Aylian StudiosAylianStudios
Back to Projects
DesktopIn DevelopmentSince Jan 2025

PEARL

Performance Evaluation & Analysis for Rocket League — an evolving esports analytics platform with a production desktop app, publisher feedback pipeline, and tri-core roadmap across in-game telemetry, mobile companion, and ML coaching.

C++RustTypeScriptReactTauriSQLiteCloudflare WorkersDiscord APIPythonKotlinBakkesMod SDKSupabasePyTorchJetpack Compose

Overview

PEARL (Performance Evaluation & Analysis for Rocket League) is a local-first analytics platform built for practical competitive improvement.

The desktop app is the active core today, but the product is evolving into a broader platform:

  • optional authentication as an account layer (never required for local use)
  • premium cloud features tied to auth
  • cross-device sync for notes, tags, and grouping configs
  • stable release operations with feature flags and release channels (stable vs beta)

The long-term direction is one consistent PEARL experience across desktop, mobile, and web while preserving a strong offline-first local workflow.

Product + Platform Direction

Local-First Core (Always Available)

  • Replay parsing, analysis, and storage remain fully usable without an account
  • Player workflows continue to prioritize local ownership and low-friction usage
  • Cloud capabilities are additive, not required

Account Layer + Premium Cloud

  • Optional sign-in for users who want synced and cross-device workflows
  • Premium candidate features include:
    • cross-device sync (notes/tags/group configs)
    • mobile companion experience
    • web app access to cloud-backed views
  • Feature flags and channel controls enable safe rollout and faster iteration

Analytics Roadmap

Replay Grouping and Aggregation

  • Collections/series for grouped replay analysis
  • Aggregate averages and totals at group level
  • Delta views between selected groups

Scrim Auto-Grouping

  • Infer Team A vs Team B from repeated roster patterns
  • Show confidence-based suggestions in UI
  • Require user confirmation when confidence is medium

Session Grouping

  • Time-window heuristic grouping (for example 10-20 minute gap thresholds)
  • Optional same-day grouping mode
  • Manual override controls with merge/split tools

Coaching and Insight Layer

Rule-Based Coaching (First)

  • Deterministic coaching rules for immediate value
  • Low operating cost and fast iteration loop
  • Better baseline guidance before heavier AI infrastructure

AI Coach (Later)

  • Deeper coaching generation once data quality, funding, and hosting are ready
  • Designed to build on top of the rule-based foundation

Insight Features

  • Key-moments auto-highlights (kickoff mistakes, boost starvation windows, defensive over-commit chains)
  • Trend diagnostics over 7/30/90 day windows
  • Replay quality scoring by match/session (good, neutral, bad habits index)
  • Player archetype profiling (aggressive challenge, support, boost-control styles)
  • Objective pack builder ("next 3 focus points" from recurring weak patterns)

Competitive and Team Workflow

  • Opponent/team profile pages with winrate and pattern summaries
  • Scrim report export (PDF, markdown, share link targets)
  • Roster-aware comparisons (same lineup vs mixed lineup)
  • Team comms notes linked to grouped scrims
  • Coach mode dashboard for multi-player side-by-side review

Reliability and Operations

  • Background crash reporting and diagnostics bundle support
  • Anonymous telemetry for update success/failure operations
  • Data migration/versioning hardening for long-lived local databases
  • Stable updater endpoint strategy under PEARL-owned domain
  • Release governance via stable/beta channels and feature flags

Roadmap

Phase 1: Grouping Core and Workflow Controls

  • Deliver manual grouping plus session and scrim suggestions
  • Ship merge/split overrides and confidence-driven review UX

Phase 2: Aggregation Views and Exports

  • Add group-level averages, totals, and deltas
  • Expand report/export pipeline for team and coaching workflows

Phase 3: Optional Auth and Premium Cloud Basics

  • Introduce account layer without blocking local usage
  • Launch first sync capabilities across desktop/mobile/web surfaces

Phase 4: Rule-Based Coach

  • Release deterministic coaching guidance and key-moment highlights
  • Build trend diagnostics and objective sequencing loops

Phase 5: AI Coach

  • Add ML-assisted coaching once infrastructure budget and hosting are ready
  • Scale model-backed insights on top of validated product signals

Related Projects