Contents
Harrah’s led the nine Atlantic City casinos with 91.40% observed payback from September 2025 through August 2026—12 complete months. That is an aggregate slot-floor result derived from reported dollars, not a way to identify the best machine or a promise about your session.
Harrah’s also led each of the 12 monthly comparisons in this window. Even that consistent lead does not reveal individual game settings: the aggregate combines different machines, denominations, and amounts wagered.
If you are looking for the loosest slots in Atlantic City, use these figures to compare historical slot-floor results. Choosing a particular game still requires checking its full spin stake, rules, and any available information about the theoretical RTP of that version and configuration.
Observed payback at all nine Atlantic City casinos
Observed payback ranged from 89.94% at Bally’s to 91.40% at Harrah’s for September 2025–August 2026. The ranking uses each casino’s total slot win divided by its total handle over the full period.
All dollar totals below are exact USD amounts, not rounded millions or billions. Casino win means the amount retained by the casino, not winnings paid to players. Handle, also called coin-in, is the total amount wagered, including payouts wagered again.
| Casino | Observed payback, Sep. 2025–Aug. 2026 | Slot win, USD | Handle, USD | Monthly payback min–max within period |
|---|---|---|---|---|
| Harrah’s | 91.40% | 180,096,853 | 2,093,734,466 | 91.19–91.70% |
| Borgata | 90.68% | 572,514,567 | 6,141,963,521 | 90.30–90.98% |
| Ocean | 90.57% | 339,507,595 | 3,601,912,012 | 90.31–90.79% |
| Hard Rock | 90.46% | 384,390,658 | 4,031,119,897 | 90.06–91.05% |
| Resorts | 90.30% | 132,804,542 | 1,369,637,315 | 90.09–90.65% |
| Golden Nugget | 90.29% | 109,490,250 | 1,127,927,691 | 90.03–90.60% |
| Caesars | 90.17% | 157,991,485 | 1,608,024,906 | 89.78–90.86% |
| Tropicana | 90.03% | 174,533,737 | 1,750,830,795 | 89.55–90.42% |
| Bally’s | 89.94% | 100,067,857 | 995,017,657 | 89.50–90.41% |
Across all nine casinos, slot win totaled USD 2,151,397,544 on USD 22,720,168,260 in handle. Combined observed payback for September 2025–August 2026 was 90.53%, rounded to two decimal places.
The monthly ranges show how each floor’s aggregate result varied during the period. They are not ranges for individual machines and do not predict next month’s payback.
Treat close rankings cautiously. Resorts and Golden Nugget differed by approximately 0.01 percentage point using unrounded inputs. Their order describes the reported results; it does not establish a statistically significant difference. Numerical precision and statistical significance are separate issues.
What observed payback does—and does not—measure
Observed payback measures completed play over a stated reporting period; theoretical RTP describes the mathematical payout rate of a specific game version and configuration. Neither is the probability of winning on the next spin.
| Term | Meaning | What it does not tell you |
|---|---|---|
| Observed payback | The share of handle paid to players in reported results for a specified period | An individual machine’s setting or your future session payout |
| Theoretical RTP | The mathematical payout proportion stated at certification for a particular game version and configuration | A guaranteed payout during short-term play |
| Hit frequency | The proportion of spins with any payout | Payout size or overall payback; a payout can be smaller than the stake |
| Volatility | How widely game outcomes vary | Whether a game has high payback |
| Hold, or win percent | Casino win divided by handle | The proportion of spins won by players |
| Denomination | The monetary value of one wagering unit | The full cost of a spin |
These distinctions matter in practice: frequent payouts do not necessarily mean high payback, and a penny denomination does not necessarily mean inexpensive spins.
New Jersey has a general theoretical RTP floor of 83%, subject to exceptions, under current rules, section 13:69E-1.28A(a). It is not a minimum payout for your session. Page 220 of the same document addresses statistically independent random number generator values; neither the clock nor a short USD 20 trial establishes better odds. Specific game rules and bonus states still need to be considered.
How the ranking was calculated and verified
The ranking uses summed dollar amounts across all 12 months, rather than an arithmetic average of monthly percentages. This weights each month by its handle instead of treating a quiet month and a busy month equally.
The slot formulas are:
- Hold = total slot win ÷ total slot handle × 100%.
- Observed payback = (1 − total slot win ÷ total slot handle) × 100%.
The verification covered 108 slot rows: nine casinos across 12 months. For each row, calculated hold was checked against the reported rounded hold, and table-game win plus slot win was checked against total casino win.
In the New Jersey Division of Gaming Enforcement (DGE) reports, “Win” means the casino’s retained amount, “Handle” means total wagering turnover, and “Win %” expresses the casino’s share of that handle. Do not substitute player payouts for casino win.
Harrah’s August 2026 example
Harrah’s observed slot payback for August 2026 was 91.32%, not 91.40%. The latter figure covers the full September 2025–August 2026 window.
Page 6 of the August 2026 DGE report lists:
- Slot win: USD 18,313,841
- Slot handle: USD 210,929,878
Applying (1 − 18,313,841 ÷ 210,929,878) × 100% gives 91.32% observed payback, rounded to two decimal places. Calculated hold rounds to 8.68%. Subtracting the report’s printed, rounded hold of 8.7% from 100% yields only 91.3%; use the raw dollar amounts for a more precise calculation.
Original DGE August 2026 report fragment showing Harrah’s slot win and handle
Source: New Jersey DGE, August 2026 report, page 6; original English fragment. Harrah’s observed payback was 91.32% for August 2026 and 91.40% for September 2025–August 2026.
This illustrates why a monthly result cannot replace a rolling 12-month result. A longer window helps smooth short-term variation, but it does not identify machine settings or establish statistically significant differences between casinos.
Primary reports for reproducing the ranking
You can reproduce the ranking from the 12 direct DGE PDFs below: sum each casino’s slot win and handle, then apply the formula above.
| Reporting month | Primary report |
|---|---|
| September 2025 | DGE PDF |
| October 2025 | DGE PDF |
| November 2025 | DGE PDF |
| December 2025 | DGE PDF |
| January 2026 | DGE PDF |
| February 2026 | DGE PDF |
| March 2026 | DGE PDF |
| April 2026 | DGE PDF |
| May 2026 | DGE PDF |
| June 2026 | DGE PDF |
| July 2026 | DGE PDF |
| August 2026 | DGE PDF |
These sources support the specified 12-month comparison. Averaging 12 rounded monthly percentages can produce a different result because it ignores differences in monthly handle.
Why denomination and slot-floor mix matter
A whole-floor average can differ substantially from a denomination category’s result. In 2024, Harrah’s 1¢/2¢ category had 85.83% observed payback, while its whole slot floor had 91.63% for that same year.
Official denomination data are available in the DGE annual slot and table-game report for 2024, page 1.
| Casino and category | Reporting period | Slot win, USD | Handle, USD | Observed payback |
|---|---|---|---|---|
| Harrah’s, 1¢/2¢ category | 2024 | 56,599,492 | 399,348,458 | 85.83% |
| Harrah’s, whole slot floor | 2024 | 179,236,961 | 2,140,346,368 | 91.63% |
| Borgata, penny-slot category | 2024 | 173,341,291 | 1,120,548,202 | 84.53% |
The same-year Harrah’s comparison demonstrates the limitation of applying a whole-floor result to a category. The exact dollar inputs above allow the calculations to be reproduced; displayed percentages are rounded to two decimal places.
Original DGE annual report crop showing the 1¢/2¢ slot category for 2024
Source: New Jersey DGE, 2024 annual data, page 1; published table as of July 11, 2025. This original English crop shows only the 1¢/2¢ category, not the entire report.
Multi-denomination slots are reported as their own category. The penny-slot category therefore does not automatically include every wager made at a one-cent denomination.
The mix of games and the distribution of handle among categories influence a floor’s aggregate result. These figures cannot identify the best cabinet, establish a particular game’s configuration, or justify choosing a larger denomination solely because a category had higher historical observed payback. See the slot denomination guide for more on choosing a full spin stake.
The 2024 figures are also not a denomination breakdown of the September 2025–August 2026 ranking. Those are different reporting periods.
What equal turnover could cost in a conditional model
At USD 600 in turnover, the rounded September 2025–August 2026 rates for Harrah’s and Bally’s produce a difference of USD 8.76 in modeled expected loss. This illustrates scale, not a prediction of your session.
Assume 600 rounds at a full stake of USD 1 per round, producing USD 600 in turnover. For this illustration only, historical whole-floor hold is used as an assumed fixed loss fraction. It is not a certified theoretical RTP for your chosen machine.
| Conditional model | Observed payback used, Sep. 2025–Aug. 2026 | Assumed hold | Turnover | Expected loss |
|---|---|---|---|---|
| Harrah’s rounded rate | 91.40% | 8.60% | USD 600 | USD 51.60 |
| Bally’s rounded rate | 89.94% | 10.06% | USD 600 | USD 60.36 |
For the Harrah’s-based model, expected loss is 600 × 0.086 = USD 51.60. The expected gross payout is USD 548.40: USD 600 minus USD 51.60. That is gross payout, not net profit or a guaranteed ending balance.
This comparison uses rounded rates. The difference would be slightly different using unrounded report inputs. Actual session results can vary substantially from the model.
Turnover is not cash deposited. Rewagering payouts can produce more than USD 100 in turnover from a USD 100 starting deposit.
A higher stake can erase the modeled benefit of a higher percentage. In a separate hypothetical example over 600 rounds:
- USD 1 per round at an assumed 90% theoretical RTP: USD 600 turnover and USD 60 expected loss.
- USD 2 per round at an assumed 95% theoretical RTP: USD 1,200 turnover and USD 60 expected loss.
Those theoretical RTP assumptions do not refer to a casino or reporting period. At equal turnover, a one-percentage-point difference equals USD 10 in expected loss per USD 1,000 wagered.
How to reproduce the example in the calculator
Use the house edge calculator with 8.6% custom house edge, USD 1 per round, and 600 rounds to reproduce the conditional model based on Harrah’s rounded September 2025–August 2026 observed payback.
- Open the calculator.
- Enable “Enter a custom house edge”.
- Enter 8.6 in the custom house edge field.
- Enter 1 in the bet field.
- Select “By rounds”.
- Enter 600 for total rounds.
- Enter 600 for rounds per hour.
English calculator settings with an 8.6% custom house edge, USD 1 bet, and 600 rounds
Local English calculator interface illustration. The entered historical hold is a hypothetical proxy—an assumed fixed loss fraction—not a casino or machine test.
With those settings, the result shows:
- House edge: 8.6%
- Interface label “RTP”: 91.4%
- Wagered: USD 600
- Rounds: 600
- Expected loss: USD 51.60
- Expected return: USD 548.40
Local interface illustration, not a casino test. The “RTP” label here is only the complement of the entered 8.6% rate; it does not verify any game’s theoretical RTP.
At the assumed pace of 600 rounds per hour, the example also models USD 51.60 in expected cost per hour. That pace is an input assumption, not a measured speed of play at Harrah’s.
The calculator does not identify a casino, machine, or configuration. It does not calculate winning probability or the probability of exhausting your budget, and it does not predict the next spin. Its rating badges are not evidence that a casino is the best choice.
How to choose a game within an affordable budget
Start with an affordable full spin stake and rules you understand; use the casino ranking only as secondary context. Higher historical observed payback does not automatically offset more expensive spins.
Before playing:
- Set a firm spending limit. Expected loss is not a cap on what you could lose during a session.
- Check the full spin cost. Denomination is only the value of a wagering unit; the total depends on how many units the spin uses.
- Read feature and bonus rules. Check any added cost and eligibility requirements before increasing your stake.
- Check available information about the game version. Do not apply another version’s published theoretical RTP to the machine you selected.
- Consider pace and turnover. More rounds or larger stakes increase total wagering, even when the starting budget stays the same.
- Do not treat a short streak as a payback test. A few wins or losses cannot establish a machine’s long-run characteristics.
The ranking’s practical conclusion is limited: Harrah’s led aggregate observed payback for September 2025–August 2026. Finding an affordable stake and understanding the game remain separate decisions.
Can an RTP catalogue or online comparison help?
Yes, but they answer different questions: neither establishes that a particular machine is available in Atlantic City nor replaces land-based casino reports.
The RTP catalogue shows the published RTP for a game version and links to its supporting source. Use it to check published specifications and their source—not to assume the same version or configuration is installed on a particular slot floor.
For a separate online discussion, see the guide to loosest slots online. A published online game-version RTP, a monthly floor result, an annual denomination result, and a rolling 12-month casino result are not interchangeable. They concern different objects, periods, and measurement methods.
No same-period, like-for-like Atlantic City versus Las Vegas comparison is established here. The broader guide to casinos with the loosest slots provides geographic context, not proof of such a comparison.
Commercial disclosure
The Bovada link is a commercial affiliate link, not a recommendation based on this ranking. It has no role in calculating observed payback for the nine Atlantic City casinos.
Its inclusion does not establish that Bovada is approved by New Jersey regulators or available in the state. This article does not recommend offshore sites or make claims about offers or player acceptance.
Frequently Asked Questions
A players card is not connected to the random number generator or paytable. It does not establish that a machine has higher theoretical RTP.
Neither the clock nor a short USD 20 sample establishes better odds. Random number generator values are statistically independent, although the specific game’s rules and bonus state still matter.
No. The general requirement applies to theoretical RTP, is subject to exceptions, and is not a minimum payout for an individual session.
No. No comparison on the same basis and for the same period has been established here. This ranking covers only nine Atlantic City casinos from September 2025 through August 2026.
Evgeniy Volkov
Verified ExpertFullstack developer with a background in mathematics. I build the calculators and game-style tools on ToolsGambling with Pixi.js and modern web tech, and every result uses transparent probability formulas you can verify yourself.