> ## Documentation Index
> Fetch the complete documentation index at: https://kb.talostrading.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Changelog

## 2026.10.23.19

* Added prediction market (`event_contract`) support:
  * `reference_data_events()` for the new `/reference-data/events` endpoint, with `events`, `exchange`, `event_tickers`, and `include_expired` filter parameters.
  * `include_expired` parameter on `reference_data_markets()` and `catalog_market_trades_v2()`, and prediction-market response fields on `reference_data_markets()` (`event`, `event_ticker`, `title`, `result`, `settlement_value`, `floor_strike`, `cap_strike`, `custom_strike`, `price_ranges`, and others).
  * `type` parameter on `get_market_trades()`, `get_market_candles()`, and `get_market_open_interest()`; pass `type="event_contract"` to select prediction markets, which are excluded by default.
* Free-form JSON response fields (such as `tags` on `reference_data_events()` and `custom_strike` on `reference_data_markets()`) are returned by `.to_dataframe()` and Parquet export as JSON-encoded string columns. `.to_list()` still returns them as parsed objects.

#### Changed

* `contract_count` from `get_market_open_interest()` is now typed as a decimal instead of an integer (upstream schema change), because prediction-market open interest can be fractional. `.to_dataframe()` returns it as `Float64` (or a string with `decimal_as_string=True`) instead of `Int64`, and Parquet export stores it as `decimal256`.

#### Fixed

* Fixed `format="json_stream"` responses failing with `JSONDecodeError` when a string value contained a Unicode line or paragraph separator (U+2028/U+2029).

## 2026.9.2.14

### Changed

* Added `reference_data_asset_groups()`.
* Added `lei` and `sedol` return fields for reference\_data\_markets.to\_dataframe()

## 2026.8.13.18

### Changed

* Fix get\_index\_candles default format behavior leading to an error.

## 2026.8.11.20

### Changed

* Added `get_protocol_metrics()`.
* Added denomination as a return field for blockchain-v2/accounts.

## 2026.7.15.21

### Changed

* MD-5897 Add asset\_class, source\_type, and identifier (CUSIP/FIGI/ISIN) filters to markets catalog

## 2026.4.30.17

### Fixed

* `get_market_candles()` no longer raises overflow errors for high-precision decimal columns; the client now materializes those columns as `decimal256`, with a fallback for older Polars versions that lack `decimal256` support.

### Changed

* The upstream documentation is now rendered with Sphinx (replacing MkDocs). No change to the published GitBook space served from this repo.

## 2026.4.16.18

### Added

* `to_dataframe(dataframe_type="polars")` and `to_lazyframe()` now run in parallel on `ParallelDataCollection`, giving end-to-end speedups of roughly 2-5x on representative endpoints. Internally the parallel path fans out via `ThreadPoolExecutor` and concatenates Arrow tables with `pl.concat(how="diagonal_relaxed")`.

### Fixed

* `to_dataframe()` no longer raises overflow errors when casting numeric columns to `decimal`.

### Changed

* PyArrow was upgraded to pick up upstream bug fixes. The Nix build target for Python 3.10 is dropped as a side effect; Python 3.10 itself remains supported via the standard install path until its EOL in October 2026.

## 2026.4.9.19

### Changed

* `.to_list()` is now \~1.5-1.9x faster on large result sets because `format="json_stream"` is the default (single streaming HTTP connection instead of paginated round trips). When a `limit` is set the client falls back to `format="json"` so behavior is unchanged for paginated callers.

## 2026.4.6.18

### Added

* Several `CoinMetricsClient` methods that were missing a `format` parameter now expose one, so callers can opt into `json_stream` / `csv` consistently across the API surface.

## 2026.4.2.14

### Fixed

* `to_dataframe(dataframe_type="polars")` on an empty result now returns an empty Polars DataFrame instead of an empty pandas one.
* `client.reference_data_markets(include="talos").to_dataframe()` no longer raises `ParserError`.
* Polars exports no longer suffer from dtype mismatches on nested and decimal columns.

### Changed

* `to_dataframe()`, `to_lazyframe()`, and `export_to_parquet()` are faster on large payloads and free of CSV-parsing edge cases. They now build outputs via PyArrow `RecordBatch.from_pylist()` driven by the schema from `get_schema()` instead of round-tripping through CSV.

### Added

* `decimal_as_string` parameter on `to_dataframe()` for callers that need full decimal precision instead of the lossy `float64` conversion.

### Deprecated

* `optimize_dtypes`, `header`, and `columns_to_store` parameters on `to_dataframe()` now emit a `DeprecationWarning`. Calls keep working.

### Removed

* Unused `TransactionTrackerData`, `CoinMetricsAPIModel`, and `AssetChainsDataCollection` helpers, plus the dead `SchemaWrapper` methods (`get_field_types`, `get_pandas_dtypes`, `get_datetime_columns`, `has_nested_fields`). These were not part of the documented client surface.

## 2026.3.31.13

### Added

* `include` query parameter on `reference_data_assets()`, `reference_data_exchanges()`, and `reference_data_markets()`.

## 2026.2.19.20

### Changed

* Internal release-pipeline cleanup. No user-facing API changes.

## 2026.2.19.19

### Fixed

* `to_dataframe()` and `export_to_parquet()` now produce correct dtypes for nested and decimal columns, end-to-end. The client now drives type inference from the precise schema generated off `openapi.yaml` (`coinmetrics/_schema_gen.py`, `_schema_base.py`, `_schema.py`) instead of inspecting columns heuristically.

### Added

* Several previously-missing endpoints and a routine schema regeneration.

## 2026.2.9.18

### Added

* `get_stream_market_contract_prices()` for the `/timeseries-stream/market-contract-prices` WebSocket endpoint.

## 2026.1.14.15

### Fixed

* Several parallelization-related parameters are now passed through correctly to the underlying data-collection layer; they were previously dropped before reaching `_data_collection.py`.

## 2026.1.5.19

### Added

* Python 3.13 support.

## 2025.12.16.20

### Added

* Exchange-pair endpoints: `reference_data_exchange_pair_metrics`, `catalog_exchange_pair_metrics_v2`, `catalog_full_exchange_pair_metrics_v2`, and `get_exchange_pair_metrics`.

## 2025.12.12.16

### Added

* `start_time` / `end_time` (and related) filters on the `catalog-v2/market-*` endpoints so callers can scope catalog queries to a time window.

## 2025.10.21.15

### Added

* Callers can now branch on the specific failure mode of a request: explicit exception classes for `401`, `403`, `414`, and `429` responses are now raised (`CoinMetricsClientUnauthorizedError`, `CoinMetricsClientForbiddenError`, `CoinMetricsClientBadParameterError`, `CoinMetricsClientRateLimitError`).

### Fixed

* Transient `ChunkedEncodingError` and `ConnectionError` failures are now retried more reliably.

### Changed

* The `examples/` tree was simplified: obsolete Flat Files and legacy patterns were removed and the troubleshooting docs were expanded. `docs.coinmetrics.io` is now the primary reference for example workflows.

### Removed

* The Data Exporter (`coinmetrics/data_exporter.py`) and its `typer_cli` entry point. Use the standard client methods plus `to_dataframe()` / `export_to_parquet()` instead.

## 2025.9.30.16

### Added

* `blockchain-metadata/locations` and `blockchain-metadata/owners` endpoints. The `owner_names` query parameter is now exposed on `blockchain-metadata/tagged-entities`, and tagging methods that did not follow the standard naming convention are now also reachable under canonical aliases.

### Fixed

* Taxonomy endpoints no longer drop nullable columns when the result is materialized via `to_dataframe()`.

## 2025.9.17.17

### Changed

* JSON-stream processing is now faster on heavier endpoints. Lighter endpoints continue to use `format="json"` so small queries stay fast.

## 2025.9.9.13

### Added

* `api_path_prefix` argument on `CoinMetricsClient` so callers can point the client at non-default API hosts/paths.

### Documentation

* Docstrings for `CoinMetricsClient` and `DataCollection`.

## 2025.9.3.18

### Added

* `ignore_unsupported_errors` and `ignore_forbidden_errors` parameters on `get_market_metrics()`, matching the existing pattern on other `get_market_*` methods.

## 2025.9.2.14

### Fixed

* Fixed the error where `_schema_constants.py` module is not in the Python Client package.

## 2025.8.28.15

### Fixed

* Fixed a bug where nullable columns for `blockchain-v2` endpoints and `timeseries/market-trades` were not returned when calling `.to_dataframe()`.

### Added

* Added `.to_dataframe()` for `get_full_block_v2()`, `get_full_transaction_v2()`, and `get_full_transaction_for_block_v2()`.
* Added the API schema as part of the package build.

## 2025.8.15.5

### Fixed

* Fixed a bug where nullable columns from `timeseries/*-metrics endpoints` are sometimes not returned when calling `.to_dataframe()`. This happened when the first row's columns are not the same as any of the subsequent rows.

## 2025.8.8.16

### Added

* Parallelization support to `get_predicted_funding_rates()`.

### Changed

* Changed the default to format=json\_stream for the following functions:
  * All of `reference_data`.
  * All of `catalog_*_v2`.
  * `get_market_orderbooks`, `get_market_quotes`, `get_market_open_interest`, `get_market_trades`, `get_market_candles`, `get_asset_metrics`

## 2025.5.6.13

### Fixed

* Timezone normalization when using the parallel option with datetime.timedelta.

## 2025.4.24.14

### Fixed

* Install CHANGELOG to correct location for documentation site.
* Fix broken image link in Best Practices user guide.

## 2025.4.15.13

### Added

* Allow `optimize_pandas_types` argument as an alias for `optimize_dtypes` in `DataCollection.to_dataframe()` calls but issue a deprecated warning. Prevents exception for code using the old name.

## 2025.3.12.17

### Changed

* Made Pandas and Polars mandatory packages.

## 2025.3.3.16

### Added

* `get_network_profiles`, missing `format` param to `get_market_orderbooks`, `ignore_*_errors` to `get_stream_asset_metrics`,

## 2025.2.26.22

### Added

* Enhanced documentation for API flows

## 2025.2.12.22

### Fixed

* Polars dependency issues.

## 2025.2.11.16

**Note: This release may be unstable if you do not have polars installed. Please update to 2025.2.12.22 for a patched version.**

### Added

* Added Polars dataframes and LazyFrames. They can be accessed using `DataCollection.to_dataframe(dataframe_type='polars')` and `DataCollection.to_lazyframe()` respectively.

### Changed

* `DataCollection` attribute `optimize_pandas_types` changed to `optimize_dtypes`.

## 2025.2.4.18

### Fixed

* Type annotations in docstrings

## 2024.12.23.19

### Added

* [CHANGELOG.md](http://changelog.md)

## 2024.12.20.17

### Added

* pd.DateOffset as a valid data type for `time_increment` in `parallel()`

### Changed

* Update pandas dependency to >= 2.0 and websocket-client >= 1.6.0

## 2024.12.16.21

### Added

* Allowed pandas Timestamp data type to be passed in `client` API calls
* Added "deprecated" flag to catalog v1 endpoints

## 2024.12.11.19

### Added

* Catalog-v2/blockchain endpoints

## 2024.12.10.20

### Changed

* Removed unused columns for `reference_data_*().to_dataframe()` return
* Improved casting for return data types

## 2024.11.21.20

### Changed

* Upgraded the typer dependency to >= 0.6.1

## 2024.11.18.19

### Changed

* Set `format=json_stream` for `catalog` and `reference_data` functions by default, drastically improving speed

## 2024.10.31.17

### Fixed

* Bug where requests have double '/' in URL

## 2024.10.15.19

### Changed

* Updated the request header to denote User-Agent as a Python API Client user

## 2024.10.9.20

### Added

* `txids` as a valid `parallelize_on` variable

## 2024.10.4.15

### Added

* Transformation logic for `catalog_*_v2().to_dataframe()` that flattens these dataframes

### Fixed

* Type casting for `coin_metrics_id` field from pandas dataframes for very large integers

## 2024.9.18.17

### Added

* Auto-retry logic for Websockets

### Removed

* Redundant tests

## 2024.9.18.16

### Fixed

* Bug on parallelization where `end_time` uses user's timezone instead of UTC

### Added

* `base_native` and `quote_native` fields to `reference_data_markets()`

## 2024.8.20.13

### Changed

* Removed JSON parsing on `on_error` for `CmStream` default

## 2024.8.16.10

### Removed

* Atlas V1 (`get_blockchain_()`) endpoints

### Added

* Warnings for upcoming `catalog` deprecation

## 2024.8.14.17

### Added

* `txid`, `accounts`, `block_hashes`, `heights`, and `sub_accounts` as valid `parallel` variables
* `height_increment` as a valid `parallel` parameter

## 2024.8.5.13

### Added

* Temporary patch for returning all columns for `to_dataframe()` call for `reference_data_*` and `security_master_*` functions

## 2024.7.12.14

### Added

* Allowed timezone aware datetimes to be passed to client API calls

## 2024.2.6.16

### Added

* Functions `get_market_funding_rates_predicted`, `catalog_{full}_market_funding_rates_predicted_v2`
* Generic examples for Python API Client functions

## 2024.1.17.17

### Fixed

* Bug where `blockchain_metadata` functions were not being called properly

## 2023.11.27.17

### Added

* `blockchain_metadata_tags()` and `blockchain_metadata_tagged_entities` functions

### Changed

* Updated README to shorten example code and remove catalog v1

## 2023.11.13.14

### Added

* Functions `get_snapshots_of_asset_metric_constituents` and `get_timeframes_of_asset_metric_constituents`

## 2023.10.30.13

### Added

* Function `get_stream_market_open_interest`

## 2023.10.19.17

### Added

* Function `get_stream_market_liquidations`

## 2023.9.29.14

### Added

* `metrics` parameter to `catalog_{full}_markets_v2`
* `catalog_index_levels_v2` and `reference_data_markets`

## 2023.9.22.21

### Added

* Parallelization support for `blockchain` endpoints

## 2023.9.11.14

### Changed

* Replaced `frequency` parameter with `granularity` for `get_market_quotes` and `get_market_orderbooks`

### Added

* Functions `reference_data_assets`, `reference_data_indexes`, `reference_data_pairs`

## 2023.8.30.20

### Added

* Functions `security_master_assets`, `security_master_markets`

## 2023.8.28.16

### Added

* Functions `catalog_{full}_pair_candles_v2`, `catalog_{full}_index_candles_v2`, `catalog_{full}_asset_chains_v2`, `catalog_{full}_mempool_feerates_v2`, `catalog_{full}_mining_pool_tips_summaries_v2`, `catalog_{full}_transaction_tracker_assets_v2`

## 2023.8.25.15

### Added

* Ability to parallelize API request for significantly improved data pull speed

## 2023.8.24.13

### Added

* Functions `reference_data_asset_metrics`, `reference_data_institution_metrics`

### Fixed

* Added `frequency` parameter to `get_market_orderbooks` (fixed in 2023.9.11.14)

## 2023.8.22.14

### Added

* Functions `catalog_{full}_asset_metrics_v2`, `catalog_exchange_{full}_metrics_v2`, `catalog_{full}_exchange_asset_metrics_v2`, `catalog_{full}_pair_metrics_v2`, `catalog_{full}_institution_metrics_v2`

## 2023.8.10.19

### Added

* `on_close` handler to `CmStream`

## 2023.7.11.17

### Added

* Functions `catalog_{full}_contract_prices_v2`, `catalog_{full}_market_trades_v2`, `catalog_{full}_market_candles_v2`, `catalog_{full}_market_orderbooks_v2`, `catalog_{full}_market_quotes_v2`, `catalog_{full}_market_funding_rates_v2`, `catalog_{full}_market_contract_prices_v2`, `catalog_{full}_market_implied_volatility_v2`, `catalog_{full}_market_greeks_v2`, `catalog_{full}_market_open_interest`, `catalog_{full}_market_liquidations_v2`, `catalog_{full}_market_metrics_v2`.

## 2023.6.8.20

### Fixed

* Market metrics catalog implementation to prevent duplicate rows
* Added test to verify one row per frequency

## 2023.5.26.17

### Added

* Transaction tracker parameters
* Include heartbeats functionality

## 2023.5.17.19

### Changed

* Fixed catalog performance issues

### Added

* Walkthrough notebook for DS UA Workshop

## 2023.5.2.20

### Added

* Rate limiter for community users
* Multithreading to CI pipeline
* Sample script for exporting atlas balance updates

## 2023.4.26.13

### Fixed

* URL fixes

## 2023.4.24.14

### Added

* Missing catalog endpoints
* Support for optional columns in API data
* New endpoint and tests

## 2023.3.16.17

### Changed

* Updated Dockerfile and dependencies
* Updated poetry lock file
* Bug fixes in examples

### Added

* Missing timeseries stream endpoints

## 2023.2.27.22

### Added

* Missing functions and parameters

## 2023.2.23.0

### Added

* Missing functions parameters
* Improved documentation for DataCollections usage

## 2023.1.26.23

### Added

* Unauthorized error handling to FlatFilesExporter
* Updated CI pipeline

## 2023.1.10.21

### Added

* Debug mode to help figure out performance issues
* Documentation updates for to\_dataframe() method for catalog
* Automated testing code coverage
* Error handling for large requests (URI too long)
* Support for secondary\_level parameter in to\_dataframe()
* Support for index levels via WS in the client

## 2022.11.14.16

### Changed

* Modified API Client to use python requests.Session for improved performance

## 2022.11.3.18

### Added

* Taxonomy endpoints
* Updated Atlas V2 balance endpoints

### Fixed

* Fixed broken examples using `type` parameter

## 2022.10.18.18

### Changed

* Updated build pipeline

## 2022.10.14.20

### Added

* New catalog endpoints for metrics
* Automated version updates on release
* Documentation generation
