-
-
Day Shift
-
Casino Night
Write Up: Dashboard Royale — Luck, Location, or a Leak?
Recommendation: don't replace management's bottom 10. Inspect one machine, fix the Back Forty, and run one measured promotion. Keep $75,100.
After adjusting for exposure (days on floor, available hours), luck (house win − theo, scaled by payout variance) and location (theo per available hour vs zone × title peers), none of the GM's 10 is a clear replacement case:
- Exposure (3): M-1118 and M-1120 are already retired, and M-5076 was installed 13 May.
- Leak (1): M-1158 hits on 38.6% of spins, but its game publishes 31.5% (z = 124, every week). Its hold is 3.0% against 10% configured, and none of the 80 tampering reports involve it. Inspect it before replacing anything.
- Location, luck or low-hold title (6): Nine of the 10 sit in the Back Forty. While its walkway was closed (3 Mar–6 Apr), the zone fell from 0.67× to 0.46× the rest of the floor, about $163k of theo lost. It has recovered only to 0.64×.
- Replacement was already tried: M-5076 replaced M-1120 in slot L-38-5 (13 May). Net of same-game peers it added ≈$8/day of theo against a $55/day lease, and still runs 12% below its neighbours.
Demand: Flat at $1.38M coin-in/day (hold 9.50% vs 9.46% theo). Saturday runs 127% of an average day, Tuesday 80%; events add ~10%. Downtime: ~5,000 machine-hours ≈ $99k theo (0.4%). Ridgeline: 30% of machines, 56% of fault downtime. May tech waits spiked with short-staffed shifts. Players: Tracked players lost $15.6M, ≈ theo. 16% finished ahead; the top 5% of losers carry 44% of losses. Comps: Points per comp dollar fell 15 → 12 (1 Feb) → 5.7 (1 May). Redemptions +3%, no lift in play, but the potential cost of ~156M unredeemed points doubled (≈$3.6M → $7.7M).
Model: Weekday mean × event lift (events only if announced before the forecast date), trained Jan–Apr. May–Jun MAE $46.7k vs $218.3k for a constant baseline (−79%). It makes a ~5% three-day promotion lift detectable (21% with the baseline), so the promotion can be measured.
Budget ($24,900):
| Action | Cost | 30-day effect | Measure | | Inspect M-1158 + control | $2,400 | ≈$5.8k if paytable fault | Hit rate vs 31.5% | | Back Forty wayfinding | $7,500 | +2% zone theo ≈ $5.3k, persists | Diff-in-diff vs rest of floor | | Promotion weekend | $15,000 | +8% ≈ $36k theo (past: 4–18%) | Actual − forecast; loser-share guardrail |
Sensitivity sliders: Budget tab.
Fictional casino with synthetic data. Anomalies are leads, not proof of wrongdoing.
