Hold percentage
Hold percentage is the share of the money taken in by a game that the casino actually keeps as win over a set period, written as win divided by drop on table games or by coin-in on slots. It is an accounting result, not a property of the game itself. Operators and regulators use it to track how a table or machine performed, while players mostly meet it when it gets mixed up with house edge or RTP.
How hold percentage is calculated
The top line is always gaming win, meaning what the house retained after paying out winnings. The bottom line changes with the game, and that is where most confusion starts. US federal gaming rules, for instance, define actual hold as win divided by drop or coin-in, calculated per table, per machine or per game type, daily or cumulatively.
On a table, drop is the value of chips bought and cash placed in the drop box, so it counts each buy-in once. On a slot, coin-in is every credit staked, including winnings played back through the reels. Because the two bases measure different things, a 15% table hold and a 15% slot hold describe quite different economics, and definitions vary between jurisdictions and operators.
Hold can even turn negative for a period.
Worked example: hold against house edge
The rows below follow one illustrative roulette session, using the published 2.70% edge of single-zero roulette and round figures chosen for clarity, not taken from any casino’s reports.
| Item | Figure | How it arises |
|---|---|---|
| Drop (buy-in) | €100 | Chips bought once at the table |
| Total wagered | €1,000 | 100 spins at €10, money recycled |
| House edge | 2.70% | Single-zero roulette, per bet |
| Expected loss | €27 | 2.70% of €1,000 wagered |
| Expected hold | 27% | €27 kept from €100 drop |
| Slot equivalent | Coin-in base | Long-run hold nears 100% minus RTP |
Why hold is not RTP or edge
House edge and RTP are theoretical figures attached to each bet, set by the game’s rules and maths. Hold is what happened to real money over a real period, so a single night proves nothing about whether a game is fair or “due”.
The gap in the example comes entirely from recycling: the same €100 was staked ten times, and the 2.70% edge applied to every stake. That is why a table hold of 27% sits alongside a modest edge without any contradiction, and why a high hold figure is no evidence of a rigged game.
When a site quotes a percentage, what I check first is which base it uses.
