TL;DR500 Casino says Towers has
"a fixed house edge set at 5% for all 3 difficulty modes". The game actually takes the 5% again on
every row you climb, so the edge grows the higher you go:
Cash out at You get back House edge
------------ ------------ ----------
Row 1 95.00% 5.00%
Row 2 90.25% 9.75%
Row 5 77.38% 22.62%
Row 10 (top) 59.87% 40.13% <- 8x the advertised 5%
The same paytable is documented
since 2023. We estimate players were paid
$3.5m to $21.6m less than advertised over three years.
If you played Towers since 2023 and cashed out past row 1, you were paid less than 500 advertised. We're asking 500 to pay it back. Save your bet history and bet IDs.
How it worksFixed 5%: payout = 0.95 x (1/p)^n -> 95% back at every row
500: payout = (0.95/p)^n -> 0.95^n back
p = chance of a safe pick per row, n = rows cleared
open Towers on Medium. Row 2 shows 3.61x. A 5% edge pays 3.80x.
$100 win on Medium 500 pays Fixed 5% pays Paid less
------------------ ---------- ------------- ----------
Row 4 $1,303.21 $1,520.00 $216.79
Row 10 $61,310.66 $97,280.00 $35,969.34
500's own pages define house edge per bet. The Towers game page says
"The game's RTP is typically around 95%". Their house-edge article says Towers works like Duels, where 500 takes
"a fixed 5% rake for every played duel": once per game.
Edge by cash-out row Row 1 Top row
--------------------- ----- -------
Stake Dragon Tower 2% 2%
Shuffle Tower 2% 2%
500 Towers 5% 40.13%
It's the server, not just the display500's server records, Nov 2024 (270 row payouts, 27 paid) 270 / 270 compounded
Public bet feed, Oct 2026 (other players' wins) 330 / 330 compounded
Our live test, 9 Oct 2026 (33 games) all balances reconcile
The 2023 help pages already list a ~613x Medium top multiplier. A fixed 5% would top out at 972.8x.
Provably fair passes. The payouts don't.We replayed all 33 test games from the revealed seeds:
2/2 commitments,
33/33 results,
53/53 tiles. The verifier proves where the bombs were.
It never checks what a win pays. 500's iTech Labs RNG certificate doesn't cover the paytable either.
Reproduce itgit clone https://github.com/ProvablyFair-org/500casino-towers
cd 500casino-towers
python3 -I -B analysis/run_all.py
python3 -I -B tests/run_negative.py
# Python 3 + Node.js, no packages, no account, no network
What 500 should do1. Publish the payout formula's history and the dates each version ran.
2. Review every affected player and pay the difference back.
3. Fix the paytable to a true 5%, or clearly advertise the real edge.
We'll update the report and this thread if 500 responds with reproducible evidence. Happy to answer questions here too, or run any extra check someone suggests.
-PF Team
