statline 3.0.0rc2

StatLine — weighted player scoring, efficiency modeling, CLI + adapter tooling

StatLine

StatLine — Advanced weighted player scoring and analytics, with modular tools for awards, performance evaluation, and real-time integrations.

“StatLine” is a trademark of StatLine LLC (in formation), registration pending. Source code is licensed under the GNU Affero General Public License v3 (see LICENSE). Brand, name, and logo are not covered by the AGPL (see TRADEMARK_POLICY.md).

---

What is StatLine?

StatLine is an adapter‑driven analytics framework with an optional remote API (SLAPI). StatLine..

• normalizes raw game stats,
• computes per‑metric scores,
• aggregates into buckets and applies weight presets (e.g., pri, mvp, role weights),
• exposes a clean CLI and Python library API,
• optionally ingests Google Sheets and caches mapped rows,
• integrates with SLAPI for secure, multi‑client deployments,
• and (new) stabilizes PRI to an adapter defined normalized set of ranges and percentiles (see below).

Supported Python: 3.10 – 3.14 (CI: Linux, macOS, Windows)

---

What’s new (v3.0.0)

• Keys & Auth: Access is gated between local and SLAPI. CLI explains further on authorization procedure.

• PRI normalization: output scales are adapter defined for clearer tiers and saner UX.

• Adapters: tightened demo adapter + docs; versioned schemas.

• Percentiles: now available as defined in adapter schematics

• Roadmap (ships in v3.1.0):

  • Batch processing filters (by position, games played, and adapter‑defined stat predicates [BACKEND COMPLETE])
  • Output toggles (e.g., show weights, hide pri_raw, include per‑metric deltas [BACKEND COMPLETE])

---

Install

Base install:

pip install statline

With extras:

# Google Sheets ingestion
pip install "statline[sheets]"

# Developer tools (linters, types, tests)
pip install -e ".[dev]"

---

CLI Basics

The console script is statline (also callable via python -m statline.cli).

statline --help
python -m statline.cli --help

The CLI --help command will guide you through authorization (contact the repository maintainer for a key).

---

Remote API (SLAPI)

Use SLAPI for remote/scaled workflows. SLAPI validates REGKEY or API Access Token on every request. Revocation invalidates both. Requires validation from a project maintainer. (access not public yet)

---

Input Formats

StatLine reads CSV or YAML. Columns/keys must match adapter expectations.

CSV

• First row is the header.
• Each subsequent row is an entity (player).
• Provide raw fields your adapter maps (e.g., pts, ast, fgm, fga, tov).

display_name,team,pts,ast,orpg,drpg,stl,blk,fgm,fga,tov
JordanRed,RED,27.3,4.8,1.2,3.6,1.9,0.7,10.2,22.1,2.1

Example adapter (YAML)

Adapters define schema for raw inputs, buckets, weights, and derived metrics. Place in statline/core/adapters/defs/.

# Key - name your adaper
# Version - SemVer; useful roadmapping
# Aliases - Alternative names/lookup
# Title - Non case sensitve naming option
key: example_game
version: 3.1.0
aliases: [eg, sample]
title: Example Game

dimensions:
  # Indirect levers for future implementation and/or filter I/O
  # eg: offense, defense, as values, side, map as unweighted definitions/buckets
  map:   { values: [MapA, MapB, MapC] }
  side:  { values: [Attack, Defense] }
  role:  { values: [Carry, Support, Flex] }
  mode:  { values: [Pro, Ranked, Scrim] }

buckets:
  # Weighting organization (metrics are unweighted, it's why you use buckets);
  # Great for grouping stats or interpolations together
  scoring: {}
  impact: {}
  utility: {}
  survival: {}
  discipline: {} # Weighted separate for punishable stats, like mistakes; see below
  telem: {} # Useful for metrics used in efficiency DSL but not weighted directly; see below

metrics:
  # Direct fields (ppg, apg, ast, kill, etc..) (clamped; discipline is inverted so fewer mistakes score higher) 
  # Clamps usually auto derived from leader and Foor in Batch mode
  - { key: stat3_count, bucket: utility,    clamp: [0, 50],               source: { field: stat3_count } }
  - { key: mistakes,    bucket: discipline, clamp: [0, 25], invert: true, source: { field: mistakes }    }
  # Still invert the bad metric; discipline is just a separate weight lever; the bucket isn't inverted
  - { key: gp,          bucket: telem,      clamp: [0, 12],               source: { field: gp }          }

# New ratio/derived DSL with field names, clamps, and minimum denominators
efficiency:
  # Per-round scoring output
  - key: stat1_per_round
    bucket: scoring
    clamp: [0.00, 2.00]
    min_den: 5
    make: "stat1_total"
    attempt: "rounds_played"

  # Impact success rate
  - key: stat2_rate
    bucket: impact
    clamp: [0.00, 1.00]
    min_den: 10
    make: "stat2_numer"
    attempt: "stat2_denom"

  # Survival quality (good vs total)
  - key: stat4_quality
    bucket: survival
    clamp: [0.00, 1.00]
    min_den: 5
    make: "stat4_good"
    attempt: "stat4_total"

weights:
  # Separate profiles of weight distribution
  # Doesn't need to sum to one; scorer evenly redistributes as necessary
  pri:
    scoring:    0.30
    impact:     0.28
    utility:    0.16
    survival:   0.16
    discipline: 0.10
  mvp:
    scoring:    0.34
    impact:     0.30
    utility:    0.12
    survival:   0.14
    discipline: 0.10
  support:
    scoring:    0.16
    impact:     0.18
    utility:    0.40
    survival:   0.16
    discipline: 0.10

penalties:
  # our discipline bucket(s) is/are self weighted here; 
  # acts as a downweight after normalization (eg: pri says discipline matters 8% less under support)
  pri:     { discipline: 0.10 }
  mvp:     { discipline: 0.12 }
  support: { discipline: 0.08 }

filters:
# include-only = output whatever passes the filter (eg: >= outputs only >= because else failed)
# exclude-only = output whatever fails the filter (eg: <= outputs > because it passed <=)
  gp: { metric: gp, accepts: ["<", ">", "<=", ">=", "=", "!="], values: ["include-only", "exclude-only"] }

sniff:
# helps auto-select
  require_any_headers: [stat1_total, rounds_played, stat2_numer, stat2_denom, stat4_good, stat4_total, stat3_count, mistakes]

See HOWTO.md for adapter authoring.

---

Versioning & Stability

• Adapters: semantic (MAJOR.MINOR.PATCH). Breaking schema changes → new MAJOR (suggested to match with repository release version).
• PRI scale: now adapter defined.
• SLAPI: deprecations target ≥60 days notice unless security requires faster action. (Whoops! V2 is now deprecated, so much for 60 days!)

---

Security & Legal

• Revocation: We may revoke keys for AUP violations; revocation invalidates both REGKEY and any derived API tokens immediately.
• ToS / Privacy / AUP: see /legal links below.
• Trademark: "StatLine" and associated logos are trademarks; see TRADEMARKS.md.

Links:

• Terms of Service — /legal/tos
• Privacy Policy — /legal/privacy
• Acceptable Use Policy — /legal/aup
• Contributor License Agreement — CLA.md
• License — LICENSE
• Trademark Policy — TRADEMARKS.md

---

Quick FAQ

Is the CLI required? Currently, yes. You can score locally without API access.

Can I use my own game? Yes—write an adapter; any game with a box score + valid datasheet can work.

Do I need env vars? No. A maintainer should assist you should you require gated access.

How are percentiles computed? Per dataset/sample window defined in the request or adapter context (v3.0.0 feature).

Discord support? We aim to make a central Discord server soon for ongoing development of our Discord app. As of current, no.