read Model pending Total -- NRFI -- 6:45 PM ET · Kalshi market context LAA at WSH No Mithrandir side probability is published yet. Read Model pending Total -- NRFI -- 7:05 PM ET · Kalshi market context NYM at NYY No Mithrandir side probability is published yet. Read Model pending Total -- NRFI -- 7:07 PM ET · Kalshi market context BAL at TOR No Mithrandir side probability is published yet. Read Model pending Total -- NRFI -- 7:10 PM ET · Kalshi market context KC at BOS No Mithrandir side probability is published yet. Read Model pending Total -- NRFI --
MODELS 34 aligned · 1 split LINEUPS 90 confirmed PLAYER STATUS 2396 tracked

Which models are live?

Availability comes from today’s saved outputs. “Not live” means Mithrandir is intentionally withholding a number.

Full coverage details
Home Runs
Palantir · live Fangorn · live Valinor · live
Strikeouts
Palantir · live Fangorn · not live Valinor · not live
Total Bases
Palantir · live Fangorn · not live Valinor · not live
Hits
Palantir · live Fangorn · not live Valinor · not live
Pitcher Outs
Palantir · live Fangorn · not live Valinor · not live
Strikeouts · Palantir calibration review Last 14d ECE 0.087, 14d error rising. This flags probability calibration, not a blanket model failure.
Strikeouts · Valinor calibration review Last 14d ECE 0.112, 14d error rising. This flags probability calibration, not a blanket model failure.
Hits · Palantir calibration review Last 14d ECE 0.073, 14d error rising. This flags probability calibration, not a blanket model failure.
Pitcher Outs · Palantir calibration review Last 14d ECE 0.048, 14d error rising. This flags probability calibration, not a blanket model failure.
Lane health Drift flags mean a market/lane needs calibration review because recent probability buckets moved, not that every model is getting worse. Rows with insufficient sample stay labeled until enough outcomes accumulate.
Filter rules Lineup filter defaults on. Rows hide players not confirmed in the lineup when lineup data is available. Positive gaps require context. Gaps below 2pp are treated as noise; 5pp is consensus, 6-10pp is soft consensus, and above 10pp is disagreement. Young players are labeled. Hitters under 50 MLB AB or pitchers under 20 MLB IP carry limited-sample context.
How the three-lane strip reads

Aligned

≤ 5pp
P 38.0
F 40.0
V 39.0

Three methods converging tightly. Useful, but still filtered through line quality and sample context.

Soft consensus

6-10pp
P 31.0
F 37.0
V 33.0

Same direction, wider uncertainty. Median probability matters; the spread tells you how much humility to carry.

Disagreement

Watch
P 22.0
F 39.0
V 26.0

One lane sees something the others smooth through. That is a research signal, not noise to hide.

Today’s slate
Market navigation
Custom Combo Combo Analyzer Build your own 2-to-6 leg combo from the current prop universe and get joint probability, model-market gap, correlation notes, and an analyst read. 2026-09-10 slate · 2 to 6 legstracking Season Futures Futures Gap Board A season-long fair-price comparison surface for division winners, pennants, World Series, win totals, and awards, framed around uncertainty rather than picks. 20260910 snapshot · 51 rowstracking Market Structure Market Watch A source-aware line shopping and CLV-readiness surface that separates consensus-backed markets from sparse or single-book rows. 20260910 snapshot · 9 anchor-ready rowstracking Home Run Props Home Run Hub The central home run overview page, built to summarize Mithrandir's model lanes, market context, and supporting HR boards without replacing the dedicated model pages. 2026-09-10 slate · 5 signalstracking Palantir Model Palantír Stable daily board 2026-09-10 slate · Stable daily boardtracking Fangorn Model Fangorn Random Forest engine built to identify nonlinear home run market gaps, surfaced through the saved core, select, conviction, and parlay layers. 2026-09-10 slate · 7 singlestracking Total Bases TB Value Board A model-vs-market pricing board for total bases that surfaces under values first, over leans as context, and the biggest line gaps without forcing official picks. 2026-09-10 slate · 62 board rowstracking Strikeout Props Strikeout Value Board An internal strikeout projection board built from Mithrandir starter and matchup artifacts, with no live sportsbook dependency in the website path. 2026-09-10 slate · 8 board rowstracking Hits Hits Board A cleaner contact-volume prop lane focused on expected hits, 0.5 and 1.5 hit probabilities, and market-aware decision layers. 2026-09-10 slate · 40 board rowstracking Pitcher Outs Pitcher Outs Board Starter workload, leash, and traffic-risk informed pitcher outs projections with a clean fair-odds fallback. 2026-09-10 slate · 7 board rowstracking Game markets · unified view Game center Answer the matchup question first, then open the supporting model and market layers. 15 gamesstart here Game markets · pilot NRFI / YRFI Daily artifact is wired; production inference is still a carried debt. 15 rowspilot Event markets · live exchange context Kalshi vs Mithrandir Game-winner and first-inning exchange prices beside the model's fair probability. 45 marketsstale

Operational context

840 sample flags
Lineups90confirmed rows
Player status2396transactions / roster
Triple-confirmed0cross-domain signals

These artifacts feed row chips on the boards. Missing lineup rows mean pre-confirmation pool, not fabricated “IN” labels.

Lane Health 2026-09-10
Palantir 19542 props Brier 0.118 ECE 0.006 14d stable
Fangorn 19542 props Brier 0.118 ECE 0.006 14d stable
Valinor 13866 props Brier 0.114 ECE 0.003 14d stable
Palantir 1715 props Brier 0.248 ECE 0.011 14d error rising
Fangorn 53 props Brier 0.257 ECE 0.119 sample building
Valinor 1211 props Brier 0.249 ECE 0.024 14d error rising
Palantir 9239 props Brier 0.238 ECE 0.048 14d stable
Fangorn tracking in progress (0 logged)
Valinor tracking in progress (0 logged)
Palantir 747 props Brier 0.166 ECE 0.036 14d error rising
Fangorn tracking in progress (0 logged)
Valinor tracking in progress (0 logged)
Palantir 973 props Brier 0.256 ECE 0.056 14d error rising
Fangorn tracking in progress (0 logged)
Valinor tracking in progress (0 logged)
Palantir tracking in progress (0 logged)
Fangorn tracking in progress (0 logged)
Valinor tracking in progress (0 logged)