CSDB · CS2 Gambling
The Maths
Three calculators for the questions people ask after a session rather than before one: where the money actually went, whether that losing run was unusual, and what a bonus is really worth — plus what a case opening returns against what it costs, for the cases where every outcome can be priced. Nothing on this page links to an operator.
Gambling involves risk and most players lose money over time. These sites are for adults only, and skin gambling is restricted or illegal in some countries — check your local law before depositing. Never stake more than you can comfortably lose or gamble to recover losses. Free, confidential help: Nothing on this page is a strategy. No stake size, cash-out point or system changes the expected outcome — that is the one thing the maths is completely clear about. BeGambleAware · National Gambling Helpline 0808 8020 133 (UK, 24h) · NCPG 1-800-GAMBLER (US)
Where the money goes
Every game on every site in this section returns less than it takes. The gap is the house edge, and the crucial detail is when it is charged: on every wager, not on every deposit. A balance staked repeatedly pays the edge repeatedly, so the amount you deposit and the amount you wager can differ by an order of magnitude while both numbers are perfectly true.
The edge is charged on every wager, so what matters is not how much you deposit but how many times you stake it. Recycling the same 100.00 through 500.00 of wagers costs the same as losing 15.00 outright — which is why “I only deposited 100.00” and “I wagered 500.00” are both true at once.
The same arithmetic, applied to a case
Valve publishes the drop rates but not the return, so the return has to be built from the outside: take every outcome a container can produce, weight it by its published probability, price it, and compare the total to what an opening costs. That is possible for 16 of the 42 weapon cases we hold data for — and only those 16appear in the table, because a case whose knife tier we cannot price cannot honestly be given a percentage at all. The 0.26% gold tier is a rounding error in probability and a large share of the value — frequently most of it — so leaving it out would not make the answer slightly rough. It would make it far too low, in the direction that flatters the argument this page is already making.
This is not a ranking.The table is alphabetical, and the ordering is meant to carry no information whatsoever. Every case listed returns less than it costs; the differences between them are differences in how much is lost, not in whether. Nothing here identifies a case worth opening, because on these numbers there isn’t one.
| Case | Cost to open | Expected item value | Estimated return | Expected loss per open |
|---|---|---|---|---|
| Chroma 2 Case | $5.90$3.41 case + $2.49 key | $4.60 | 78% | −$1.30 |
| Chroma 3 Case | $5.48$2.99 case + $2.49 key | $4.38 | 80% | −$1.10 |
| Chroma Case | $6.55$4.06 case + $2.49 key | $4.79 | 73% | −$1.76 |
| Clutch Case | $3.00$0.51 case + $2.49 key | $0.60 | 20% | −$2.40 |
| Dreams & Nightmares Case | $3.59$1.10 case + $2.49 key | $1.89 | 53% | −$1.70 |
| Gamma 2 Case | $5.22$2.73 case + $2.49 key | $3.38 | 65% | −$1.84 |
| Gamma Case | $5.27$2.78 case + $2.49 key | $3.52 | 67% | −$1.75 |
| Glove Case | $13.06$10.57 case + $2.49 key | $3.75 | 29% | −$9.31 |
| Operation Broken Fang Case | $7.94$5.45 case + $2.49 key | $2.19 | 28% | −$5.75 |
| Operation Hydra Case | $31.09$28.60 case + $2.49 key | $9.56 | 31% | −$21.53 |
| Operation Riptide Case | $10.85$8.36 case + $2.49 key | $3.16 | 29% | −$7.69 |
| Recoil Case | $2.79$0.30 case + $2.49 key | $0.60 | 21% | −$2.19 |
| Revolution Case | $2.72$0.23 case + $2.49 key | $0.51 | 19% | −$2.21 |
| Snakebite Case | $2.90$0.41 case + $2.49 key | $0.81 | 28% | −$2.09 |
| Spectrum 2 Case | $5.08$2.59 case + $2.49 key | $3.11 | 61% | −$1.97 |
| Spectrum Case | $5.90$3.41 case + $2.49 key | $3.44 | 58% | −$2.46 |
How the figures are built. Every outcome valued at its cheapest listed wear across markets (price snapshot 2026-08-27), with StatTrak folded into non-special drops at 10%. A real drop can roll a better wear, so this is a floor rather than a central estimate; those are also buyer prices, before whatever a market charges to turn a skin back into money. Knife and glove pulls are valued as non-StatTrak, because the odds we work from publish no StatTrak share for the gold tier. Skin prices are a snapshot taken and case prices ; both move, and the return percentages move with them. Valuing every drop at its cheapest listed wear means the real average sits somewhat above these numbers — and market prices are what a buyer pays, so getting that value back out costs a fee on top. The two pull in opposite directions, and neither is anywhere near large enough to turn a case that returns a fraction of its cost into one that returns the cost.
The remaining 26 cases are listed here rather than quietly dropped, because an absent case looks like an oversight and a stated reason does not:
- 26 cases — at least one knife or glove it can drop has no price in our data — plain, un-finished knives are the usual case. For these, RTP cannot be independently calculated from published data: CS:GO Weapon Case, CS:GO Weapon Case 2, CS:GO Weapon Case 3, CS20 Case, Danger Zone Case, eSports 2013 Case, eSports 2013 Winter Case, eSports 2014 Summer Case, Falchion Case, Fever Case, Fracture Case, Gallery Case, Horizon Case, Huntsman Weapon Case, Kilowatt Case, Operation Bravo Case, Operation Breakout Weapon Case, Operation Phoenix Weapon Case, Operation Vanguard Weapon Case, Operation Wildfire Case, Prisma 2 Case, Prisma Case, Revolver Case, Shadow Case, Shattered Web Case, Winter Offensive Weapon Case.
Was that losing streak actually unusual?
This is the question behind most “this site is rigged” posts, and it has a real answer. Runs that feel impossible are usually likely: over a few hundred rounds, a streak most people would call suspicious is closer to inevitable.
A run of 8 losses happens in roughly 35.8% of 200-round sessions. Uncommon, entirely normal, and not evidence of anything. Either way, the next round is unaffected by the last one: the odds do not “correct” after a bad run, and no stake size changes the edge.
What a bonus is really worth
A deposit bonus is not money until it has survived its wagering requirement, and the requirement is what determines whether it is worth anything at all. The turnover has an expected cost, and where that cost exceeds the bonus the offer has negative value before you start.
At 30× and a 3% edge the expected value is 10.00. That is an average across many attempts rather than what any one person gets, and it assumes every published term is honoured exactly as written. Game weighting usually makes this worse: where slots count 100% and other games count 10%, the real turnover needed multiplies accordingly.
The published promo-code termswe hold for each operator record the requirement where one is stated — and note where the operator publishes no figure at all, which is common.
Frequently asked
Why do I always lose at CS2 gambling?
Because the games are built to return less than they take, and the shortfall is charged every time you stake — not every time you deposit. Recycling the same balance through many rounds pays the edge repeatedly, which is why a modest deposit can produce a large total wagered and a balance of zero. That is the designed outcome, not a malfunction and not evidence the site is rigged.
Is a long losing streak proof a site is rigged?
Almost never. Streaks that feel impossible are routinely likely across a few hundred rounds — the calculator above works out how likely for your own numbers. If you want to check a specific result rather than a feeling, use the provably-fair verifier, which settles that exact question for operators that publish seeds.
Does changing my stake size or cash-out point improve my odds?
No. Every staking system changes the shape of your results, not their expected value. Doubling after a loss wins small amounts repeatedly and then loses everything at once when a long run arrives, which it eventually will. A low cash-out multiplier wins often and loses heavily; a high one loses almost always and occasionally pays. They lose at the same rate over time.
Can a CS2 case’s RTP actually be calculated?
Sometimes. It needs every outcome the container can produce to have a price, including the 0.26% knife or glove tier, which holds most of the value — price only the outcomes that happen to be listed and the answer comes out far too low. That test is passed by 16 of the 42 weapon cases we hold data for, and their figures are in the table above. For the other 26, usually because the gold pool contains a plain un-finished knife with no market price, RTP cannot be independently calculated from published data, and no number is published for them.
Which CS2 case has the best return?
That question has no answer worth acting on. Every case whose outcomes can all be priced returns less than it costs to open, so the differences between them are differences in the size of the loss. The table above is alphabetical deliberately: sorted by return it would read as a shopping list, and the ordering is not information anyone should be using to choose a case.
Are deposit bonuses worth taking?
It depends entirely on the wagering requirement, and often the answer is no. A bonus at 30× with a 3% edge is expected to cost more in turnover than the bonus is worth, which makes it a reason to deposit rather than free money. Work it out with the calculator above before accepting one, and check the maximum-cashout term as well as the multiplier.
If this page landed badly
Working out that a run of losses was mathematically ordinary does not make it feel any better. If gambling has stopped being fun, taking a breakcovers deposit limits, blocking software and free independent help — none of it run by any operator. The case simulator costs nothing and uses the real published CS2 drop rates if you want the opening without the loss.