I have been trading Counter-Strike skins since 2015. That means I was around for the era when jackpot _was_ CS:GO gambling — you threw skins into a pot, a wheel spun, one person walked away with everything. If you played in 2016, you played jackpot. There was not much else. Last month I sat down and did something I had been putting off: I went through every site I have an account on and tabulated what each one actually runs today, game mode by game mode. Not what their landing page advertises — what is actually in the lobby when you log in. I expected jackpot to be down. I did not expect this. Jackpot is on 11 of 42 sites ---------------------------- Twenty-six percent. It is now rarer than plinko. Here is the full breakdown, sorted by how many of the 42 sites run each mode:

Game mode	Sites	Share
Coinflip	27	64%
Roulette	25	60%
Case opening	24	57%
Crash	23	55%
Raffles	20	48%
Plinko	18	43%
Case battles	18	43%
Upgrader	16	38%
Blackjack	15	36%
Poker	14	33%
Esports betting	13	31%
Jackpot	11	26%

Below that it falls off a cliff — dice and mines on four sites each, keno and towers on two, and baccarat, limbo and hi-lo on exactly one apiece. Why it happened, and it is not what people usually say ------------------------------------------------------ The usual explanation is Valve's 2016 crackdown, and that is part of it. But Valve hit every mode, not just jackpot, and coinflip survived at 64%. The real reason is structural, and it is about **liquidity**. Jackpot needs a crowd. A pot with two people in it is just a slow coinflip. A pot with twenty needs twenty people online at the same moment, on the same site, willing to wait for it to fill. That was fine in 2016 when four or five sites had all the traffic. Coinflip needs one other person. Crash needs nobody at all — the multiplier climbs whether anyone else is playing or not, so a crash game with three players online works exactly as well as one with three thousand. When the market fragmented across dozens of sites, the modes that survived were the ones that work with thin traffic. Jackpot did not fragment well. The handful of sites still running it properly are mostly the ones with enough concurrent users to fill a pot in seconds — and there are not many of those left. The thing I got wrong --------------------- I used to think mode count was a decent proxy for whether a site was serious. More games, bigger operation, right? The numbers killed that idea. The average site in my table runs 5.8 modes, but four run ten or more and three run exactly one. And the single-mode sites are not the bad ones. A site that only does case opening and does it properly is a more honest proposition than one that bolted nine half-finished mini-games onto a landing page to look like a casino. Mode count is a menu. It says nothing about whether the food is any good. What I actually check now ------------------------- After going through all of them, the order that matters turned out to be almost the reverse of what gets advertised: 1. **Can you withdraw?** Depositing is frictionless everywhere — that is the whole design. Getting money back out is where sites differ, and the verification steps that only appear when you try to leave are the ones worth knowing about before you deposit, not after. 2. **House edge.** This is the number that decides your result over time. On case opening it is routinely above 10%, which is far worse than any other mode in the table. Most sites do not publish it, and on several you have to derive it from the drop odds yourself. 3. **Can rounds be verified?** Provably fair is checkable in about two minutes if a site implements it properly. Copy the server seed hash before you play, rotate the seed afterwards, run the revealed seed through SHA-256, compare. If it does not match, the site changed it. That test is binary and there is no "close enough". 4. **The welcome bonus.** Last, and by a distance. A large offer attached to a 40x wagering requirement is worth less than a small one you can actually withdraw. Where the data lives -------------------- I keep the full table on my own site — 42 sites against 13 game types, sorted alphabetically rather than ranked, because it is an index and not a top list: [https://www.csgopromocodes.com/cs2gamblingindex](https://www.csgopromocodes.com/cs2gamblingindex) It also names the twelve sites I do **not** have verified game-mode data for, rather than filling those gaps in from their marketing pages. I would rather publish a hole than a guess. * * * **Disclosure, because it matters and Hive deserves it straight:** that site earns affiliate commission when someone signs up to a listed platform. It costs you nothing extra and commission rate is not part of how I score sites, but you should know it before you read anything I write about this subject. I am telling you because every comparison site in this niche has the same incentive and most of them bury it. **Source of the numbers:** my own accounts on each platform, checked during August and early September 2026. Where a site had changed its lobby since I last logged in, I re-checked rather than trusting my notes. **18+ only.** Every mode in that table has a house edge, which means the expected result over enough bets is a loss. That is arithmetic, not an opinion, and verifying a round does not change it — it only confirms the loss was honest. Never deposit skins or money you cannot afford to lose. Free confidential help at [BeGambleAware](https://www.begambleaware.org/).

I Tested 42 CS2 Gambling Sites. The Mode That Defined the Era Is Almost Gone.