Review: Analytics dashboard: find winning conditions

review-1841-2026-07-09 Review

review needs-refinement

Project

[x] story:bot-strategy label -- "Bot operator identifies winning conditions"
[ ] story note MISSING -- story-bots-bot-strategy listed as "(pending)" on project-botfire user-stories section. [SCOPE] Create story note story-bots-bot-strategy.
[x] arch:bots label -- bots architecture component
[ ] arch note MISSING -- search for arch-bots returned no results in pal-e-docs. [SCOPE] Create architecture note arch-bots for the bots component.
[x] Forgejo issue -- https://forgejo.tail5b443a.ts.net/ldraney/botfire/issues/5, open

[x] app/controllers/dashboard_controller.rb -- verified: exists (2.5k), has data endpoint querying BotFire with joins to bot, trigger, market, game. Already accesses g.sport_type via join.
[x] app/services/scoreboard.rb -- verified: exists (3.0k), has bot_summary method with P&L, win/loss, stop-loss/take-profit counts. Good foundation to extend.
[x] db/migrate/ -- verified: directory exists with 13 migrations.
[ ] db/migrate/ -- ISSUE: Issue mentions adding sport_type to bot_fires, but sport_type already exists on games table and the controller already accesses it via join (g.sport_type). Clarify whether denormalization is intended or join-based computation suffices. [BODY]
[ ] db/migrate/ -- ISSUE: game_phase_at_entry is proposed but "game phase" is not defined in terms of existing data. The games table has period (string) and clock (string) fields. How to map these to early/mid/late across 5 different sports (each with different period structures -- innings, quarters, halves, periods) needs explicit definition. [BODY]
[x] app/views/dashboard/index.html.erb -- verified: exists (261 lines), live position monitoring dashboard with dark theme, summary cards, bot cards, open-positions table.

OK -- issue filed on ldraney/botfire, all file targets in ldraney/botfire. Single-repo scope.

Board item #1839 (Rework Volatility Bot) -- in QA column, deployed. Generating the trade data this ticket will analyze. Not a blocker.
Board item #1840 (Rework Consensus Bot) -- in backlog, ON HOLD pending analytics. This ticket (#1841) is a prerequisite for #1840. Dependency direction is correct and documented in Related section ("Issue #2 -- Consensus Bot rework (ON HOLD until analytics prove what to amplify)").
No blockers for this ticket. Dependencies are documented.

8 acceptance criteria. Most are testable, with these concerns:

AC 3 (P&L by Game Phase) -- "early/mid/late" mapping from existing period/clock fields is undefined. Different sports have different period structures (innings, quarters, halves, periods). An agent cannot implement this without a definition.
AC 7 (Cross-filter) -- complex interactive behavior: "click a sport, price band, or game phase to filter all other views." Significant frontend engineering effort, especially with no chart libraries allowed.
AC 8 (Hypothesis Log) -- persistence mechanism undefined. Is this localStorage, a new DB table, or ephemeral? "Simple text area or section" is too vague for agent implementation.

Dashboard changes are visually isolated (single page, no new routes).
Scoreboard service is used by bot services for console logging. Extending with analytics methods is additive and non-breaking.
BotFire model is used by 6 services (volatility_bot, consensus_bot, bot_engine, position_monitor, portfolio_grouper, scoreboard). Adding columns via migration is non-breaking.
No chart libraries constraint means all visualization (6+ chart types including line charts, bar charts, tables with breakeven lines) must be custom HTML/canvas. This is the highest-effort constraint.

NEEDS DECOMPOSITION -- fails the 5-minute rule on multiple criteria:

File targets: 3-4 files in 1 repo -- borderline acceptable.
Acceptance criteria: 8 criteria -- exceeds the >5 threshold.
Estimated agent work: Well over 15 minutes. Backend analytics endpoint + 6 chart views (all custom HTML/canvas, no libraries) + cross-filter interactivity + responsive design + hypothesis log + possible migration + tests. At minimum 3 separate agent passes.

Suggested decomposition into 3 sub-tickets:

Analytics backend + static tables (3pt) -- analytics endpoint with P&L breakdowns by sport, price band, game phase, exit reason. Static HTML tables with breakeven threshold. AC 1-5 data + display.
Cumulative P&L trend chart (2pt) -- custom canvas line chart per bot showing P&L over time. AC 6. Responsive.
Cross-filter + hypothesis log (3pt) -- AC 7-8: interactive filtering across all views and hypothesis persistence. Depends on tickets 1 and 2.

[DECOMPOSE] 8 AC across backend analytics + 6 custom chart types + cross-filter interactivity + hypothesis log. Route to skill-decompose-ticket.
[SCOPE] Create architecture note arch-bots for the bots component.
[SCOPE] Create story note story-bots-bot-strategy (currently listed as pending on project-botfire).
[BODY] Define "game phase" mapping: how do period and clock fields on the games table map to early/mid/late for each sport type (baseball innings, basketball quarters, football quarters, hockey periods, soccer halves)?
[BODY] Clarify sport_type migration: the field already exists on games table and is accessible via join. State explicitly whether denormalization to bot_fires is intended or join-based access suffices.
[BODY] Define hypothesis log persistence: localStorage, new DB table, or ephemeral?

Verdict: NEEDS_REFINEMENT