Does Leans.ai (Remi) explain why a pick was made?

No. Per the leans.ai homepage, Remi assigns each pick a unit weight from 2 to 15 representing model confidence. There is no per-pick feature attribution, no SHAP-style breakdown, and no public model card listing the input features. The unit weight is the only window into Remi's reasoning.

Does Leans.ai publish closing-line value on individual picks?

Leans.ai surfaces aggregate win-loss and units-staked tallies but does not publish a per-pick closing-line value metric. CLV is the most predictive single sharpness signal in sports betting, and its absence on Leans.ai means the user cannot verify whether Remi's picks beat the market close on any individual line.

How does NuroPicks explain each pick?

Every NuroPicks pick carries a SHAP top-features payload in the picks.shap_top JSONB column. The values render inside Discord embeds, on /record/[id] permalinks, and in the chip rows on /record as ranked signed contributions in basis points (rest advantage plus 280bp, injury minus 140bp, line movement plus 120bp). The same module renders both the bot embed and the web chip rows, so they cannot drift.

Does NuroPicks capture closing-line value automatically?

Yes. The closing-line-capture.js service runs on a 2-minute tick comparing posted odds against the book's closing number when the game starts. Every pick gets a CLV reading auto-stamped on /record, not buried in an aggregate report. Migration 0017 ships the picks_immutable_guard trigger so the captured CLV cannot be retro-edited.

Are NuroPicks and Leans.ai priced the same?

Pro tier matches the same price band at $29.99 per month. NuroPicks adds a Free tier with 1 pick per sport per night and an Elite tier at $79.99 per month with line-move trackers and capper rev-share. Leans.ai is structured around the unit-weight pick feed and excludes per-pick explainability at every tier.

Alpha

Today—Model 30d—0 LIVE

start

Learn / Leans.ai Remi vs NuroPicks SHAP CLV

Remi gives you a unit number. NuroPicks gives you the receipt.

Name: NuroPicks
Availability: InStock
Author: NuroPicks LLC

What Leans.ai actually ships, why a unit weight is not an explanation, and how NuroPicks closes the per-pick CLV and SHAP gap at $29.99 Pro.

Plain-English answer

Leans.ai is the consumer AI-pick app whose model is branded Remi. Per the leans.ai homepage, Remi runs Reinforced Recursive Machine Learning that learns from past mistakes and improves over time, then assigns each pick a unit weight from 2 to 15 representing the model's confidence. The product ships a daily pick feed, an aggregate win-loss record, and a units-staked tally. What Leans.ai does NOT ship: per-pick feature attribution, a SHAP-style breakdown of why Remi liked the pick, a model card describing the input feature set, or a public closing-line value metric on individual picks. The unit weight is the only signal you get into the model's reasoning. NuroPicks publishes the same surface differently. Every NuroPicks pick carries a SHAP top-features payload (picks.shap_top JSONB) that breaks the model's edge into ranked signed contributions: rest advantage plus 280 basis points, injury minus 140 basis points, line movement plus 120 basis points. The values render inside Discord embeds, on the per-pick permalink at /record/[id], and in the chip rows on /record. The schema is canonical (one explainer module, one TypeScript twin), and the picks row is immutable past insertion via a Postgres trigger added in migration 0017. Closing-line value is captured automatically by a 2-minute tick that compares the posted price to the book's closing number when the game starts; the CLV value renders on every pick on /record, not just in aggregate. The contrast is Remi gives you a unit number, NuroPicks gives you the unit number AND the receipt for how the model arrived at it. Pro tier matches the price band at $29.99 per month. NuroPicks adds a Free tier with 1 pick per sport per night and an Elite tier at $79.99 with line-move trackers and capper rev-share.

What a SHAP block looks like in a real NuroPicks embed

A NuroPicks pick embed includes a single SHAP line at the bottom of the why-edge block. The string is generated by one canonical module (src/bot/services/pick-shap-explainer.js) and reads like this on a real Lakers minus 3.5 example: plus 280 basis points rest advantage, minus 140 basis points injury report, plus 120 basis points line movement. The same values render in the chip rows on /record and on the per-pick permalink at /record/[id]. The TypeScript twin at web/src/lib/pick-shap-explainer.ts is parity-tested against the bot module so the web page and the Discord embed cannot drift. The data comes from the picks.shap_top JSONB column populated at insertion time. Migration 0017 ships the picks_immutable_guard trigger that rejects UPDATE and DELETE on the picks table, so the SHAP payload you read tonight is the exact one the bot wrote when the pick posted.

Remi's surface is the unit weight: 12 units on Lakers minus 4.5, backed by Reinforced Recursive Machine Learning trained on past pick outcomes. The user gets a confidence proxy, not the feature breakdown that produced it. NuroPicks ships the model output and the receipt the model produced, then auto-captures the closing-line value to verify whether the price beat the close.

Leans.ai Remi vs NuroPicks SHAP plus CLV, line by line

AreaLeans.ai (Remi)NuroPicks SHAP + CLV

Per-pick explanationUnit weight 2-15 only (no feature attribution)SHAP top features in basis points, ranked by contribution

Public CLV per pickAggregate units only, no per-pick CLVAuto CLV on every pick via 2-min closing-line-capture tick

Pick storagePick feed (no public immutability guarantee)Postgres picks table, append-only via picks_immutable_guard trigger

Public model cardBrand name (Remi) + Reinforced Recursive ML claim/learn/shap-betting documents the explainer end-to-end

Per-pick permalinkPick history within the app feed/record/[id] public, with SHAP chip rows, CLV, OG card

Stake sizing logicUnit weight 2-15 (model confidence proxy)Kelly Criterion + bankroll fraction via /bankroll kelly

PricingTiered around the unit-weight pick feedFree + Pro $29.99 + Elite $79.99

Where to read the explainer documentation

/learn/shap-betting walks through the SHAP block end-to-end: what each feature means, how basis points are signed, and how to read the chip rows on /record without a stats background.
/learn/clv covers closing-line value as a sharpness signal, why per-pick CLV beats aggregate units, and how the 2-minute capture tick stamps every NuroPicks pick.
/compare/leans covers the full feature comparison: pricing, surfaces, AI picks, unit weights, refund policy, capper marketplace.
/record is the public per-pick track record where the SHAP chip rows render alongside CLV and the result badge.
/glossary ships 461 sports-betting terms, including SHAP, edge percent, closing-line value, Kelly fraction, and rest advantage.

Not financial advice. 21+ only. Bet within your bankroll. If gambling stops being fun, call 1-800-GAMBLER.

Must be 21+. Gambling problem? Call 1-800-GAMBLER. NJ/PA: 1-800-GAMBLER · MI: 1-800-270-7117 · VA: 1-888-532-3500 · CO: 1-800-522-4700 · NY: 877-846-7369 (HOPENY).

NuroPicks LLC · 30 N Gould St Ste R, Sheridan, WY 82801