CS2 Trade-Up Calculator

Compute the exact output skins, probabilities, and predicted floats for any Counter-Strike 2 trade-up contract. Supports knife/glove 5-input contracts, multi-collection mixing, StatTrak, and reverse search.

CS2 Trade-Up // Contractmode std-10 · stattrak off

Note: This calculator is an educational tool that computes Counter-Strike 2's publicly documented trade-up math. It does not facilitate gambling, skin betting, or real-money transactions. Skin prices fluctuate; values shown are estimates based on third-party marketplace data when displayed.Counter-Strike, Counter-Strike 2, CS:GO, Steam, and Valve are trademarks of Valve Corporation. This site is not affiliated with, endorsed by, or sponsored by Valve.

Input rarity

StatTrak™

All inputs must be StatTrak™; the output will also be StatTrak™.

Inputs00 / 10

Input #01

Float

Input #02

Float

Input #03

Float

Input #04

Float

Input #05

Float

Input #06

Float

Input #07

Float

Input #08

Float

Input #09

Float

Input #10

Float

Add 10 inputs to see possible outcomes, or load an example contract below.

Or load an example contract:

CS2 eDPI Calculator OpenTTD Cargo Income Calculator

Updated: May 15, 2026

CS2 trade-up calculator. Output odds, predicted floats, and the math behind every contract.

A CS2 trade-up calculator averages 10 normalised input floats (or 5 in knife/glove mode) and maps the result into every possible output skin's range. The probability of each output equals the share of inputs from that skin's collection divided by the number of next-tier skins inside that collection.

What is a CS2 trade-up contract?

A CS2 trade-up contract converts 10 skins of the same rarity into one skin one tier higher on the standard ladder: Consumer → Industrial → Mil-Spec → Restricted → Classified → Covert. All 10 inputs must share one rarity, must be all StatTrak or all non-StatTrak (the in-game contract menu blocks any mix), and Souvenir skins are not allowed. The output's collection is decided probabilistically — the share of inputs from a given collection equals the share of total output probability that lands inside that collection's next-tier pool.

The knife/glove variant was added by Valve on 23 October 2025 in the Re-Retakes update (the same patch that brought back Retakes mode). It takes 5 Covert skins from one or more case-feeder collections and outputs one Special — a knife or pair of gloves — drawn uniformly from the contains_rare pool of the case that collection feeds. StatTrak inputs produce a StatTrak knife; non-StatTrak inputs can produce either a knife or a glove pair because StatTrak gloves do not exist. That single asymmetry is why the StatTrak 5-Covert contract is strictly knife-only.

Floats matter because they determine the output's exterior band — Factory New (0.00–0.07), Minimal Wear (0.07–0.15), Field-Tested (0.15–0.38), Well-Worn (0.38–0.45), Battle-Scarred (0.45–1.00) — and a single band difference can swing the output's market value 5–50×. Before October 2024, CS2 averaged raw input floats directly. After the Retakes Update, the engine normalises each input against its own [minFloat, maxFloat] before averaging, then maps the average back into the output skin's range. The shift removed the old filler-skin meta where players padded contracts with tightly capped skins to drag the average down; under the new formula a tightly capped input at the top of its range contributes the same normalised value as a 0–1 skin at the equivalent relative position.

How the calculator computes each output

Pick the input rarity (Consumer through Classified in standard mode; Covert is locked when you toggle the Knife / Glove 5-input tab), choose StatTrak or non-StatTrak, then fill the 10 (or 5) slots. For each slot you select a collection, then a skin available at that rarity inside the collection, then the raw float you can read in your CS2 inventory. The calculator runs four steps on every input change:

1. Normalise each input's float against its own min/max. The expression is $\bar{f}_i = (f_i - f_i,\min) / (f_i,\max - f_i,\min)$. This collapses every skin onto a common 0–1 axis so a 0.15 float on the AK-47 Asiimov (range 0.05–0.70) normalises to 0.154 while the same 0.15 float on the AK-47 Redline (range 0.10–0.70) normalises to 0.083 — almost half — and the two inputs are now directly comparable on the same axis.

2. Average the normalised values. The expression is $\bar{f} = (1/n) \sum_i \bar{f}_i$, where n = 10 in standard mode and n = 5 in knife/glove mode. The average is the single number that drives every output's float in step 3.

3. Map the average back into every candidate output's range. For each possible output skin, the formula is $f_\text{out} = (f_\text{out,\max} - f_\text{out,\min}) \cdot \bar{f} + f_\text{out,\min}$. If the result exceeds the skin's max float the calculator clamps it and surfaces a 'capped' badge — this is the case for items like the M4A1-S Fade (covert, capped at 0.00–0.08) and other skins where the underlying float space is artificially narrowed.

4. Compute every candidate's probability. The expression is $P(\text{skin}) = (n_{\text{collection}} / n_\text{total}) \times (1 / |\text{outputs}_{\text{collection}}|)$. Sum the probabilities of every output row and you should land at exactly 1.0; the results panel surfaces a verification line so you can confirm there is no rounding leak.

The Show-the-Math accordion under the results table renders the same four steps with your actual numbers substituted, and the Min Float to land FN / MW / FT panel inverts steps 2 and 3 to tell you the maximum average normalised float that keeps the primary output skin inside a chosen exterior band — useful when you have specific input candidates and need to know whether their floats are tight enough to clear the target band.

The CS2 trade-up float and probability formulas

f_{\text{out}} = (f_{\text{out},\max} - f_{\text{out},\min}) \cdot \bar{f} + f_{\text{out},\min}, \quad \bar{f} = \frac{1}{n} \sum_{i=1}^{n} \frac{f_i - f_{i,\min}}{f_{i,\max} - f_{i,\min}}, \quad P(\text{skin}) = \frac{n_{\text{coll}}}{n_{\text{total}}} \cdot \frac{1}{|\text{outputs}_{\text{coll}}|}

$f_{\text{out}}$ = Predicted float of the output skin for the contract.
$f_{\text{out},\min} / f_{\text{out},\max}$ = Minimum and maximum float of the output skin, per its bundled CS2 data entry (sourced from ByMykel/CSGO-API).
$\bar{f}$ = Average normalised input float across all n inputs.
$f_i$ = Raw float of input skin i, the value displayed in your CS2 inventory.
$f_{i,\min} / f_{i,\max}$ = Minimum and maximum float of input skin i — these can be the canonical 0.00–1.00 or a per-skin cap (Asiimov 0.05–0.70, Redline 0.10–0.70, M4A1-S Fade 0.00–0.08, AUG Amber Fade 0.00–0.40, etc.).
$n$ = Number of inputs: 10 for standard trade-ups, 5 for the knife/glove contract added in October 2025.
$n_{\text{coll}} / n_{\text{total}}$ = Inputs from a given collection divided by the total inputs (10 or 5). This is the collection's share of total output probability.
$|\text{outputs}_{\text{coll}}|$ = Number of next-tier skins (or knife/glove contains_rare entries for the case the collection feeds) in that collection's output pool.

The float formula was rewritten in the October 2024 Retakes Update and remains unchanged after the October 2025 knife/glove addition. Each input's normalised float collapses skin-specific range differences onto a common 0–1 axis. A raw 0.10 on a tightly capped skin like the AK-47 Asiimov (0.05–0.70) normalises to (0.10 − 0.05) / (0.70 − 0.05) = 0.077, while the same raw 0.10 on a 0.00–1.00 skin normalises to 0.10. The average of those normalised values is then mapped back into each output skin's own range.

The probability formula is two factors multiplied together. The first factor — n_coll / n_total — is the share of inputs from a given collection, which after the October 2024 rewrite is the *only* knob you control through input selection. The second factor — 1 / outputs_in_collection — is fixed by the dataset: a collection with one next-tier skin concentrates 100% of its share into that single output, while a collection with four next-tier skins splits its share four ways. This is why the seasoned trade-up meta favours single-output collections like The Militia Collection (Restricted tier holds only M4A4 | Modern Hunter) — fewer downstream outputs concentrates probability into the one valuable skin.

Trade-up examples

Standard 10-input Mil-Spec, single collection, deterministic Restricted output

Fill all 10 Mil-Spec slots from one collection that has exactly one Restricted skin. Apply the probability formula: n_coll / n_total = 10/10 = 1.0, outputs_in_collection = 1, so P = 1.0 × 1/1 = 100% on the single Restricted row. The predicted float follows the formula in step 3: average the 10 normalised inputs, then plug the average into the Restricted skin's [minFloat, maxFloat]. If your inputs average a normalised 0.05 and the output's range is 0.06–0.80, the predicted float is (0.80 − 0.06) × 0.05 + 0.06 = 0.097 → Minimal Wear. Single-collection, single-output contracts are the cleanest demonstration of the math because the entire result reduces to the float step.

Multi-collection 7+3 split, weighted Restricted output

Seven Mil-Spec from a 1-output collection plus three Mil-Spec from a 3-output collection. The first collection's per-output probability is 7/10 × 1/1 = 70%. The second collection's per-output probability is 3/10 × 1/3 = 10% per skin (3 rows of 10% each). Total probability is 70% + 3 × 10% = 100%. Concentrating inputs in a smaller-output collection is the standard way to push probability mass onto a specific high-value Restricted skin while still using cheaper fillers from a wider collection to drag the average normalised float down.

Knife/Glove 5-Covert from one case-feeder collection

Five Coverts from one collection that feeds a single case (the calculator looks the mapping up in case-by-collection.json). n_coll / n_total = 5/5 = 1.0. The output pool is the case's contains_rare entries — typically 50 to 70 items depending on the case's age. Each Special's probability is 1.0 × 1/|pool|, so with a 50-item pool every knife/glove sits at 2.0% probability and with a 70-item pool at roughly 1.43%. Predicted float is computed against each Special's own range: most knife finishes use 0.00–1.00, but pattern-restricted knives (Doppler phase, Gamma Doppler phase) and other special items with artificially narrowed float ranges will clamp the predicted float to their max and earn a 'capped' badge in the results table.

Reverse mode — work backwards from a target skin

Pick a target Covert (say AK-47 Asiimov from The Danger Zone Collection, range 0.05–0.70) and a desired exterior. The calculator inverts steps 2 and 3 to compute the average normalised input ceiling that keeps the output inside the target band. For Field-Tested (upper bound 0.38), the ceiling is (0.38 − 0.05) / (0.70 − 0.05) ≈ 0.508 average normalised, so your 10 Classified inputs from the same collection must average below that or the contract will deliver a Well-Worn instead. Reverse mode also flags whether the target collection actually has enough lower-tier skins to feed the contract — some Operation collections only have a Classified tier with no Restricted feeders below them, in which case reverse mode returns 'infeasible' and tells you no input collection can deliver that output.

Trade-up tips

Concentrate inputs in the smallest-output collection that contains your target. The probability formula is (n_coll / n_total) × (1 / outputs_in_collection), so a collection with one next-tier skin concentrates its entire input share into that one output. The Militia Collection (1 Restricted: M4A4 | Modern Hunter) and several Operation collections are popular for this reason — fewer next-tier outputs amplifies probability per skin even when input cost is the same as a wider collection.
Open the float optimisation panel before committing inputs. For your primary target output, the panel solves backwards from a chosen exterior band (FN/MW/FT/WW/BS) and returns the maximum average normalised input float that still keeps the predicted output inside that band. If the ceiling is 0.082, every input has to push the average normalised float to 0.082 or lower — a strict cap that is invisible if you only watch the raw input floats.
Use the reverse mode when you have a target skin in mind. Pick the skin, pick the exterior, choose between 'lowest float in target exterior' (deepest in the band, best quality output) and 'just inside the tier' (looser ceiling, cheaper inputs). The calculator returns the input collection that can deposit the target plus the maximum average normalised float that keeps the result inside the band — so you know upfront whether the inputs you can afford can actually deliver the exterior you want.
Treat StatTrak gloves as impossible. StatTrak gloves do not exist in CS2, which means a StatTrak 5-Covert contract has a knife-only output pool. The calculator filters the contains_rare pool accordingly when you toggle StatTrak on the knife/glove tab — you can confirm this by seeing the output pool size shrink relative to the non-StatTrak case (DMarket and the Steam Community trade-up guide both confirm StatTrak Covert inputs from the Recoil, Revolution, Snakebite, Clutch, Glove, Hydra, Broken Fang and Dead Hand Terminal cases are blocked entirely because their only Specials are gloves).
Avoid float-capped inputs at the top of their range. After the October 2024 normalised-float rewrite, a 0.30 AUG | Amber Fade input (capped 0.00–0.40) normalises to 0.75 — about 2.5× higher than the same raw float on a 0.00–1.00 skin. Wide-range inputs with floats deep inside Factory New normalise to small numbers and pull the average down; capped inputs at the upper end of their range pull the average up. The old filler-skin meta where players padded with tight-range inputs is dead under the new formula.
Sum probabilities and confirm the total is 1.0 (100%). The calculator surfaces a check line under the output table — if it ever reports 0.98 or 1.02 instead of 1.00, you have either an input from a collection with no next-tier outputs (the contract is invalid in-game and the calculator flags it) or a rounding leak from a malformed dataset entry. The check is the cleanest sanity test that every column of the formula was applied correctly.

Frequently asked questions about CS2 trade-up contracts

Does this calculator support knife and glove trade-ups?

Yes. Switch to the Knife / Glove (5) tab; the input rarity locks to Covert and the slot count drops to five. The output pool is the contains_rare entries of the case that each input's covert collection feeds, mapped via the bundled case-by-collection table (snapshot 2026-05-15 from ByMykel/CSGO-API).

What changed in the October 2024 trade-up update?

The Retakes Update rewrote the float formula. Before October 2024, CS2 averaged raw input floats directly. After the update, every input is first normalised against its own [minFloat, maxFloat], then the normalised values are averaged, and the result is mapped back into the output skin's range. The change killed the filler-skin meta where players padded contracts with tightly capped low-float skins to drag the average down.

What did Valve change in October 2025?

The 23 October 2025 Re-Retakes Update added the 5-Covert knife/glove contract. Five Covert inputs produce one Special drawn uniformly from the feeder case's contains_rare pool. StatTrak Coverts produce a StatTrak knife only. The standard 10-input contract is unchanged.

Can I mix StatTrak and non-StatTrak inputs?

No. The CS2 contract menu blocks mixed contracts. Toggle StatTrak in the controls bar above the slots; the picker then filters to skins that have a StatTrak variant, and the output preserves StatTrak through the trade-up.

Why are Souvenir skins not in the picker?

Souvenir inputs are explicitly excluded by Valve's trade-up rules. The calculator filters them out at the picker, and a URL state that tries to hydrate a Souvenir input is rejected at validation with a clear error instead of computing a wrong result.

What is a normalised float?

It is the input's raw float mapped to a 0–1 scale against that skin's own min/max float range. Post-October-2024 CS2 averages normalised values across inputs, then maps the average back into the output's range. The Show-the-Math accordion under the results panel walks through every substitution with your actual numbers.

How does the calculator determine probabilities for multi-collection inputs?

Each output's probability is (inputs from that output's collection / total inputs) × (1 / number of next-tier outputs in that collection). The first factor is the collection's share of total probability; the second factor is how that share splits among the collection's eligible outputs. A 7+3 split between a 1-output collection and a 3-output collection produces 70%, 10%, 10%, 10% on the four rows — sum 100%.

Does the calculator price the inputs and outputs?

Not in this version. The focus is on float math and probability distribution against an audited dataset, not on live marketplace prices. Pricing overlays from Steam Market, Skinport, Buff163 and similar sources are planned for a later release once we have a stable price-feed contract.

Why does the predicted output float not match my in-game contract preview exactly?

Three reasons. First, ByMykel/CSGO-API is refreshed on a weekly cadence and can lag Valve by a few days on freshly released cases — the dataset footer at the bottom of the page shows the bundled snapshot date so you can verify how fresh the ranges are. Second, some skins have hidden float floors or pattern caps that the public dataset cannot model perfectly. Third, the calculator displays a clamped float when a candidate's predicted float exceeds its max; the in-game preview applies the same clamp but shows the underlying raw value in some menus.

Can I share a permalink to a specific contract?

Yes. Every input change updates a versioned querystring, and the Share button copies it to your clipboard. Recipients open the link and see the exact same contract pre-loaded — slots, StatTrak toggle, mode, and reverse settings.

Is this trade-up calculator free?

Yes. The calculator runs entirely in the browser, requires no account, never asks for your Steam login, and shows no marketplace upsells. The bundled dataset is open-source under MIT (ByMykel/CSGO-API), so you can verify every input range, collection mapping and case pool against the upstream JSON.

How accurate are the predicted output floats?

Accurate to four decimal places against the bundled dataset, and accurate to within roughly 0.001 against an in-game contract when the dataset is up to date. The four-step formula is deterministic — there is no randomness in the float prediction, only in which collection the output happens to come from. If your real contract result drifts by more than that, refresh the page to pick up any dataset update or report the deviation so we can patch the data.

CS2 trade-up glossary

Float

The wear value of a CS2 skin, a real number in [0, 1]. Lower floats sit in cleaner exteriors (Factory New, Minimal Wear) and generally command higher market prices.

Exterior

The display band of a float: FN (0.00–0.07), MW (0.07–0.15), FT (0.15–0.38), WW (0.38–0.45), BS (0.45–1.00). Upper bounds are exclusive except BS, which is inclusive to 1.00.

Normalised float

An input float mapped to 0–1 against its own min/max range: (float − min) / (max − min). CS2's post-October-2024 trade-up formula averages normalised inputs and maps the result back into the output skin's range.

Collection

A named group of CS2 skins released together (e.g. The Mirage 2021 Collection, The Cobblestone Collection). Trade-ups draw outputs from the contributing collections' next-tier pools, weighted by input share.

Float cap

A per-skin minimum or maximum that restricts the float range below the canonical 0–1. The AWP Asiimov caps at 0.18 minimum (no Factory New version exists), the AK-47 Redline caps at 0.10–0.70, the AK-47 Asiimov at 0.05–0.70. Float caps matter under the normalised formula because they change the relative position of a given raw float on the 0–1 axis.

Covert

The rarest standard rarity tier (red), one below Special. Covert is the input rarity locked when you switch the calculator to knife/glove mode.

Special

Knives and gloves. Not part of the standard rarity ladder. Reachable only via the 5-Covert knife/glove trade-up added by Valve in the Re-Retakes Update on 23 October 2025.

StatTrak

A weapon variant that tracks confirmed kills. Trade-up contracts cannot mix StatTrak and non-StatTrak inputs, and the StatTrak status of the output mirrors the inputs. StatTrak gloves do not exist in CS2 — the StatTrak 5-Covert contract is therefore knife-only.

contains_rare

The pool of Special items (knives and gloves) a CS2 case can drop as its yellow-tier (Special) reward. The 5-Covert knife/glove contract draws uniformly from this pool for the case that the input collection feeds.

Expected Value (EV)

The probability-weighted average market value of every possible output. Common metric in the wider CS2 trade-up community for ranking contract profitability. This calculator focuses on float and probability math; pricing-aware EV is planned for a later release.

Content verified by the Smart Calculators Team