Rust Case Odds Index 2026: Public Data Audit

The 2026 public-data result

The reviewed set contains eight operators from our case-opening comparison and wider review list. Public pages made case formats identifiable for most case-led sites. Four sampled RustClash pages also exposed a price, complete probability table and displayed gem value for every outcome, allowing a same-page expected-value calculation.

Other operators remain unknown where the same input set was not captured. Version 2026-08-28.1 does not combine a price from one time, odds from another page and market values from another system.

Each sampled probability table totals 100%. We multiplied every displayed gem value by its probability and divided the sum by the displayed case price. The pages can change after the check time, and the result describes internal credited value rather than Steam Wallet or cash value.

Read each row as a dated calculation, not a permanent operator average. Crim Case has the highest implied return among these four captured rows, but that does not make it the best case for a session or establish that the table is still current after the observation time. A later version must recapture the full row before changing the number.

Download and reuse the dataset

The machine-readable dataset is available as JSON and CSV. It includes row source IDs, observation time, method, limitations, null semantics and version history.

License: CC BY 4.0. Attribution should name “Best Rust Gambling Editorial,” link to this page, identify dataset version 2026-08-28.1, and retain the limitation that transactions and account-only catalogues were not tested.

Calculation standard for a future complete row

For a case costing C, with item credited values v₁…vₙ and probabilities p₁…pₙ, expected credited value is:

EV = (p₁ × v₁) + … + (pₙ × vₙ)

Implied return is EV / C, and implied edge is 1 - (EV / C). Probabilities must total approximately 100%, allowing only a disclosed rounding tolerance. Every value must come from the operator's own balance table at the same observation time; Steam Wallet or cash-market values belong in separate columns.

The case-odds explainer provides a worked example. The skin-value guide explains why expected credited value is not guaranteed resale value.

Why the remaining rows are not estimated

Estimating a missing probability from how often an item appears in a short history would produce a noisy and potentially biased result. Copying values from promotional art risks omitting common low-value outcomes. Scraping an account-only catalogue without recording rule version, unit and time can create a precise-looking number that is not reproducible.

The index therefore uses three evidence states:

• Partial: a public source establishes the format or one input but not the full calculation.
• Unknown: the reviewed public material does not support the field.
• Complete: price, probabilities and credited values can be reproduced from one dated rule set. Four sampled RustClash rows reach this state.

Update protocol

A future version can fill or refresh a row only when the full case table is captured from an operator-controlled public source. The update must store the source URL, UTC time, case identifier, balance unit and raw probabilities before calculation. Changed rows receive a version-history note; old figures are not silently overwritten as if they were current.