Win per unit is casino win divided by a defined number of operating units for a stated period. A unit may be a slot machine, table, electronic terminal, seat, pit, kiosk, or another item chosen for the report.
The phrase is incomplete unless the report identifies four things: the type of win, the unit definition, the time period, and whether unavailable or closed units are included.
Basic formula
Win per unit = casino win ÷ number of units
If 20 slot machines produce $12,000 of actual win in one day:
Win per machine = $12,000 ÷ 20 = $600
That is an average across the bank. It does not mean every cabinet won $600. One machine may have won $1,500, another may have lost money after a jackpot, and the group average may still be $600.
The metric can be calculated from actual or theoretical win:
- Actual win per unit: observed casino win divided by units.
- Theoretical win per unit: expected win from approved math and wagering volume divided by units.
Actual win is used for accounting and realized performance. Theoretical win helps judge whether a short-term result is above or below the expected level.
The denominator is part of the metric
Two reports labelled “win per unit” can produce different answers because they count units differently.
A slot report might divide by:
- all installed machines;
- active machines;
- available machines excluding out-of-service units;
- occupied machines during a time interval;
- machine-days over a month.
A table-games report might divide by:
- installed tables;
- open tables;
- open table hours;
- occupied seats;
- pits or game types.
None of these denominators is automatically wrong. The problem arises when analysts compare results built from different definitions.
Suppose a floor wins $90,000 from 100 installed machines, but 10 were unavailable for the full period. The report shows:
- $900 per installed machine using 100 units;
- $1,000 per available machine using 90 units.
Both calculations are arithmetically correct. Only the label tells the reader what was measured.
Unit-days handle changing inventory
A fixed month-end count can distort performance when machines are added, removed, or unavailable during the period. Unit-days give each unit one count for every day it was available.
Win per unit-day = total win ÷ total available unit-days
Consider 18 machines available for all 30 days and 2 new machines available for the final 15 days:
Unit-days = (18 × 30) + (2 × 15) = 570
If total win is $171,000:
Win per unit-day = $171,000 ÷ 570 = $300
Dividing by the 20 machines visible at month-end would give $8,550 per machine for the month, but that figure ignores the shorter exposure of the new units. Unit-days make the comparison more honest.
Win per day measures the time dimension directly. Combining unit and time creates the common slot metric win per machine per day.
Slots and tables need different context
For slots, managers often pair win per machine with:
- coin-in;
- actual and theoretical hold;
- denomination;
- cabinet and game theme;
- occupancy or utilization;
- progressive liability and jackpot volatility;
- lease or participation cost;
- placement and traffic zone.
A machine can have high coin-in but modest win because its hold percentage is low. Another can show strong short-term win after an unusually favorable period. A third can look weak in win per unit but still attract valuable traffic to the surrounding bank.
For tables, win per table is strongly affected by open hours, average bet, occupied positions, game speed, player mix, and variance. Comparing a roulette table open 18 hours with a high-limit baccarat table open 5 hours is not meaningful without time and demand adjustments.
A useful table metric is:
Win per open table hour = table win ÷ open table hours
If a pit wins $24,000 across 160 open table hours:
Win per open table hour = $24,000 ÷ 160 = $150
That still does not equal profit. Dealer wages, supervision, benefits, equipment, taxes, comps, utilities, and allocated overhead remain outside the numerator unless the report explicitly subtracts them.
Actual win can mislead over short samples
Casino results are volatile. A single jackpot can make a slot bank’s actual win per unit look poor. One large losing player can make a table look exceptionally productive. Neither event proves the underlying game, placement, or staffing decision is good or bad.
Managers should compare:
- longer rolling periods;
- actual versus theoretical win;
- variance and jackpot events;
- weekdays versus weekends;
- comparable denominations and customer segments;
- pre-move and post-move performance;
- unit availability and open hours.
The Nevada Gaming Control Board’s gaming revenue information reports illustrate why gaming results are commonly viewed across one-month, three-month, and twelve-month periods. Longer periods do not remove every difference, but they reduce the temptation to treat one noisy day as a stable productivity rate.
Occupancy and utilization explain why equal units differ
Two identical machines can have different win per unit because one sits on a main traffic path and the other is rarely occupied. Two identical blackjack tables can differ because one opens during peak demand while the other operates through quiet hours.
That is why unit productivity is often paired with machine utilization or occupied-seat measures. A low-utilization unit may need better placement, different opening hours, a denomination change, or removal. A highly utilized unit with modest win may indicate strong demand but low stake, low hold, or a product that supports broader customer value.
Utilization should not be used to excuse every weak result. It identifies the mechanism behind the number. The operating decision still depends on whether management can improve demand, pricing, product mix, or cost.
Theoretical win per unit
Theoretical unit performance can be estimated from the activity appropriate to the product.
For a slot bank:
Theoretical win per machine = coin-in × theoretical hold ÷ machines
If 25 machines receive $500,000 of coin-in at a weighted theoretical hold of 8%:
Theoretical win = $500,000 × 0.08 = $40,000
Theoretical win per machine = $40,000 ÷ 25 = $1,600 for the period
For a table area, a simplified model is:
Theoretical win = average bet × decisions × house edge
The result can then be divided by open tables or open table hours. The calculation is only as reliable as the ratings, decision counts, rule assumptions, and denominator.
See theoretical win, average bet, and game speed for the inputs.
Win per unit is not profit per unit
A high win-per-unit number can hide high cost. A proprietary table game may produce strong win but carry licensing expense. A premium slot cabinet may require a participation payment. A live table needs labor and supervision that a machine does not.
For capital and floor decisions, managers may extend the analysis to contribution:
Contribution per unit = win per unit − directly attributable unit costs
The definition of attributable cost must be consistent. Adding only a lease fee to one product while ignoring labor on another creates a biased comparison.
Floor optimization and game mix use unit metrics as inputs, not final answers. Customer demand, strategic positioning, service standards, and the value of a complete product mix can justify a unit whose direct ranking is not the highest.
A reliable report states the full metric
Instead of writing “WPU = $500,” a useful report says:
Actual slot win per available machine per day, trailing 90 days: $500.
That label specifies the numerator, denominator, time basis, and sample window. It allows another analyst to reproduce the number and compare it with a compatible group.
The essential questions are:
- Is the numerator actual win, theoretical win, or contribution?
- What exactly counts as one unit?
- Were unavailable or closed units included?
- Is the period expressed as a day, month, unit-day, or open hour?
- Are the products comparable in denomination, demand, cost, and volatility?
Win per unit becomes useful when those questions are answered. Without them, it is a neat ratio that may rank the wrong things for the wrong reason.