Historical Replay Dataset

The evidence behind the CR141 verdict: 28 executed Compound governance proposals, 141 collateral factor changes, 89,645 analyzed accounts and 474 verified account state transitions.

What the dataset is

For each governance proposal that changed one or more collateral factors, the dataset records the protocol state at the block before execution and the recomputation of every affected account's collateral buffer under the executed change. It is a replay of history, not a simulation of the future.

What a record contains

• Proposal identifier, execution block and the evaluated block (b − 1).
• Per market: cToken address, symbol, collateral factor before and after in wei and decimal.
• Population definition and size: accounts that entered a changed market and carried debt.
• Targets evaluated, completion rate and RPC failure count for the measurement.
• Predicted and observed state transitions, and whether the two sets are identical.
• Per account and per market: the contribution to the collateral buffer, and the driving market.

Reproducibility

The verdict is reproducible from committed artifacts without network access: the replay outputs, the dashboard data and the evidence manifest are all in the repository, and the audit-closure evaluator recomputes the gate results from them. Raw per-stage dumps are large and mechanically regenerable; they are referenced by hash in the evidence manifest rather than stored inline.

What the dataset does not contain

• No price history, no liquidation events, no post-execution behaviour beyond the answer key.
• No accounts outside the population definition — entered a changed market and carried debt.
• No forecasts. Historical replay only.

Cite it

CR141 — Governance Impact Verification.
Historical replay of 28 Compound governance proposals, 141 collateral-factor changes,
89,645 analyzed accounts, 474 predicted and 474 observed account state transitions.
https://cr141.com/dataset