Site guide
A walkthrough of every section on Powerwise — what each page shows, when to use it, and how the pieces connect from pipeline to board meeting.
Powerwise is a pre-season scheduling and ranking aid for the Missouri Girls D1 lacrosse board. It anchors on last year's finish, adjusts for graduation and returning talent, surfaces flags where data and eye test may disagree, and lets the board codify votes during meetings.
A sensible order for a ranking prep session:
Every team name on the strength or board table links to /team/{slug}. The profile combines projected power, player impact, graduation loss, and schedule context in one place. Coaches can edit roster metadata at /team/{slug}/profile.
Coaches submit preseason rankings via the Coach ballot button in the header. Board members can use /ranking/{year}/vote for drag-rank ballots. Admins publish coach aggregates from /admin/coach-votes when ready.
Log in with the site password to enter edit mode. Editors can override board ranks, add program notes, manage the glossary, and switch between draft workspaces and the canonical league board.
Some runbooks live as markdown docs with their own routes:
The website reads from Neon Postgres, not from CSV files at runtime. After running the glax-pipeline, execute website/scripts/import_data.py to refresh tables, markdown docs, and the data explorer.
Navigation group
Pre-season views the board uses to compare teams, spot schedule problems, and prepare for ranking meetings. Start with strength and board, then drill into players and graduation when a flag needs context.
Main conversation table for the forecast season.
Shows every team ranked by projected power with prior finish, recommended rank, margin and RPI inputs, domain splits, turnover, and flag codes. This is the table you argue from before opening the board view.
Board votes, flags, and inline overrides.
Same underlying data as strength, oriented for the ranking meeting: board-voted rank, program notes, flag severity, and HTMX inline editing when logged in. Editors can override recommended ranks and attach notes that persist in the database.
Conference tiers and cluster views for realignment talks.
Groups teams by schedule-strength clusters and conference membership so you can see who plays whom at similar competitive levels. Useful when discussing conference moves or scheduling philosophy.
Pairwise expected margins and blowout flags.
Matrix of every team vs every opponent with projected goal differential. Highlights matchups where the model expects a ±10 goal margin (blowout risk). Filter by schedule cluster to focus on a scheduling tier.
Per-player offensive share and impact scores.
Browse returning and graduating players by team and position. Impact scores summarize how much each player contributed to team offense; useful when validating graduation flags or roster turnover stories.
Departure shares and roster turnover by team.
Shows who is leaving, what share of offense they carried, and aggregate turnover metrics per team. Pairs with the strength board turnover columns.
Navigation group
Evidence that the model behaves sensibly, plus archived seasons for context. Use these pages to sanity-check projections and cite historical precedent.
2024–25 holdout vs actual finish; MAE by tier.
Runs the pipeline forward one year and compares projected ranks to how teams actually finished. Breaks error down by tier so you can see where the model is sharp vs noisy.
Rolling strength snapshots through the season.
Season-end RPI table plus week-by-week rolling values from completed games. Shows how strength evolved during the prior season.
Week-by-week Powerwise during a live season.
Demo of in-season ranking progression using bundled CSV snapshots. Illustrates how the model would move teams as results arrive.
Archived seasons: rankings, results, teams, and model outputs.
Index of past seasons with tabs for overview, final rankings, game results, team pages, player stats, and model artifacts. Also includes career leaders, player search, and a data-gap tracker.
Navigation group
Methodology, audit trail, and raw pipeline outputs. Read these when you need to defend a number or trace it back to a CSV column.
Plain-English math with glossary tooltips.
Non-technical walkthrough of how prior finish, graduation, returning talent, and margins combine into a recommended rank. Every defined term links to the glossary.
Auditor's manual: formulas, knobs, and CSV columns.
Complete reference for reviewers: every formula, weight, flag rule, and output column. Includes the flag codes table at the bottom of the page.
Every pipeline CSV, searchable and sortable.
Imports all glax-pipeline/output/*.csv files into Postgres so you can filter and sort without opening spreadsheets. Each table name matches the file stem (e.g. ranking_recommendations_2027).
Roadmap, scope, and how the site fits the pipeline.
High-level project goals, what's in scope for MO Girls D1, and how run_glax_analysis.py feeds this site via import_data.py.