Is Buying the Bonus Worth It? Bonus Purchases Measured Across the Lab

How the purchases were made

Of the 198 games tested so far, 134 offer a bonus purchase and 64 do not. For each game with a button, the rig bought the standard feature a series of times at its listed price, in a session of its own, after the base run was complete, so that purchased features never mix with natural triggers. Every purchase is recorded as a complete round: the price paid, the free spins that followed, the total returned, and that total divided by the price. The lab now holds 12,050 such purchases.

The average says yes more often than the typical ticket does

Each dot is a game: the mean return of its purchases as a multiple of the price. Games above the dashed line paid for their purchases across their logged series; games below did not. Across the lab, 44 of 134 games finished above the line and 90 below, with the average game returning 0.94x of the price. The strongest purchase series returned 1.60x, the weakest 0.42x.

The mean is the wrong number to stop at. Within almost every game the results are skewed the same way: a majority of tickets below the price and a few tickets several times above it, and those few carry the average. The median ticket, the one in the middle when the tickets are sorted, is the better description of what one purchase tends to do, and it sits below the price in nearly every series.

How many tickets lost money

Averaged over every game with a buy, 73% of individual purchases returned less than their price. That share stays high even in games whose series finished ahead: a purchase series can end ten percent in profit while well over half of its tickets lost, because the shape of the payout is the same everywhere and only the size of the top tickets differs. Someone who buys five features in a row is far more likely to see three losses and two modest wins than the smooth average the mean suggests.

The purchase and the natural bonus are different products

It is tempting to compare the bought feature with the one the base game triggers on its own. The lab keeps them apart on purpose. A natural bonus is a free event inside a base game that is already costing money; a purchase is a fixed price paid up front, and its return is counted against that price. Both are counted as the whole round, trigger spin included. The average natural bonus in the lab paid 77 bets, while most purchases cost a hundred, the price of the feature in 80 of 113 games on 16 October 2026; the two figures are not in the same currency and cannot be subtracted from each other. The free spins frequency ranking answers how long the base game makes a player wait; this report answers what the shortcut cost.

What a purchase series cannot tell you

A few dozen purchases are enough to describe the shape of a game’s purchase results: the median, the share of losing tickets, the size of the outliers. It is not enough to decide whether a game’s purchase RTP is above or below its base RTP, a difference that is usually a fraction of a point in the design and would need thousands of purchases to resolve. Nor is a profitable series a forecast: the games at the top of the table are there because one or two tickets in the series landed a large feature, and the next series may not include one. Each review states the game’s declared purchase RTP next to its measured series so that the two can be read together.

What this report will look like as tests are added

The charts and tables recompute from the current set on every load. As the sample grows, the share of losing tickets is expected to settle into a narrow band, and the number of games above the price line is expected to stay near the level implied by purchase RTPs a few points below 100%. If a game holds a mean well above the price across a second series, that game gets a re-test and a paragraph here.

Two games have since done so. Sword of Ares returned 1.35x of its 100x price in the series logged on 20 September 2026 and 1.15x on 14 October; Deadwood xNudge returned 1.19x and 1.22x of its 71x price on 3 and 15 October. Most of their tickets still lost money: 62 of 100 and 67 of 100.