CS2 trade up guide: how trade up contracts work
Ten skins of one grade go in, one skin of the grade above comes out. That part everybody knows. The parts that decide whether it was a good idea are the odds, the float math and the price of the ten copies you burned.
The contract in one paragraph
A trade up contract takes ten weapon skins of the same grade and returns one skin one grade higher. The output is drawn at random from the next grade pools of the collections your inputs belong to, so the collections you feed it are the collections you can win from. Every input must be either all StatTrak or all ordinary, and the output follows suit. Since October 2025 there is a second form of the contract: five Covert skins produce one knife or glove item from those same cases, which is the only route from ordinary skins to a gold item.
The rarity ladder
Trade ups walk one rung at a time. Covert is the top of the ordinary ladder, which is why Coverts cannot be ten input material: above them there is only the five Covert knife contract.
| Grade | Inputs | You get |
|---|---|---|
| Consumer Grade | 10 | Industrial Grade |
| Industrial Grade | 10 | Mil-Spec Grade |
| Mil-Spec Grade | 10 | Restricted |
| Restricted | 10 | Classified |
| Classified | 10 | Covert |
| Covert | 5 | Knife or gloves |
Knives, gloves, agents, stickers, charms, cases, capsules and the Contraband M4A4 Howl are not eligible as inputs at all.
How the game picks your outcome
Each input carries exactly one tenth of the probability mass, and it carries it to its own collection. Add up how many inputs came from a collection, then split that share evenly across every next grade skin in that collection. Nothing is weighted by price, by wear or by how badly you want the good one.
- Ten inputs from one collection with three possible outputs:
10/10 ÷ 3, so each outcome sits at 33.3 percent. - Seven from a collection with two outputs and three from a collection with five:
7/10 ÷ 2gives those two 35 percent each,3/10 ÷ 5gives the other five 6 percent each.
This is the whole lever you get. A collection with a small pool concentrates your chance; a collection with a big pool dilutes it. Mixing is not cheating, it is the design.
So what are the odds, in practice?
The formula is one thing. The pools it divides into are fixed by Valve, and they are a lot smaller than people expect. We counted every collection in the catalogue we ingest on 6 September 2026: 92 collections have at least one rung where a contract is possible at all, which comes to 315 collection and grade pairings. This is what the pool looks like at each rung, for a contract fed entirely from one collection.
| Contract | Collections that can run it | Pool sizes you meet | Most common pool | Odds per outcome there |
|---|---|---|---|---|
| Consumer → Industrial | 38 | 1 to 9 | 4 outputs (42%) | 25% |
| Industrial → Mil-Spec | 46 | 1 to 6 | 4 outputs (43%) | 25% |
| Mil-Spec → Restricted | 88 | 1 to 6 | 5 outputs (35%) | 20% |
| Restricted → Classified | 78 | 1 to 4 | 3 outputs (50%) | 33.3% |
| Classified → Covert | 65 | 1 to 2 | 2 outputs (65%) | 50% |
Three things fall out of that, and they run against the way trade ups usually get talked about.
- The top rung is not a lottery. Not one collection in the game holds more than two Coverts. 42 of the 65 collections that can run a Classified contract hold exactly two, which is a coin flip, and the other 23 hold exactly one. Feed ten Classifieds from The Arms Deal Collection and the output is an AWP | Lightning Strike, every time. From The St. Marc Collection it is an AK-47 | Wild Lotus. From The Ancient Collection, an M4A1-S | Welcome to the Jungle.
- 53 of the 315 pairings, one in six, have a pool of one. There are no odds to weigh on those at all. The question collapses into two: what do the ten inputs cost, and what float do they land you.
- The dilution lives at the bottom. The worst odds in the game are 1 in 9, and only two contracts offer them: Consumer Grade inputs from The Ascent Collection or The Boreal Collection. Every other contract in CS2 is 1 in 6 or better.
| Outputs in the collection's pool | Chance of any one of them | Pairings at this size |
|---|---|---|
| 1 | 100% | 53 |
| 2 | 50% | 87 |
| 3 | 33.3% | 75 |
| 4 | 25% | 49 |
| 5 | 20% | 43 |
| 6 | 16.7% | 6 |
| 9 | 11.1% | 2 |
Those are single collection contracts. Mix two collections and each one's share is its count of inputs over ten, then split across its own pool, so a mixed contract can land on odds no single collection offers. Counted on 6 September 2026 from the open ByMykel CS2 catalogue we ingest, weapon skins only: knives, gloves, agents and the Contraband M4A4 | Howl are not trade up eligible and are left out. Want it the other way round, from the skin you actually want back to the ten that feed it? The reverse trade-up calculator names the collection, the pool and the odds against the rest of it.
The float math nearly every guide gets wrong
The formula you will read on most sites is average input float × output range + output min.
That was true years ago. CS2 normalises first, and on any skin whose float range is capped the difference is
big enough to change the wear tier you get out.
Three steps:
- Normalise each input inside its own range:
(float − min) ÷ (max − min). - Average those normalised positions across the ten inputs (or five, for the knife contract).
- Map the average into the range of whichever outcome you got:
min + average × (max − min).
Take ten copies of the AK-47 Green Laminate from the 2021 Vertigo Collection, each at float 0.10. In the item
name that is a Minimal Wear. But Green Laminate runs 0.02 to 0.40, so
(0.10 − 0.02) ÷ 0.38 = 0.2105. The contract behaves as if you handed it skins sitting
21 percent of the way through their wear, not 10 percent. Every outcome then lands at a different raw float,
because each one has its own range:
| Outcome | Chance | Output float | What it sells for |
|---|---|---|---|
| Galil AR | CAUTION! | 33.3% | 0.1158 (Minimal Wear) | €45.05 |
| MAG-7 | Prism Terrace | 33.3% | 0.0779 (Minimal Wear) | €21.14 |
| P250 | Digital Architect | 33.3% | 0.1032 (Minimal Wear) | €26.32 |
Floats are the contract's own arithmetic and never move. Prices are ours, measured 6 September 2026, and they do.
Same contract, same average, three different floats. That is the normalisation at work, and it is why a calculator that just averages the raw numbers will quote you a cleaner skin than the game will hand over. Our trade up calculator normalises per input, the way the game does.
What that contract actually costs
Odds and floats are the fun half. The half that decides it is arithmetic. For the contract above, priced from
our own feed on 6 September 2026, one Minimal Wear Green Laminate costs €18.70 and ten
cost €187. Every outcome is equally likely, so the value of the result is just the
average of the three: (45.05 + 21.14 + 26.32) ÷ 3 = €30.84. You would be paying
€187 for a coin flip worth €30.84.
There is no lucky roll that rescues it. The best of the three outcomes pays €45.05, a quarter of what the inputs cost, so this contract loses money on every branch. That is worth checking before the float maths, because a contract whose best case is under its cost cannot be fixed by feeding it cleaner skins. We repriced the same ten inputs three days apart, 3 September and 6 September: the inputs moved €185 to €187 and the odds weighted result €31 to €30.84. The gap does not close.
That is not a trap we invented to make a point, it is what the first plausible looking contract in the catalogue happened to do. Inputs get costed at what it would really take to buy ten copies, walking down the order book rather than quoting the single cheapest listing, and outcomes get valued on what copies have actually sold for. Prices move, so treat the numbers here as a snapshot and rerun the contract live.
Price a contract before you commit →Five Coverts to a knife or gloves
The October 2025 contract is its own thing. Five Covert skins, all from cases that contain a gold pool, return one knife or glove item drawn from the gold pools of the cases you used. Odds work the same way, share of inputs split across the pool, so a case with a small gold pool concentrates the chance, and some cases are single family: feed only Operation Breakout Coverts and the only knife you can get is a Butterfly.
We keep a live table of every case: the pool, the odds per item, what the five cheapest Coverts cost right now and the floor to ceiling spread of the gold pool. That is the CS2 knife trade up page, and there is a page per case under it.
What changed in 2026
Souvenir skins are allowed now
Until May 2026 a souvenir skin could not go into a contract at all. It can now, and you can mix souvenir and ordinary inputs in the same contract. The catch is total: a souvenir input loses every souvenir attribute, the gold stickers, the match, the tag, the history, and the contract still returns one ordinary skin one grade up. Ten souvenirs in does not get you a souvenir out. Anything you read that says souvenirs are banned was written before that update.
The seven day hold, both ways
The item a contract produces arrives with a seven day trade hold, so you cannot craft and flip in the same session. In the other direction, a skin you received in a player to player trade is trade protected for seven days and cannot be fed into a contract during that window, while a skin bought on the Steam Community Market can go straight in. Plan the buying a week before the crafting. On the desk a protected or locked copy is marked in both the cards and the table, so you can see what is eligible before you start clicking.
How to read a contract before you commit
A trade up is a tool for turning skins you are not using into a skin you want, at a known set of odds. It is not an income. The honest checklist is short:
- Price both sides at sellable numbers. What the ten copies cost to buy, against what each outcome would fetch after the market's fee, weighted by its odds.
- Look at the worst outcome, not the best one. You will get the ordinary result most of the time. If the common outcome makes you unhappy, the rare one will not fix it.
- Check the float you will actually get. Normalise first. An outcome that lands in Field-Tested instead of Minimal Wear can be worth half as much.
- Watch the thin ones. An outcome with two listings on the whole market has a price that means very little until you try to sell into it.
PatinaAI computes the contract the way the game does, prices every leg from live market data, and tells you when a leg has no honest price rather than guessing one. Two ways in, depending on which end you are starting from. Got the ten skins and want to know what they are worth rolling? The trade up calculator takes the exact inputs and their floats. Got a skin in mind and no idea what feeds it? The reverse trade-up calculator runs it the other way: name the output and it gives you the collection, the ten feeders with live prices, the odds against the rest of the pool, and the average normalised float those inputs need for your wear. Knives and gloves included.
Related
- CS2 knife trade up: five Coverts to a knife, every case compared
- CS2 reverse trade-up calculator: the inputs for any skin
- The cheapest CS2 knives, priced by wear
- CS2 float and wear, explained
- Is a CS2 trade fair? How to check
- Where to sell CS2 skins for the most
- CS2 trade protection, explained: the seven day window
Lists, updated hourly: best CS2 skins to buy · skins under 10 EUR · under 25 · under 50 · most expensive · Doppler phases · Case Hardened blue gems