The Gilded Ace · Floor analyticsby Beta Dancers
Dashboard Royale · built on Databricks Free Edition
Luck, location, or a leak?

Don't replace the bottom 10.Move the question.

The manager's list mostly measures where machines sit, not what they are. One machine shows a real payout anomaly worth inspecting. Every number below is recomputed in your browser from the Databricks gold tables for the period you choose.

♦ The Gilded Ace is fictional and its data synthetic (TAMU Datathon Dashboard Royale). Only Las Vegas visitation (LVCVA) and Missouri slot benchmarks are real.
JACKPOT
Published hit 31.5%M-1158 38.5%
Loading data…
House win (period)
–
–
Observed hold vs theo hold
–
–
Budget spent
–
3 action types, 30 days from Jul 1
Budget kept
–
Of $100,000

The manager's bottom 10, explained

Lowest raw house win in the selected period, with what is really driving each one.

Source: g_machine_day → luck z, hit z, exposure per machine · same logic as g_machine_scorecard

What we'd do with $100k

Three actions, 30 days from Jul 1. Most of the budget stays put.

  • Inspect 5 machines $6,000

    M-1158 pays on far more spins than its published math, every month. The other four are the most negative luck results; inspection rules them in or out.

  • One promotion weekend $15,000

    The February slot tournament added +18% coin-in (≈$80k theo) with no dip afterwards.

  • Wayfinding for Back Forty $7,500

    Weak for every theme and down 30% while its walkway was closed: access is the problem.

  • Keep $71,500 71.5%

    Replacing Back Forty machines didn't help (M-1120 → M-5076: $6.00 → $6.39 theo/hr). Extra technicians would recover only $1–4k.

Priced plan

Cost = quantity × (unit cost + daily cost × max(0, 30 − lead days)) · g_budget_plan

Machine scorecard

Click a column to sort, a row for monthly detail. Zone/bank filters apply.

Same theme, different zone

Theo per available hour ($). Every theme is weakest in Back Forty.

Demand vs luck

Each dot is a machine. Low-left = weak demand; below −2 = unusually unlucky.

Daily coin-in

Shaded days had an event.

Average coin-in by weekday

No-event days in the period.

Theo per day by zone, weekly

Shaded: walkway closed. Pick a zone above to highlight it.

Event lift vs weekday baseline

Full period · g_event_lift

Casino vs Las Vegas visitors

Index, January = 1.0. LVCVA data is real but published after 2025, so it is context, not a forecast input.

Did new machines grow the floor?

Eight Aurora Link machines (bank B-51) opened Mar 24. Theo per day, excluding walkway-closure weeks.

Downtime
–
–
Lost theo (upper bound)
–
Down hours × machine's own theo per available hour
Fault tickets
–
–
Average wait for a tech
–
Ticket opened → work started

Lost theo by month and manufacturer

Reliability (exposure-adjusted)

Faults per 1,000 machine-days on the floor.

Slot technicians on shift

Daily average, actual vs planned. Day shift fell short in Apr–May.

Average wait for a technician (hours)

By month and shift the fault was opened in.

Tracked players
–
Wagered
–
Got back
–
Players' net
–
Expected loss (theo)
–
Share ahead
–

House win by player decile

Deciles by full-period theo (1 = top 10%).

Players' net and comps by tier

We do not recommend targeting heavy losers with promotions.

Points price by comp item

Comp cost and reinvestment rate

Bars: comp cost to the house. Line: comp cost ÷ tracked theo.

Actual vs forecast daily coin-in

Shaded: holdout.

Mean absolute error

Lower is better · g_forecast_scores

Claims & evidence

Every table is built by databricks/02_build_gold.sql in Databricks Free Edition and exported by 03_export_public.py.

AI use: an AI coding assistant (Cursor) helped write the SQL, the Databricks notebooks, this page and its visual styling, and draft text. All numbers come from the queries above and were re-run in Databricks. The Ask button is not a remote model: it reads the tables already loaded in this page, and what you type is not sent anywhere.