How do casinos calculate theoretical loss?

Casinos calculate theoretical loss by applying an expected house advantage to the amount of wagering action recorded for a player.

For a table game, the common model is:

Table theoretical loss = average bet × decisions per hour × hours played × house edge

For slots and other tracked machines, the common model is:

Machine theoretical loss = coin-in × theoretical hold percentage

These are estimates of long-run expected value. They are not invoices, guarantees, or statements of what the player actually lost.

Step 1: record the action correctly

The calculation is only as good as its inputs.

A table-game rating generally needs:

• player identity or account;
• game and table;
• start and stop time;
• average wager;
• relevant side action;
• an assumed or measured pace;
• the edge assigned to that game and rule set.

A machine system can usually record wagered volume more directly when the player uses a card or account. It captures coin-in, game identification, time, and other meter or transaction data.

The casino is not primarily asking how much cash entered the player’s wallet or how much was inserted once. It is asking how much money was repeatedly put at risk.

A player may buy in for $500 and generate $8,000 of table action by recycling chips. A slot player may insert $200 and produce $2,500 of coin-in by replaying credits. Theoretical loss is based on action, not the initial cash amount.

Step 2: estimate a real average bet

The largest wager is not the average. Neither is the opening wager.

If the bet changes during the session, the best simplified estimate is time- or decision-weighted:

Weighted average bet = total estimated action ÷ total decisions

When only time blocks are available, a time-weighted approximation can be used.

Suppose a blackjack player wagers:

• $25 for 90 minutes;
• $100 for 30 minutes.

The time-weighted average is:

[($25 × 1.5 hours) + ($100 × 0.5 hours)] ÷ 2 hours = $43.75

Calling this player a $100 average would greatly overstate the action. Calling the player a $25 average would understate it.

Real rating systems vary. Supervisors may update averages during play, use chip observations, record separate main and side wagers, or apply approved automated table data. The Average Bet page explains the measurement problem in more detail.

Step 3: convert time into decisions

Sitting at a table for two hours does not mean every table produces the same amount of action.

Pace depends on:

• number of players;
• dealer speed;
• game type;
• shuffle method;
• side-bet procedures;
• player decisions and disputes;
• fills, credits, markers, and interruptions;
• live-dealer presentation such as baccarat squeezing.

If the casino assumes 70 blackjack hands per hour and records two hours, it estimates 140 decisions for one player who wagered on each hand.

Theoretical loss can then be written as:

Average bet × estimated decisions × house edge

The Decisions per Hour glossary entry shows why a pace assumption can materially change a rating.

Step 4: assign the game edge

The edge used for rating may be a standard internal assumption rather than a personalized calculation of every decision.

A perfectly played blackjack game and a poorly played one do not have the same actual expectation. A baccarat Player wager and a high-edge side bet do not have the same price. Craps players can combine line bets, odds, place bets, and propositions with different edges.

Casinos deal with that complexity in different ways. A system may:

• use one standard edge for the main game;
• apply separate rates to major wager categories;
• include or exclude free-odds action;
• use a blended edge for typical play;
• adjust a rating when unusual strategy or side action is material.

No single universal rating edge applies to every casino. The formula is common; the operating assumptions are property-specific.

Full table-game example

Use the weighted blackjack average from above:

• weighted average bet: $43.75;
• estimated pace: 70 hands per hour;
• time: 2 hours;
• assigned house edge: 0.8%, or 0.008.

Theoretical loss = $43.75 × 70 × 2 × 0.008 = $49.00

The player might actually win $900 or lose $700. Neither result changes the $49 estimate produced by those rating assumptions.

The model says that if comparable action were repeated many times under the assigned edge, the casino would expect to win about $49 per such session on average.

Machine example: coin-in, not cash inserted

Suppose a tracked slot session records:

• coin-in: $8,400;
• theoretical hold: 8.5%, or 0.085.

Theoretical loss = $8,400 × 0.085 = $714

The player may have started with much less than $8,400 because wins and remaining credits were wagered again. Coin-in counts every bet.

Theoretical hold is the casino-side complement of theoretical RTP:

Theoretical hold = 1 − theoretical RTP

A 91.5% theoretical RTP corresponds to an 8.5% theoretical hold, before considering any separate promotional value.

Regulated reporting often distinguishes designed or theoretical RTP from actual achieved RTP. The UK Gambling Commission’s live RTP monitoring definitions make that distinction explicit.

Theoretical loss for mixed play

A trip can contain several games, each with a different price and measurement method. The casino should calculate each segment and then add them.

The total is not the amount the player must lose. It is the combined expected value under the assumptions used for each segment.

This method is more defensible than applying one percentage to all gambling, because the games do not have the same edge, pace, or tracking quality.

Why table ratings are less exact than slot meters

Machine coin-in is recorded transaction by transaction when tracking works correctly. Table ratings often include human observation and standard pace assumptions.

Possible table-rating errors include:

• start time opened late;
• stop time left open after the player departed;
• average bet not updated after a sustained change;
• one large wager mistaken for the normal amount;
• side bets omitted or overstated;
• breaks counted as active play;
• wrong game or edge selected;
• multiple spots not recorded correctly.

Casinos use supervisory review, exception reporting, player-tracking controls, and reconciliation to reduce these errors. Nevada’s table-games minimum internal control standards provide one official example of the control environment around table records and operations.

A player can politely ask whether the rating is open and whether the approximate average is being captured, but the floor is not required everywhere to disclose the full model.

Actual loss and theoretical loss answer different questions

Actual result is what happened:

Actual result = cash and chips leaving play − cash and chips entering play, adjusted for credit and other transactions as required.

Theoretical loss is what the casino expected from the measured action.

Casinos use both. Actual loss matters for accounting, credit, service recovery, risk, and trip context. Theoretical loss is often more stable for marketing because one unusually lucky or unlucky result does not redefine the long-run value of the play.

The dedicated explanation Why Casinos Track Theoretical Instead of Only Actual Loss covers that policy choice.

Comps are a second calculation

A common mistake is to treat theoretical loss as the amount of comps owed.

A property may apply a reinvestment guideline:

Preliminary comp budget = theoretical loss × reinvestment rate

If the blackjack theo is $49 and the illustrative reinvestment rate is 20%:

$49 × 0.20 = $9.80

If the mixed trip theo is $546.80 at the same rate:

$546.80 × 0.20 = $109.36

These figures are planning examples, not universal entitlements. Casinos value rooms, food, free play, events, host discretion, prior offers, and internal cost differently. Credit status, profitability, market strategy, and responsible-gambling controls can also affect a decision.

Read How Casinos Calculate Comps for that separate step.

What theoretical loss cannot tell you

Theo cannot predict the next hand or guarantee a trip result. It does not measure enjoyment, affordability, harm, or the wisdom of continuing. It can be wrong when the rating inputs are wrong. It can also create false precision when a system displays cents from estimates that were rough to begin with.

For a player, the safest use is cost awareness:

• estimate total action before the session;
• apply the relevant edge;
• treat the result as an average, not a maximum loss;
• allow for variance and a much larger plausible cash swing;
• never increase play only to improve a rating or earn a comp.

Theoretical loss is a planning estimate built from measured action. The arithmetic is simple; the quality of the inputs is the difficult part.

Related reading

Continue with Theoretical Loss, Average Bet, Coin-In, Player Rating, and Why Average Bet Matters.