DRAFTROOMMLBB DRAFT ASSISTANT
RUNS LOCALLY

LIVE DRAFT

Your next pick,
by the numbers.

Base win rates. Every listed matchup. Every listed synergy.

Our team

0 / 5Synergy

Enemy team

0 / 5Matchups
BANNED
No bans added

Recommended picks 0

Mythic·All lanes·Base rates first

Prepare your draft room

Load a rank’s data, then add picks as the draft unfolds.

How to read the score Base WR + matchup Δ + synergy Δ. An additive score, not a predicted win probability. Unlisted pairs add 0 and remain unknown.

OUR TEAM

Choose a hero

All lanes are shown here. Already picked or banned heroes are unavailable.

SCORE BREAKDOWN

Hero details

Positive and negative entries are applied exactly as provided, from this candidate’s own rows.

DATA WORKSPACE

Data & settings

Current rank snapshot

The timestamp is when the app retrieved data, not the game’s patch date. Cache freshness is six hours; explicit Refresh fetches again. A failed refresh never invents new data.

Rate units

Both main_hero_win_rate and increase_win_rate use the same selected scale. Auto assumes fractions when every base rate is ≤ 1. Check the raw values in a score breakdown; override this setting if needed.

Import saved data

Use an exported app snapshot, or save a JSON response body from Postman. A collection file contains requests, not statistics, and cannot be imported as a response.

Raw per-hero responses add only that hero’s profile. Full recommendations need each candidate’s own matchup and synergy records. Missing profiles stay visibly provisional. A snapshot restores its own rank. Raw responses without retrieval timestamps have unknown source age; importing them does not make them fresh.

The scoring rule

Base WR + Σ enemy deltas + Σ ally deltas

Both sub_hero and sub_hero_last are included. No reversal of an enemy’s rows, no inferred symmetry, no weighting or score clipping. A listed 0 differs from an unlisted relationship. Individual lane filters only narrow our next-pick list. Full-team lineups and tournament advice separately enforce five-role coverage; picked-hero role locks can override API lane tags.

Team averages and tournament bans

A completed team’s score-based win-chance estimate is the arithmetic mean of its five adjusted hero scores, not a calibrated probability. Each hero excludes itself from synergy and is rescored when enemy picks change. Missing base rates block the estimate; out-of-range scores stay unclipped.

Starting lineups optimize all five heroes jointly. Search status distinguishes proven results from best-found results under the runtime limit. At three picks each, Net ban value = enemy best-completion loss − our best-completion loss. Future opposing picks are unknown; this is one-ban-at-a-time advice, not a prediction of the opponent’s behavior.