Pragmatic Play · 2021 · tested 2026-09-29
Starlight Princess
Base-game orbs made the 430x record of the first Starlight Princess demo run without free spins, and that run's fifty bonus buys still finished ahead at 1.07 times their price.
- RTP measured / declared 100.0% 96.5%
- Hit frequency 29%
- Max win · no bonus 858x
- Bonus buy · mean result 100x · 1.03x
- Free spins, natural 74 / 30,000
- Bonus frequency 2.5 bHz
In numbers: measured RTP 100% against 96.5% declared over 30,000 logged rounds in 3 runs; 28.7% of rounds paid; a natural bonus landed once per 405 rounds (74 in total) and a bonus round paid a median 56.1x the stake, mean 96.9x, trigger spin included; 150 bonus buys at 100x returned 1.03x of their price on average, 67% of them below the price; the biggest round paid 858.0x; the longest dry run was 23 spins.
Test card
30,000 spins · USD 2.00 bet · every request logged| Spins | 30,000 · 3 runs |
| Bet | USD 2.00 |
| RTP declared | 96.50% |
| RTP measured | 99.95% (approximate 95% CI 88.0–111.9) |
| Hit frequency | 29% · every 3.48 |
| Dead streak max / avg | 23 / 3.45 |
| Median hit | 0.80x |
| Max win (no bonus round) | 858.0x · run 3, round 7933 |
| Free spins, natural | 74 · 2.5 bHz · every 405.4 spins · mean bonus round payout 96.9x stake |
| Balance start → end | 100,000 → 99,971 |
- 012026-09-0410,000 rounds · 50 buysfiles & checksums ↓88.17%RTP5 files · SHA-256 ↓
- 022026-09-2110,000 rounds · 50 buysfiles & checksums ↓101.78%RTP5 files · SHA-256 ↓
- 032026-09-2910,000 rounds · 50 buysfiles & checksums ↓109.90%RTP5 files · SHA-256 ↓
each run keeps its own spin log, CSV and raw server journal with checksums; the figures on this card combine all runs
Play Starlight Princess demo
PRAGMATIC PLAY · official demo · same math as real play
United States
Starlight Princess available at Cafe Casino
Same reels, real payouts
Cafe Casino runs the same Starlight Princess with the same math and adds a welcome bonus for players from the United States:
300% up to $2,000 + 150 FS 18+ · T&Cs apply · the demo stays free
Bonus buy lab
150 buys × 100x · USD 200 eachmean 1.03x · median 0.602× price (USD 120.35) · 101 of 150 below price · best 10.85x · worst 0.03x
| Spins per bonus (avg / min–max) | 15.4 / 15–25 |
| Retriggered | 6.7% |
| Multiplier total (avg / max) | 16 / 56 |
| Bonuses that paid nothing | 0% |
Three scatters during the feature add five spins; only an increase beyond the initial 15-spin award counts.
- Spent
- USD 30,000150 × USD 200
- Returned
- USD 30,895.50102.99% · approximate 95% CI 83–123%
- Net
- +USD 895.50+4.5 buys
- Break-even or better
- 49 / 15033%
- Paid 2× price or more
- 22 / 15015%
- Price in base spins
- 100natural trigger every ~405 spins
One ticket equals 100 ordinary stakes. Natural bonuses occurred once per 405 ordinary rounds in this sample. Stakes multiplied by this interval measure gross wagering, not the net cost of obtaining a bonus. Purchase returns above include the complete purchased round and are divided by the ticket price.
Risk profile
how much bankroll the 3 runs in sequence needed| 50 bets | bust at spin 70 |
| 100 bets | bust at spin 193 |
| 200 bets | bust at spin 750 |
The runs are played back to back as one series, one bet size. A different session can bust earlier or never; treat this as a measured example, not a rule.
How it plays
7 rules screens from the game
Starlight Princess runs on the same engine as Gates of Olympus, on a 6×5 scatter-pays grid: a set of eight identical symbols is enough to pay, then those cells clear and new ones drop in. The difference is cosmetic: sun, heart, moon and star emblems at the top of the paytable, coloured gems below, and the princess as the scatter. The longest base-game chain of the first run was 11 steps (round 8996).
Orbs are the volatility: any spin can land multiplier orbs worth 2x to 500x, and when the tumbling stops every orb on screen is summed and applied to that spin's total. In the base game that sum is applied once and then forgotten, which is what turned a four-step chain into 430x on round 8581 of the first run, and a seven-step chain into 232x on round 7751 of the same run.
Fifteen free spins start when four or more princess scatters land; inside the feature the orb values accumulate into one total multiplier that grows through remaining free spins, and three scatters add five spins. In the first run none of the 22 natural bonuses retriggered and 4 of the 50 bought ones did.
Two optional levers exist: paying 25% more per spin doubles the scatter count on the reels, and the buy button starts the feature for 100x the stake. The server quotes 96.5% whether the stake is ordinary, Ante, or a purchase, and the game advertises a 5,000x ceiling.
Cascading reels
Multiplier
Free spins
Retrigger
Scatter symbol
Bonus buy
Extra bet
Bet range USD 0.20 – 300 per spin · payouts read from the game's paytable at bet USD 2.00
Gallery
9 frames from the demo session
Verdict
Records still come from ordinary spinsThree runs of 10,000 spins returned 88.17%, 101.78% and 109.9%, 99.95% pooled over 30,000 rounds and an approximate 95% CI from 87.98 to 111.92%, with the printed 96.5% inside it and the combined balance 15 stakes down, USD 29.40 short of where it started. The shape did not change from run to run: a hit on 29.34%, 27.96% and 28.84% of spins, a 0.8x median win in each, blank stretches of 18, 23 and 21 spins at the longest, and 7,066, 7,204 and 7,116 rounds paying nothing.
The records still come from ordinary spins: 858x on round 7933 of run three, 430x on round 8581 of run one and 351.6x on round 5014 of run two all arrived without a scatter, and 20 of the 43 rounds above 100x in 30,000 spins were base-game tumbles. The feature came 74 times, one per 405 spins, 2.5 bHz (natural features per 1,000 rounds), paid 96.91x on average and 56.08x at the median, 23.9% of the return, with 23 rounds over 100x, 13 under 20x and the smallest, round 1492 of run one, at 3x with free spins that paid nothing; four of the 74 retriggered, and the longest wait was 1,341 spins in run two. Without round 1644 of run three, the 642.9x feature, the other 29,999 rounds returned 97.81%.
The 150 purchases at USD 200 returned 1.030x of price, 102.98%, USD 30,895.50 for USD 30,000: a median ticket of 0.602x, USD 120.35, 101 losing tickets, 49 that covered the price, 22 above 2x and two above 5x, purchase 39 of run three at 10.85x and purchase 8 of run two at 5.101x. The three batches went 1.0738x, 1.0898x and 0.926x, so the buy has now lost money in one run of three, and the bought feature averaged 15.4 spins against 15.3 for the natural ones.
- Twenty base-game rounds above 100x in 30,000 spins, led by 858x, 430x and 351.6x without a scatter
- Hit rates of 29.34%, 27.96% and 28.84% at a 0.8x median across three runs, no blank stretch past 23
- The 150 purchases returned 1.030x of price, with 22 tickets at 2x or more
- Pooled 99.95% over 30,000 rounds, above the declared 96.5%
- Triggers arrived once per 405 spins, with waits of 1,177, 1,341 and 979 spins across the runs
- One feature's free spins paid nothing and 13 of 74 paid under 20x, against a 56.08x median for all 74
- Two tickets in three lost money over 150 purchases, and run three's batch returned 0.926x
- The first run's 1,941-stake trough is still the deepest point on the page
Run 01·2026-09-04·88.17%
The base game had real teeth on this path. Eighteen base-game rounds passed 50x in ten thousand spins, five passed 100x, and the 430x on round 8581 arrived two spins after the balance hit its lowest point, 1,940 bets down. That single spin is why the last thousand spins of the log show a 144% return while the first thousand show 82%.
The feature is steady rather than spectacular in this run: 22 triggers, all 15 spins long, 63x on average, with 202x on round 3216 and 167x on round 977 at the top and one round (1492) whose free spins paid nothing at all. Waiting for it took 455 spins on average, and the gap between two triggers went as long as 1,178 spins.
The buys deserve a careful read. Fifty purchases at 200 returned 10,738 in total, 1.07x of their price, which makes Starlight Princess the first game in this lab where buying finished in profit. But the median buy was 0.8325x, 32 of 50 lost money, 19 returned less than half, and the whole edge came from nine rounds above 2x, the best 4.48x on buy 29. Treat the 1.07x as one sample, not a promise.
Run 02·2026-09-21·101.78%
The second series returned 101.78% of stake against the printed 96.5%, with an approximate 95% band of 83.1 to 120.5%, and closed 178 stakes up after a 792-stake drawdown that bottomed out at spin 7634. Something paid on 27.96% of spins at a 0.8x median, the longest empty run was 23, and the ordinary reels delivered the two biggest rounds: 351.6x on round 5014, where a six-step chain closed with USD 673.90 of orb-multiplied wins, and 340x on round 7699.
Free spins landed 25 times, one per 400 spins with a 1,341-spin gap at the longest, and paid 97.9x the stake on average, 62.35x in the middle; nine rounds cleared 100x, led by 324.05x on round 7874, and two retriggered. Fifty purchases at USD 200 returned 1.0898x of price, 108.98%, but the median ticket came back at 0.7732x, USD 154.65, 32 lost and 21 returned less than half; ticket 8 paid USD 1,020.20 with a 28x orb total.
Run 03·2026-09-29·109.90%
Round 7933 was a four-tumble base spin worth USD 1,716, 858 stakes, the largest round of the series so far, and it lifted the balance to a peak of 1,361 stakes up on the next spin. The ordinary reels produced seven rounds above 100x in all, 224.7x on round 4689, 179.2x on round 9904 and 175.5x on round 68 behind the record, and the free spins added eleven more. Hits landed on 28.84% of spins at a 0.8x median, 7,116 rounds paid nothing and the longest blank stretch ran 21 spins from round 36, and the run finished at USD 101,980, 990 stakes ahead, 109.9% of the USD 20,000 staked, having been 534.7 stakes down on spin 1643, one round before its biggest feature.
Twenty-seven features came one per 370 spins, 2.7 bHz, and paid USD 6,685.90 in all, 30.4% of the return, 123.81x on average and 81.35x at the median: eleven cleared 100x, four finished under 20x, with round 3494 the smallest at 5.8x, and two rounds retriggered to twenty spins. Round 1644 was the feature of the run, USD 1,285.80, 642.9 stakes, over twenty spins with a 291x tenth spin and a 211.2x thirteenth; the next three, rounds 9957, 5407 and 7056, finished between 282x and 286x. The longest wait was 979 spins, from round 7302 to 8282, and the test bankrolls ran out on spins 67, 792 and 1049, smallest to largest.
Tickets bought at USD 200, fifty of them, produced USD 9,259.10, 0.926x of price, 92.59%, the median ticket paying USD 81.30: 37 lost money, 13 covered the price, four passed 2x and one, purchase 39, paid USD 2,169.70, 10.85x, USD 1,797 of it on its eleventh spin under a 32x orb total. Forty-eight bought rounds ran fifteen spins and two started with twenty from five scatters; the worst, purchase 48, returned USD 6.50.
Test record
3 runs · 30,000 rounds · 150 buys · open data · closed rigThis game has been through 3 separate test runs on the official demo server, each a fresh session of 10,000 rounds at a fixed bet. Every figure on this page combines the runs into one sample, as if they were one long session; the record below keeps each run apart, with its own numbers and its own downloadable log.
- 30,000rounds, all runs
- 29%rounds that paid
- 858xbiggest round
- 1 / 405natural bonus rate
- 1.03xmean of 150 buys
- 23longest dry streak
01run Run 012026-09-04 lab note measured return88.17%-8.3 pts vs declared
- 95% band
- 73.4–103.0%
- rounds that paid
- 29%
- biggest round
- 430x · round 8,581
- longest dry streak
- 18
- natural bonuses
- 22 · one per 455 rounds
- bonus buys
- mean 1.07x · median 0.833x · 64% below price
- balance start → end
- 100,000 → 97,635
- deepest drawdown
- 1,941x
Corrected 29 September 2026: this first run was recorded by an early rig that dropped the tumbles of the last free spin, so 7 of its 22 natural features looked smaller than they were. Free spins money now comes from the final bonus response, and the mean natural feature rises from 53.26x to 58.63x the stake in free spins alone, 62.78x for the whole round. Round wins, the RTP and the trigger count are unchanged, and the published spin log keeps the figures as first recorded.
02run Run 022026-09-21 lab note measured return101.78%+5.3 pts vs declared
- 95% band
- 83.1–120.5%
- rounds that paid
- 28%
- biggest round
- 352x · round 5,014
- longest dry streak
- 23
- natural bonuses
- 25 · one per 400 rounds
- bonus buys
- mean 1.09x · median 0.773x · 64% below price
- balance start → end
- 100,000 → 100,356
- deepest drawdown
- 792x
Second run, seventeen days after the first: the return rose from 88.17% to 101.78%, and the 83.1–120.5% band holds the printed 96.5%. Features came slightly more often, 25 against 22, and paid more, 98x the stake on average against 63x; the base game again produced the two largest rounds, 351.6x and 340x, and the 50 buys returned 1.09x of price after 1.07x in the first run.
03run Run 032026-09-29 lab note measured return109.90%+13.4 pts vs declared
- 95% band
- 83.1–136.7%
- rounds that paid
- 29%
- biggest round
- 858x · round 7,933
- longest dry streak
- 21
- natural bonuses
- 27 · one per 370 rounds
- bonus buys
- mean 0.93x · median 0.407x · 74% below price
- balance start → end
- 100,000 → 101,980
- deepest drawdown
- 827x
Dated 29 September, the third run came in at 109.9% after 88.17% on 4 September and 101.78% on 21 September, and the three runs together sit at 99.95% over 30,000 rounds, above the printed 96.5% for the first time as a series. The base game kept its records, 858x on round 7933 against 430x and 351.6x before, and the features paid more again, 123.81x on average against 62.78x and 97.9x; the fifty purchases slipped to 0.926x of price after 1.0738x and 1.0898x.
3 test runs combined into one sample (2026-09-04, 2026-09-21, 2026-09-29): 30000 ordinary paid rounds, including naturally triggered features, at a fixed USD 2.0 bet; 150 bonus buys at 200.0 each (100.0x) on the official Pragmatic Play demo server (jurisdiction 99, RTP profile 96.5%); every request and response logged.
The test rig plays the official demo build over the network and records every server round: bet, win, intermediate steps (cascades, respins), free spins, collect. Screenshots and video were captured in a separate browser session and are not part of the statistical samples. The text was read against the log by an editor before publication. How we test →
FAQ
7 questions · answered from the logDoes the base game or the feature pay more in Starlight Princess?
The base game, by share: natural features made 23.9% of the pooled return, so ordinary tumbles made the rest, and the three largest rounds in 30,000 spins, 858x, 430x and 351.6x, came without a scatter. The feature still owns the median: 56.08x per round against 0.8x for a typical paying spin.
At what rate do the princess scatters trigger?
Seventy-four triggers in 30,000 spins, 2.5 per 1,000 rounds, one trigger every 405 spins on average; run one waited 1,177 spins at the longest, run two 1,341 and run three 979, and the first trigger of run three came on spin 435.
Which round produced the 858x record?
It was round 7933 of run three, four tumbles in the base game, USD 1,716 at the USD 2 stake; the balance rose to 1,361 stakes up on the next spin, and it beat the 430x record of run one, another four-tumble base round, and no free-spin feature in three runs came closer than 642.9x, round 1644 of run three.
What has 150 purchases shown about the 100x ticket?
Over 150 tickets the ticket returned 1.030x of its price, 102.98%, with 22 tickets at 2x or more and two past 5x, but 101 tickets came back short and USD 120.35 was the middle result. By run the batches went 1.0738x, 1.0898x and 0.926x, and the best ticket, purchase 39 of run three, paid USD 2,169.70 with USD 1,797 on one spin under a 32x orb total.
How much do orbs matter inside the feature?
They set the whole result: in round 1644 of run three a single spin under the accumulated orbs paid 291x and another 211.2x, and the round finished at 642.9x over twenty spins, while four of the 27 features that run finished under 20x. Across 150 bought rounds the orb total averaged 16.1x and peaked at 56x.
Was the Ante bet used in any of the runs?
No: all 30,000 spins and 150 purchases were played at the ordinary USD 2 stake without Ante, so the doubled scatter count it buys for 25% more per spin is not measured here. The server quotes 96.5% for ordinary play, Ante and the 100x purchase alike.
Did the balance ever recover from the first run's slide?
No, not inside run one, which closed 1,182.6 stakes down after a 1,940.5-stake trough at spin 8579; run two closed 177.8 stakes up and run three 990 stakes up, so the pooled series ended 15 stakes below its start. The 200-stake test bankroll lasted 750, 495 and 1,049 spins in the three runs.
Also tested
4 more slots on the same rig-
Might of Ra
Pragmatic Play · 40,000 spins
2026-10-03
2026-10-15
RTP
91.3% 96.5%
hit27%
max291x
buy—
-
Twilight Princess
Pragmatic Play · 10,000 spins
2026-09-15
RTP
110.9% 96.1%
hit20%
max60x
buy1.08x
-
Sugar Rush
Pragmatic Play · 40,000 spins
2026-09-04
2026-10-06
RTP
101.6% 96.5%
hit34%
max650x
buy0.88x
-
The Dog House
Pragmatic Play · 30,000 spins
2026-09-04
2026-09-29
RTP
87.8% 96.5%
hit28%
max151x
buy—
Field notes
0 notes