Where the numbers come from

A threshold can be reached two ways.

The first is derivation. A par sheet, the manufacturer's design document for a machine, lists reel strips, hit frequencies and paytables. From those, theoretical return can be calculated directly, and the point where it crosses 100% follows from the arithmetic.

Analysts at Wizard of Odds and Wizard of Vegas work this way. Both have published results and the methods behind them, across forum threads and video, for years. It is the most rigorous work in the field.

Access is the constraint. Par sheets are confidential documents distributed to operators, not players. Copies circulate. Each describes one configuration of one release. Many par sheets list no useful information.

The second method is observation: play the game, record what happens, and correct the estimates until they hold. This method is slower and it produces a starting point rather than proof. It is also how most working thresholds in this field are built, by actual advantage players.

For the most part, Edge Finder uses the second.

A new game arrives on a floor.

The first question is whether the game can be beaten at all. Advantage players read the rules screen looking for clues, and observe whether something accumulates, whether it survives a cash-out, whether the next person to sit down inherits it.

Most titles fail here. The vast majority of machines on any floor never enter a positive state.

Games that show promise get additional scrutiny. We observe how quickly the state builds, how the game dynamics work. We observe whether behaviour differs across denominations and bet levels, and any other pertinent nuances of the game.

Observing is free. At some point, as confidence builds, APs will start trying the game at different threshold amounts. For low volatility games, threshold adjustments are relatively easy to make. For highly volatile games, sometimes it takes years and a coordinated online effort to gather a dataset worth interpolating.

Once a test threshold is established, record keeping becomes crucial to solidify the findings and monitor software updates. Several of the most popular current games have changed substantially since they were released. Continued monitoring ensures that material game updates do not go unnoticed. (Playing Buffalo Ascension the same way as when it came out is a good way to lose money.)

Figures are then compared against other advantage players keeping their own logs, and against published analysis from people with reputable track records. Agreement raises confidence. Disagreement is taken seriously and investigated.

Once a threshold number becomes established and reasonably actionable, it is added to the library.

Operators configure identical titles differently. Settings vary between properties, and occasionally even among identical games in the same bank.

A single universal figure would be simpler to publish but it would not tell the whole story. The guides often give a range instead. You can start with a more conservative number and adjust toward more aggressive play depending on the property and any other goals and considerations. (Promotions, mailers, and so on.)

False precision is less useful than a tight range that can be adjusted.

A threshold is not final. It moves when:

• A manufacturer revises the game or issues new software
• A property alters its configuration
• Accumulated session data shifts the estimate
• Recent results contradict the published figures

A downloaded list decays, and the cost of that surfaces on the floor. Staying on top of thresholds is the subscription. At least 4 new guides are added each month alongside it.

A threshold describes where a game turns favourable across a substantial number of trials. It carries no information about any individual session, and variance is severe enough on some games to produce extended losses on entirely correct plays. If a game is particularly volatile, we generally include a warning and minimum bankroll recommendations. A cardinal sin in advantage play is to start a play one can't finish.

Good thresholds are not a replacement for good personal records.

None of it is financial advice. Read the full disclaimer.