Gates of Olympus 1000 on Duel: the Rakeback Pilot, Tracked to Eight Decimals

Tested 17 July 2026 · 200 settled paid wagers in two blocks · USDT · Account with the 50% Instant Rakeback badge · Protocol v1.0 pilot (replication audit)

Gates of Olympus 1000 at a glance

RTP values were read from the in-game information screen on Duel on the test date. Pragmatic Play ships lower configurations of the same title (95.5%, 94.5%) to some casinos, so the information screen where you play is the only number that counts — how to read it in a minute.

What this page proves — and what it doesn't

Two hundred spins cannot measure a slot's RTP. On a 5/5-volatility game with a 15,000x ceiling, short-session returns swing so widely that any "observed RTP" from a test this size would be noise dressed up as data, and we did not try. What a short session can verify is the rakeback model, because accrual is supposed to be deterministic: 50% of the house edge on every settled paid wager, win or lose. That claim is falsifiable with modest turnover, and it is the claim we tested — this was the pilot for the whole series, and the only test read at eight decimal places rather than to the cent. The full protocol is on the methodology page.

What Duel's tooltip claims

A caution on the last line. "Duel RTP" is the interface's own aggregate, not engine RTP: the rebate is not part of the game's mathematical model. It accrues outside the round, sits in a separate counter until you claim it, and depends on your account carrying the badge. We call the combined figure a conditional effective return, and we treated 98.25% as the claim under test, not as a given.

How we tested

We claimed the counter down to zero, confirmed the zero reading twice a few minutes apart, then ran two independent blocks with different stakes and different game modes:

Both blocks ran through Duel's Hyperplay batch mode, which logs every spin in a per-spin table — a cleaner evidence trail than screenshots of a counter. The accrual readings come from the expanded rakeback widget, which shows eight decimal places; the compact display truncates to cents. Every bet reconciles against the balance ledger to the cent: $500.01 − $50.00 + $21.50 = $471.51 after Block A, then − $40.00 + $16.56 = $448.07 after Block B.

Results: one rate, replicated to eight significant figures

The rate is identical across both blocks to eight significant figures — different stake, different mode, same share of turnover. Both readings sit inside the range implied by the rounded published RTP (96.495–96.505%), so both grade as Matched within displayed RTP precision, and the identical rate across blocks earns the Replicated flag. The +0.10% deviation from the naive expectation is the most interesting number on this page.

The accrual appears to use an unrounded internal reference rate

0.875 is what "96.50%" predicts. What accrued implies a house edge of 3.5035% — a base RTP of 96.4965%, which Pragmatic rounds to the published 96.50%. Assuming Duel applies exactly the advertised 50% share, the observed rate implies a reference edge of about 3.5035% for this game — more precision than the information screen shows. One honest caveat: a single measured rate cannot distinguish "exact edge × exactly 50%" from, say, "exactly 3.5% × 50.05%". We report the measured accrual rate as the fact; the decomposition assumes Duel's stated 50%, which is the more plausible reading but not the only one.

Ante Bet made no difference — as this game's maths says it shouldn't

Block A ran with Ante Bet on, Block B without it, and the accrual rate did not move. That fits the information screen, which lists the same 96.50% for the ante mode. It cannot be generalised from this game: where ante or buy modes carry a different RTP, whether the rebate follows the mode or a flat badge figure has to be tested separately — Sweet Bonanza 1000 was that test, and it found one flat badge rate. Later tests confirmed the same on every game where the cents could tell the difference.

What it costs across the modes

The rebate is the same 1.75% of turnover in every mode, because it runs off the badge; on this game the base and ante modes are the badge mode, so the rebate is exactly half of what you pay. Halved is still not cheap: a low-edge table game costs less. Whatever the mode, the cost is (edge − rebate) × turnover; the rakeback guide has the arithmetic for any game. Net edge is rounded to two decimals.

Claiming: instant credit — and it withdraws

Claiming moved $1.57 to the main USDT balance on the spot — $448.07 became $449.64 — and the counter reset to zero. The credited funds sit alongside deposits, and we checked the last step the obvious way: claimed rakeback was later withdrawn from the account like a deposit, with no wagering requirement applied. One sub-cent footnote: displays truncate to cents, so whether the fractional tail of a claim (always under one cent) transfers or is discarded sits below display resolution. Accrual itself is cumulative and precise, so claiming infrequently makes the question academic.

Session return: 42.3% — and why that is not an RTP measurement

Across 200 spins the game paid back $38.06 on $90.00 wagered: a 42.3% session return, 43.0% in Block A and 41.4% in Block B. An anecdote, not an estimate — it says nothing about the long-run RTP of Gates of Olympus 1000 and everything about what 5/5 volatility does to a 200-spin sample. It makes the series' point better than a diagram: the slot returned less than half our stakes, and the rakeback still accrued at the same reproducible rate in both blocks. Accrual tracks turnover, not outcomes.

Limitations

All of this was measured on one account, on one date, in USDT, with the 50% badge active — rates could differ by account, region or time, and Duel's terms can change after publication. The conclusions cover the accrual formula only, not session outcomes. Counter readings were taken from the site's own expanded widget; they reconcile with the ledger to the cent, but we did not intercept API traffic to verify the backend independently. The sub-cent claim behaviour described above remains unresolved below display precision.

Verdict

On this account, on this date, Gates of Olympus 1000 produced one reproducible rakeback rate across both tested blocks, in both game modes, at two different stakes — a rate consistent with Duel applying its advertised 50% share to an unrounded internal reference edge — and paid the claim into the main balance immediately, funds we later withdrew with no strings attached. The observed rate was reproducible and fell within the range implied by the displayed 96.50% RTP.

Related: the game's row in the catalog · bonus buys on Duel on the feature-buy page · the full results table on the rakeback guide.