Hit cadence, free games, the jackpot tail and volatility all have to hold at once
Per-game samples of nearly a hundred PG slots, read by shape: a hit roughly every three spins, free games every hundred-odd spins that actually pay, a jackpot tail that really carries return, and deliberately spread volatility. Why copying the max win or the art alone never reproduces the feel.
Almost every slot team has been asked the same question: "Can you make one that feels like PG?" So they study the art, the animation, the max win multiplier, and build something that looks the part and plays nothing like it. This article skips the art and talks only about math: the PG "feel" is several numeric components holding at the same time. Copy one, leave the other three behind, and players notice within ten spins.
The usual approach is to push the max odds up to PG levels and pour effort into the free-game presentation. But the max odds are a single number most players never touch in an evening, and however lavish the free game looks, if it does not trigger twice in a session it is just a video clip. Both are outcomes, not causes.
Players do not compute RTP, but they remember cadence: how often they hit, whether a hit feels like anything, how long until the next free game, and whether it pays when they get there. Every one of those is a distribution, not a number. Two games with identical RTP, one hitting every three spins and the other every five, feel like they came from different companies.
This article is based on per-game samples of tens of thousands of spins across nearly a hundred PG titles, with hit rate, free-game trigger interval, max odds, odds distribution and volatility tabulated for each. We deliberately avoid absolute RTP values, for a simple reason: RTP is the last dial anyone turns, while the shape of the distribution is the design intent. Each section below covers one component: where it sits, then why it sits there.
A player's first impression of a game is set in the first ten spins. In those ten spins PG wants exactly one thing: the player hits at least three times, and every hit is visible.
Line up the hit rates of nearly a hundred titles and the vast majority crowd into a narrow band around "a hit roughly every three spins", with very few drifting either way. That is not coincidence, it is a spec: whatever the theme, whether the cap is in the hundreds or the thousands, the first ten spins should feel the same.
A sizeable share of hits are "you won, but less than you bet". In PG titles nearly a fifth of all spins are these small wins. Their job is not return, it is sound and light: during the hundred-odd spins spent waiting for a free game, something happens every few spins. Remove them and concentrate the return into larger wins, and the hit rate drops and the game feels dry, even with RTP untouched.
If hit rate decides the first ten spins, free games decide the first hundred or two. Whether a player settles in for the evening depends on "how often does it trigger" and "is it worth it when it does". PG gives a clear answer to both.
In more than half the titles, the free-game trigger interval lands between one and two hundred spins; most of the rest only drift slightly either way. In other words, whichever PG title a player picks, they can expect a free game within a reasonable session. The cadence is deliberately pinned: too frequent and free games become cheap, too rare and players quit before seeing one.
The trigger interval is only the doorway; what matters is what you get inside. Overall, PG places roughly a third of total return inside free games, and in some titles more than half. That means a single free game pays, on average, about what a player would accumulate over dozens of base-game spins. Nobody walks out empty-handed, and that is where "worth waiting for" comes from.
Moving return into free games makes the base game thinner. That is the trade PG makes: the base game carries cadence and spectacle, the free game carries return. Many teams build the opposite, a fat base game and a thin free game, so players enter the feature and think "that's it?". That disappointment hurts more than never triggering at all.
The jackpot is the easiest component to imitate and the easiest to misunderstand. Teams see PG titles advertising thousands of times the stake, raise their own cap to match, and find it changes nothing. The difference is that PG's tail really does carry return.
PG's max odds mostly start in the hundreds, and close to a fifth of titles reach the thousands. That cap is not decoration on the marquee; it is the apex of the whole distribution. From sub-stake wins to a few times, tens, hundreds of times the stake, every band has probability, all the way to the top.
The key fact is the allocation: wins above a hundred times the stake take nearly a quarter of total return. Put another way, of every four units a player bets, roughly one comes back only through those rare big wins. Parking that quarter in the tail means the base game and free games must leave room in the middle, which loops back to why the base game is thin.
The first three components are tightly clustered; volatility is the one PG deliberately lets loose. The product line has mild titles, medium titles, and roughly a third sitting in the high-volatility zone. The calmest and the wildest can differ several-fold.
This is a product-line strategy: hit rate and free-game cadence make every title "feel like PG", volatility makes every title "feel different". Switch games and the first ten spins feel familiar, but the curve over a whole evening can be completely different. That retains more kinds of players than a line where every game is equally calm or equally wild.
Many teams assume a high hit rate means low volatility, so to build a high-volatility game they suppress the hit rate. But split the PG sample into the higher-hit-rate half and the lower half, and the median volatility is nearly identical on both sides. Volatility comes from how heavy the tail of the odds distribution is; hit rate comes from how many spins win. The two dials do not pull on each other. To go high-volatility, move return toward the tail and leave the hit rate at every three spins.
Most PG titles let players buy the free game directly. Numerically it is not a different mode of play; it is the option to remove the waiting.
Look at a bought free game on its own and its volatility is a fraction of the base game's. The reason is straightforward: most of the base game's variance comes from the uncertainty of when the free game will arrive. Buy it and that uncertainty disappears, leaving only the free game's own distribution.
Same title, one player spinning through and another buying in: the two experience completely different curves. The spinner lives through the full distribution of hit rate plus free-game cadence plus jackpot tail; the buyer lives only through the free-game segment. That is also why buy-feature pricing needs independent verification: its distribution is not the same object as the base game's.
Looking back, "feels like PG" is four things holding simultaneously: a hit every three spins, a free game every hundred-odd spins that actually pays, a jackpot tail that really carries return, and volatility that is high overall but different per title. The four constrain each other; drop one and the other three stop working.
The most common failure is "cap right, everything else wrong": raise the max odds, then suppress the hit rate to balance RTP, and the result is tight from spin one with sparse free games, the exact opposite of the PG feel. The second most common is "hit rate right, free game thin": a fat base game, nothing inside the feature, anticipation collapses. Neither is an art problem or an RTP problem. The components are mismatched.
If you are building a game that is "supposed to feel like PG", set RTP aside, run your own sample, and answer four questions:
Answer those four and you know which component is off. That is also the first thing we do for clients: no art changes, no RTP changes, just run the distribution and look at the shape.
Further reading: we applied the same method to JILI's three math choices, a completely different answer that holds just as well. For how hit rate and odds distributions are designed mathematically, see natural probability design; for verifying that a measured distribution matches its spec, see the risk-control algorithm; and for how these cadences are actually felt on the client side, see the reel engine.