A slot machine’s house edge is the mathematical share of total wagers that the game is expected to retain for the casino over very large amounts of play. If the theoretical return to player is 94%, the corresponding theoretical house edge is 6%.
That does not mean a player who inserts $100 will lose $6. It also does not mean a session will finish anywhere near the theoretical percentage. The edge applies to coin-in—the total amount wagered across repeated spins—and short-term results can sit far above or below the long-run expectation.
The relationship is simple:
House edge = 100% - RTPSo:
94% RTP -> 6% house edge
96% RTP -> 4% house edge
90% RTP -> 10% house edgeFor the player-return side of the same equation, read slot RTP explained.
The percentage is charged against action, not the starting balance
Suppose a player begins with $100 and wagers $1 per spin. Wins are credited back to the balance and reused. After 500 spins, total coin-in is $500 even though the player never had $500 in cash at one time.
At a 6% theoretical house edge:
Expected loss = $500 × 0.06 = $30The expected ending balance is therefore $70 in a long-run average sense, but a real 500-spin session can end with much more or much less. The machine does not deduct 6 cents mechanically from every $1 spin. Individual outcomes remain random within the approved game mathematics.
This distinction between bankroll and coin-in is fundamental. A bankroll can be recycled many times. The edge is applied to each wager as part of the game’s expected-value structure.
Coin-in explains why small bets can become expensive
Total action is:
Coin-in = average bet per spin × number of spinsThen:
Expected loss = coin-in × house edgeConsider the same 6% edge at three levels of play:
| Average bet | Spins | Coin-in | Expected loss at 6% |
|---|---|---|---|
| $0.50 | 400 | $200 | $12 |
| $1.00 | 400 | $400 | $24 |
| $2.00 | 400 | $800 | $48 |
Nothing about the underlying percentage changed. The difference comes entirely from how many dollars were put through the game.
This is why “I only bet a dollar” is incomplete information. A dollar wager made 50 times is $50 of action. The same wager made 700 times is $700 of action.
Speed belongs in the cost calculation
Slots can resolve wagers quickly, so time matters.
A useful hourly estimate is:
Expected loss per hour = spins per hour × average bet × house edgeFor a $1 average bet at a 6% edge:
| Spins per hour | Coin-in per hour | Expected loss per hour |
|---|---|---|
| 250 | $250 | $15 |
| 500 | $500 | $30 |
| 750 | $750 | $45 |
These figures are not forecasts of the next hour. They are theoretical averages that show how pace multiplies the cost of a given percentage edge.
The time-on-device calculator is useful when you want to combine stake, pace, and session length rather than looking at house edge in isolation.
House edge and RTP are theoretical; actual hold is observed
Players often use house edge, hold, and RTP as though they are the same statistic. They are not.
- Theoretical RTP is the expected percentage of wagers returned to players under the game’s mathematical model.
- Theoretical house edge is 100% minus that RTP.
- Actual hold is what the casino actually retained over a measured period divided by coin-in.
- Actual player return is what players actually received over that period divided by coin-in.
A 94% theoretical RTP game can show 90%, 100%, or another actual return over a short sample because jackpots, bonuses, and ordinary variance do not arrive in perfectly even intervals.
Over sufficiently large play volume, actual performance is expected to move toward the theoretical model, but the speed of that convergence depends heavily on the game’s variance.
This is why a casino does not conclude that a game is “wrong” because one day or one player produced an unusual result.
The same RTP can produce very different sessions
House edge describes average price. Volatility describes how unevenly the return can arrive.
Two games could both have 94% RTP and therefore 6% theoretical house edge, while one pays frequent small awards and the other concentrates much more return in infrequent bonuses or jackpots.
The expected loss per dollar wagered is the same under the simplified 94% assumption, but the session experience can be radically different.
That is why these statements can all be true at once:
- a slot can have a relatively modest edge;
- a player can still lose most of a bankroll quickly;
- another player can win a large jackpot;
- neither result changes the published theoretical RTP;
- a short session cannot reveal the machine’s true long-run edge reliably.
See slot volatility for the separate question of session swing.
A 30-cent return on a $1 spin is still a 70-cent loss
Slot interfaces can make partial returns feel like wins because the machine may animate, count credits, or play celebratory sounds whenever an award occurs.
Mathematically, though, a $1 spin returning $0.30 has produced a net loss of $0.70.
This matters because hit frequency and house edge measure different things. A game can award credits often while still returning less than the amount wagered on many of those outcomes. Frequent events can change the rhythm of play without changing the expected-value equation.
The player should therefore evaluate the balance change, not merely whether the screen called an outcome a win.
The casino does not need every machine to win every hour
From an operations perspective, slot analysis is built around long-run performance data, not a belief that each machine must retain its theoretical percentage every shift.
Casino systems and accounting procedures commonly track measures such as:
- coin-in;
- actual win;
- actual hold percentage;
- theoretical hold percentage;
- denomination and paytable;
- changes affecting theoretical hold;
- performance variance between actual and theoretical results.
Nevada’s current Minimum Internal Control Standards for slots require records of theoretical hold information and changes affecting it, which illustrates the difference between a game’s approved theoretical characteristics and the actual results accumulated on the floor.
That recordkeeping also shows why “the casino turned this machine tighter because it paid yesterday” is the wrong model. Approved paytables and theoretical hold are controlled configuration data, not a live reaction to one player’s session.
A configuration can matter, but not in the way players often imagine
A cabinet may support multiple games, denominations, or approved paytables. Different configurations can have different theoretical returns. That means two visually similar machines are not guaranteed to have identical RTP.
But this is different from saying that a floor supervisor can press a button after a jackpot and secretly lower the next player’s odds. In regulated operations, game software, paytables, and changes affecting theoretical hold are subject to technical and internal-control requirements.
The practical player lesson is to use disclosed paytable or RTP information where available and not infer a configuration from recent wins or losses.
House edge is not the casino’s profit margin on a player
Another common mistake is to treat the theoretical edge as if it were the casino’s guaranteed profit margin on every customer. A 6% house edge does not mean the property earns exactly $6 whenever a player brings $100, nor does it mean the machine keeps 6% of cash inserted.
Theoretical win is tied to wager volume. Actual casino win over a period is affected by the exact sequence of outcomes, jackpots, promotional credits, progressive contributions, and how much play occurred. Operating profit is a different accounting question again because the casino also has labor, equipment, licensing, marketing, utilities, taxes, and other expenses.
For a player, the useful number is still the mathematical one: expected loss per dollar of eligible wagered action. For an operator, actual win and theoretical win are compared over large samples to understand performance rather than to guarantee a fixed margin from an individual session.
Bet size can change more than dollar exposure
On some slot games, changing denomination or wager level simply scales the same mathematical game. On others, a different denomination or bet option can activate a different paytable, bonus eligibility rule, progressive contribution, or jackpot schedule. If those changes alter the theoretical return, then the house edge can change as well as the number of dollars at risk.
That is why the correct comparison is not merely “50 cents versus $1.” Check whether the underlying paytable and feature eligibility are the same. If they are the same, doubling the wager doubles coin-in and expected dollar loss while leaving the percentage edge unchanged. If the configuration changes, both pieces of the calculation may move.
The slot bet-size guide deals with that distinction in more detail.
Expected loss example: bankroll versus recycled action
Suppose a player brings $150, wagers $1.25 per spin, and makes 600 spins on a game with a theoretical 92% RTP.
First calculate house edge:
House edge = 1 - 0.92 = 0.08 = 8%Then coin-in:
Coin-in = $1.25 × 600 = $750Then expected loss:
Expected loss = $750 × 0.08 = $60The $750 figure does not mean the player inserted $750. It means the same bankroll may have been wagered, partially returned, and wagered again until cumulative action reached $750.
The expected loss of $60 is not a guarantee that the $150 bankroll will finish at $90. A volatile game could end the session at $0, $300, or another value. The formula prices the action; it does not predict the path.
Comparing two RTPs requires the same amount of action
Suppose two games are identical for this example except for theoretical RTP:
- Game A: 96% RTP, 4% edge
- Game B: 92% RTP, 8% edge
At $1,000 coin-in:
Game A expected loss = $1,000 × 0.04 = $40
Game B expected loss = $1,000 × 0.08 = $80The lower-edge game halves the theoretical cost for the same $1,000 of action.
But if the player stays twice as long on Game A because the lower edge or session path keeps them playing, total expected dollars lost can converge again. Percentage edge and total action always need to be read together.
The RTP comparison tool helps make that tradeoff visible.
What house edge cannot tell you
A house-edge number cannot tell you:
- whether the next spin will win;
- how frequently bonuses will appear;
- how large the biggest session drawdown will be;
- how long a bankroll will last with certainty;
- whether a progressive jackpot will hit soon;
- whether a game has been “cold” or is “due.”
Those questions involve randomness, volatility, feature design, or misconceptions about independent outcomes.
House edge answers a narrower question: What is the long-run expected cost per dollar of action under the specified game configuration?
The most useful way to apply the number
If you want to reduce theoretical slot cost, there are only a few levers that matter directly:
- choose a higher RTP when reliable information is available;
- wager fewer dollars per spin;
- make fewer spins;
- shorten the session;
- avoid increasing stakes to recover losses.
Those actions reduce expected cost because they reduce the edge, the amount of action, or both. They do not change the random result of the next spin.
For a session-level estimate, use the expected-loss calculator or read slot expected loss per hour. For the broader probability model, continue to slot machine odds.
- “/slots/par-sheets/”