Realized hold is the percentage a casino actually retained over a defined period using a stated activity base. It is an observed result after real play has occurred. It is not the same thing as theoretical hold, house edge, or expected value.
The definition becomes useful only when the denominator is named. In slot reporting, realized or actual hold is commonly statistical win divided by coin-in. In table-game reporting, hold is commonly win divided by drop. Coin-in and drop measure different kinds of activity, so the resulting percentages cannot be compared as though they mean the same thing.
Slot realized hold uses coin-in as the wagering base
For a slot product, a common performance calculation is:
[ \text{Realized slot hold %}=\frac{\text{statistical win}}{\text{coin-in}}\times100 ]
If a bank records $2,000,000 coin-in and $120,000 statistical win:
[ \frac{120{,}000}{2{,}000{,}000}\times100=6% ]
The realized hold for that sample is 6%. The corresponding realized return on that same coin-in base is 94%.
That 94% is an observed result for the period. It is not automatically the machine’s theoretical RTP.
Table hold usually uses drop, not total wagers
A common table-game operating calculation is:
[ \text{Table hold %}=\frac{\text{table win}}{\text{drop}}\times100 ]
If a blackjack pit records $500,000 drop and $75,000 win:
[ \frac{75{,}000}{500{,}000}\times100=15% ]
That 15% is hold against drop. It is not a 15% blackjack house edge.
Drop represents money or approved instruments entering the table-game inventory under the property’s accounting definition. The same chips can then be wagered repeatedly. A player who buys in once and recycles chips can create substantial wagering volume without creating equivalent new drop.
This is why Hold Percentage and House Edge must remain separate concepts.
Realized hold answers “what happened?”
Theoretical measures answer what the casino expects under defined assumptions over a sufficiently large sample. Realized hold reports the actual outcome observed in the selected data.
A slot bank with a 7% theoretical hold can realize 3%, 9%, or even a negative result over a short interval. A blackjack pit with a low mathematical edge on individual wagers can show a double-digit hold percentage against drop.
Neither result is a contradiction. The percentages use different denominators and are affected by real variance.
Theoretical Win explains the expected side of the comparison, while Actual Win covers the observed dollar result.
Hold variance needs percentage points and dollars
A useful slot performance comparison is:
[ \text{Hold variance}=\text{realized hold}-\text{theoretical hold} ]
Suppose realized hold is 6.0% and the applicable weighted theoretical hold is 7.2%:
[ 6.0%-7.2%=-1.2\text{ percentage points} ]
The percentage difference can be translated into an approximate dollar difference on the same coin-in base:
[ \text{Projected dollar variance}=\text{coin-in}\times(\text{realized hold}-\text{theoretical hold}) ]
On $2,000,000 coin-in:
[ 2{,}000{,}000\times(-0.012)=-24{,}000 ]
The observed result is $24,000 below that theoretical expectation for the selected sample.
That does not mean $24,000 is missing. It means actual statistical win differed from the theoretical benchmark by that amount. Investigation is needed only when the magnitude, pattern, controls, or data quality give a reason to investigate.
Sample size controls how much confidence the percentage deserves
A realized percentage based on very little activity can move violently. One jackpot or one high-value session can dominate the period. As volume grows, realized performance often becomes more informative, but volume alone does not guarantee a clean comparison.
Management should also verify:
- the correct time period;
- complete meter or drop data;
- consistent win definitions;
- game and denomination mix;
- paytable or configuration changes;
- progressive and jackpot treatment;
- promotional treatment;
- known system adjustments;
- and concentration in a few players or machines.
A daily percentage can be too noisy. A lifetime percentage can be too slow to expose a recent problem. Good review uses multiple windows.
See Sample Size and Variance.
Mixed slot products need a weighted theoretical benchmark
If a bank contains products with different theoretical holds, a simple average can be misleading. The correct comparison weights each product by its wagering volume.
Assume:
- Game A: $800,000 coin-in at 6% theoretical hold;
- Game B: $200,000 coin-in at 10% theoretical hold.
The weighted theoretical hold is:
[ \frac{800{,}000\times0.06+200{,}000\times0.10}{1{,}000{,}000}=6.8% ]
A simple average of 6% and 10% would be 8%, but that would ignore the fact that four-fifths of the wagering occurred on the lower-hold product.
The same principle applies when denomination, paytable, or game mix changes over time. A floor can realize a different aggregate hold simply because wagering shifted toward different products, even when every machine is operating exactly as designed.
Realized hold can be negative
Negative realized hold is possible over a short period when players win more than the selected wagering base produces in casino win for that interval.
For slots, a large jackpot can push statistical win below zero for the day. For table games, a major player can leave with more than the period’s measured drop. A negative result does not mean the game has a negative long-run house advantage. It means the selected sample captured a favorable player outcome large enough to overwhelm the casino’s other win.
This is one reason short-term hold should never be interpreted as a new “true house edge.”
A high realized hold can also be misleading
Management can make the opposite mistake by treating a high hold percentage as proof of excellent operating performance.
A high short-term result may come from favorable variance. A table can hold an unusually large share of drop because one major player lost quickly. A slot bank can show an unusually high percentage because high-volatility games happened to run strongly for the house during a small sample.
The result is real revenue, but it may not be repeatable.
A better review pairs realized hold with volume, theoretical expectation, player concentration, jackpot exposure, active hours, and comparable historical periods.
Reporting definitions can change the numerator
“Win” is not always identical across every operational, accounting, and tax report. Promotional credits, free play, progressive contributions, jackpot accounting, adjustments, and timing conventions can be treated differently depending on the report’s purpose.
Therefore, the formula is only as reliable as its definitions. If one analyst uses statistical win and another uses a tax-reporting revenue figure, both can calculate a percentage correctly and still be talking about different measures.
Report labels should state the numerator, denominator, period, and scope instead of assuming everyone uses “hold” the same way.
Current control procedures illustrate the slot comparison
Nevada’s slot internal-control materials provide one jurisdictional example of formal actual-versus-theoretical performance review. The state’s current slot procedures include controls for comparing actual hold with theoretical hold and examining percentage and projected-dollar variance. The applicable materials are published through the Nevada Gaming Control Board.
That example should not be turned into a universal reporting rule. Casinos must use the definitions and control requirements that apply to their own jurisdiction, approved systems, and accounting framework.
Realized hold is a starting point for investigation, not the explanation
A useful performance review asks:
- What exact win figure is in the numerator?
- What exact activity base is in the denominator?
- What volume produced the percentage?
- What theoretical benchmark applies to that same mix?
- Did configuration or mix change during the period?
- Is the variance plausible given volatility and concentration?
- Are the meters, drop records, jackpots, and adjustments complete?
Only after those questions are answered does the hold percentage become meaningful evidence.
The practical definition to keep
Realized hold is the actual retained percentage for a stated sample and stated denominator. For slots it is often statistical win divided by coin-in. For table games it is often win divided by drop. Those measures describe what happened in different operating systems; neither should be confused with the house edge on individual wagers.
The number is useful because it gives management an observed result to compare with expectation. It is limited because it cannot explain the cause by itself. Volume, variance, game mix, accounting definitions, and data quality determine whether the percentage is routine, informative, or worth investigating.