Casinos usually calculate comps in two steps: estimate the theoretical value of the player’s rated gambling, then decide how much of that value the property is willing to reinvest. The result may fund food, rooms, free play, event access, transportation, or discretionary host benefits.
Actual loss can influence a decision, especially during an unusual trip or host review, but it is not a reliable substitute for theoretical value. A player can lose heavily in twenty minutes or win during four hours of valuable rated action. The casino needs a measure that is less dependent on one session’s luck.
The core calculation
For a table game, a common model is:
Theoretical loss = average bet × decisions per hour × hours played × house edge
The variables mean:
- average bet: the rated average amount risked per decision;
- decisions per hour: the estimated pace for that game and table condition;
- hours played: rated time in active play;
- house edge: the expected casino advantage for the game, rules, and assumed player behavior.
For a machine game, the calculation can be more direct:
Theoretical loss = tracked coin-in × configured house edge
Coin-in means the cumulative amount wagered, including recycled credits. It is not the cash inserted and not the final loss.
The preliminary comp budget is then often modeled as:
Comp budget = theoretical loss × reinvestment rate
There is no universal reinvestment rate. It can differ by property, player segment, benefit type, campaign, game, host authority, and whether the casino uses retail value or internal cost.
A complete table-game example
Suppose a rated blackjack player has:
- $50 average bet;
- 55 decisions per hour;
- 3 hours of rated play;
- a 1.2% house-edge assumption for the rating model.
First calculate total action:
$50 × 55 × 3 = $8,250
Then calculate theoretical loss:
$8,250 × 0.012 = $99
If the property’s illustrative reinvestment policy for that situation were 25%:
$99 × 0.25 = $24.75 preliminary comp budget
The 25% is an example, not an industry promise. Another casino can use a different percentage, a different game-speed assumption, or a different cost rule.
The player’s actual result does not enter this basic calculation. The player might finish up $600 or down $700 while the rated theoretical remains about $99.
Why two equal losses can produce unequal offers
Consider two players who each finish down $500.
Player A plays $25 blackjack for 25 minutes, loses quickly, and leaves.
Player B plays a machine for three hours, records $6,000 coin-in with a player card, and finishes down the same $500.
The final loss is identical. The underlying action is not. Player B may have generated more measured theoretical value; Player A’s result may be mostly short-session variance.
This is why “I lost more” is not the same claim as “my play generated more comp budget.” Comps are not proportional refunds.
What happens after the formula
Theoretical value gives the casino a starting point, not an automatic entitlement. The final decision can be changed by several layers.
Recorded-play quality
A table rating may be wrong because the card was not opened promptly, time was left running during a long break, the average bet missed a major increase, or chips were spread across multiple hands. Machine play is more automated, but card removal, shared cards, excluded games, and system outages can still affect the record.
Players who care about accurate ratings should present their card at the start, avoid lending it, and ask politely whether the session was captured before leaving.
Retail value versus casino cost
A $200 room rate is not necessarily a $200 comp cost to the casino. An unsold room, buffet seat, show ticket, or resort credit can have a different internal cost from its advertised retail price. Free play has another cost model because some of it returns as winnings and some creates additional wagering.
This allows two benefits with the same public value to consume different amounts of the internal budget.
Trip, day, or long-term view
Some systems calculate average daily theoretical, trip theoretical, recent rolling averages, predicted future value, or a combination. A short low-play visit can therefore dilute a daily metric at one property while being ignored or treated differently at another.
The property may also distinguish earned same-day comps from future marketing offers. A host’s available meal authority is not necessarily calculated the same way as next month’s room mailer.
Discretion and exceptions
Hosts and supervisors usually operate within limits. They may consider actual loss, service recovery, competitive matching, a special event, unused budget, player history, or management approval. Discretion can adjust the outcome, but it does not mean the underlying rating has disappeared.
The operational risk is overcomping a player whose expected value cannot support the benefit, or undercomping a valuable player because the rating is incomplete.
How the rating enters the system
At a table game, the rating usually begins when the player presents a card or account identifier. A supervisor or dealer opens the session, records the game and table, estimates the average wager, and closes the session when play ends. The system may supply a standard decisions-per-hour value rather than counting every hand.
Average bet should represent the session, not the largest wager seen for a moment. If a player bets $25 for 90 minutes and $100 for the last ten minutes, rating the entire session at $100 greatly overstates action. A weighted estimate is cleaner:
Weighted average bet = total estimated action ÷ total decisions
Suppose the player makes 75 decisions at $25 and 10 decisions at $100:
Total estimated action = (75 × $25) + (10 × $100) = $2,875
Weighted average bet = $2,875 ÷ 85 = $33.82
That is very different from using either $25 or $100 for the whole session. Why average bet matters examines this rating problem in detail.
Machine ratings usually record each qualifying wager electronically, so the theoretical calculation can use actual coin-in and the configured game model. That does not make every offer identical: excluded products, multiplier days, promotional credits, gaming-day boundaries, and account rules can still change what marketing sees.
Face value, internal cost, and offer design
After a comp budget is estimated, the casino decides what it can buy. A room, meal, resort credit, free-play award, and cash reimbursement are not interchangeable costs.
Consider an illustrative $60 internal budget:
| Benefit presented to player | Public value | Illustrative internal cost | Budget effect |
|---|---|---|---|
| Unsold standard room | $180 | $45 | May fit |
| Restaurant credit | $75 | $60 | May fit exactly |
| Free play | $75 | Depends on redemption and game return | Requires separate model |
| Cash | $75 | $75 | Exceeds budget |
These amounts are only an example of cost accounting, not a claim about a specific property. The point is that an offer can look generous at retail value while still staying inside an internal reinvestment limit. Conversely, a sold-out room night can be much more expensive to comp than the same room on a low-demand date.
Marketing also has to decide whether the benefit is meant to reward completed play, induce a return trip, fill unused capacity, protect a valuable relationship, or repair a service failure. The same theoretical value can therefore support different offers without the formula being inconsistent.
Theoretical value is used because actual outcomes are noisy
A recent peer-reviewed study of casino return behavior describes theoretical loss using average bet, decisions per hour, hours, and house advantage, while distinguishing that estimate from actual outcomes. See the study’s theoretical-versus-actual loss framework.
The regulatory side is less about prescribing a generosity percentage and more about control, accountability, and reliable records. Nevada’s table-games minimum internal control standards illustrate the broader environment in which ratings, table transactions, supervisory approvals, and audit trails must operate.
Neither source creates a universal comp rate. The formula estimates value; the casino’s policy decides how much value to return and in what form.
Four numbers players often confuse
| Number | What it measures | What it does not mean |
|---|---|---|
| Buy-in | Cash or chips brought to the game | Total amount wagered |
| Actual loss | Session result | Expected value of the play |
| Total action | Cumulative wagers | Final loss |
| Theoretical loss | Expected casino win from the rated action | Guaranteed casino win |
A player can buy in for $500, cycle $8,000 through the game, lose $100, and have a theoretical loss of $80. All four numbers can be true at the same time.
Why chasing a comp usually costs more than the comp
Suppose a player is $20 short of an internally estimated food-comp budget and decides to generate another $100 of theoretical loss to “earn” it. Even if the benefit is granted, the player has accepted $100 of expected gambling cost for a benefit with a lower retail and internal value.
The safe decision rule is simple: play only the amount you already considered affordable entertainment; treat the comp as a discount after the fact. Do not add hours, raise bets, or switch to a higher-edge game solely to move a reward meter.
For the components behind the calculation, use average bet, player rating, theoretical loss, and comp value. Why average bet matters explains rating accuracy; why casinos use loyalty programs explains the wider marketing purpose.