GeneratedGenerated from the built extension by scripts/generate_python_api_reference.py; CI fails if it is stale.

openquant._core.data

Compiled from Rust (crates/pyopenquant). Signatures and types are from the shipped .pyi stubs; descriptions are the functions’ own docstrings, as help() shows them. All Python modules.

Functions

`align_calendar`

align_calendar(
    timestamps_us: Sequence[int],
    symbols: Sequence[str],
    open: Sequence[float],
    high: Sequence[float],
    low: Sequence[float],
    close: Sequence[float],
    volume: Sequence[float],
    adj_close: Sequence[float],
    interval_seconds: int,
    return_report: bool = False,
) -> Any

Clean OHLCV columns and reindex each symbol onto a regular time grid.

Rows are first deduplicated (keeping the last duplicate), then each symbol is reindexed from its first to its last timestamp in steps of interval_seconds. Grid points without a bar get None prices and is_missing_bar = True. The grid starts at each symbol's first timestamp, so bars whose timestamp is not on that grid are dropped; the interval should divide the data's spacing. A short interval over a long span produces many rows.

Parameters

timestamps_us : list[int]: Bar timestamps in microseconds since the Unix epoch (UTC).
symbols : list[str]: Instrument identifier per row.
open : list[float]: Opening prices.
high : list[float]: High prices.
low : list[float]: Low prices.
close : list[float]: Closing prices.
volume : list[float]: Traded volumes.
adj_close : list[float]: Closes adjusted for splits and dividends.
interval_seconds : int: Grid spacing in seconds; must be positive.
return_report : bool, default False: Also return a report of what alignment removed.

Returns

tuple[list[int], list[str], list[float | None], list[float | None], list[float | None],: list[float | None], list[float | None], list[float | None], list[bool]] (timestamps_us, symbols, open, high, low, close, volume, adj_close, is_missing_bar) on the grid, sorted by symbol then timestamp. Price fields are None where the grid point had no bar. With return_report=True the result is (aligned, report), where report has keys rows_removed_by_deduplication, off_grid_bar_count and off_grid_bars (a list of (symbol, timestamp_us) for each dropped off-grid bar).

Raises

ValueError: If the column lists differ in length, interval_seconds <= 0, or a Polars operation fails.

`align_calendar_df`

align_calendar_df(
    pydf: polars.DataFrame,
    interval_seconds: int,
    return_report: bool = False,
) -> Any

Clean an OHLCV DataFrame and reindex each symbol onto a regular time grid.

DataFrame form of align_calendar. The frame needs columns symbol (str), ts_us (int64 microseconds since the Unix epoch, UTC) and open, high, low, close, volume, adj_close (float64). Duplicates are dropped (keeping the last), then each symbol is reindexed from its first to its last ts_us in steps of interval_seconds. Grid points without a bar get null prices. Bars not on the grid (which starts at each symbol's first timestamp) are dropped, so the interval should divide the data's spacing.

Parameters

pydf : polars.DataFrame: OHLCV frame with the columns listed above.
interval_seconds : int: Grid spacing in seconds; must be positive.
return_report : bool, default False: Also return a report of what alignment removed.

Returns

polars.DataFrame: The aligned frame with the input columns plus a boolean is_missing_bar column, sorted by symbol then ts_us. With return_report=True the result is (aligned, report), where report has keys rows_removed_by_deduplication, off_grid_bar_count and off_grid_bars (a list of (symbol, timestamp_us) for each dropped off-grid bar).

Raises

ValueError: If interval_seconds <= 0, a required column is missing or has the wrong dtype, a symbol or ts_us value is null, or a Polars operation fails.

`clean_ohlcv`

clean_ohlcv(
    timestamps_us: Sequence[int],
    symbols: Sequence[str],
    open: Sequence[float],
    high: Sequence[float],
    low: Sequence[float],
    close: Sequence[float],
    volume: Sequence[float],
    adj_close: Sequence[float],
    dedupe_keep_last: bool,
) -> tuple[list[int], list[str], list[float], list[float], list[float], list[float], list[float], list[float], Any]

Sort OHLCV columns by (symbol, timestamp) and drop duplicate keys.

A duplicate is a repeated (symbol, timestamp) key; one row per key is kept. Output is sorted by symbol, then timestamp. Prices are neither validated nor filled. This is data preparation ahead of bars, labels or features (no AFML chapter). The report infers the bar spacing as the most common positive spacing between consecutive bars of one symbol. For daily data a gap is a skipped weekday (weekends are not gaps; holidays are); for any other spacing it is a spacing longer than the inferred one.

Parameters

timestamps_us : list[int]: Bar timestamps in microseconds since the Unix epoch (UTC).
symbols : list[str]: Instrument identifier per row.
open : list[float]: Opening prices.
high : list[float]: High prices.
low : list[float]: Low prices.
close : list[float]: Closing prices.
volume : list[float]: Traded volumes.
adj_close : list[float]: Closes adjusted for splits and dividends.
dedupe_keep_last : bool: Keep the last occurrence of a duplicated key if True, the first if False.

Returns

tuple[list[int], list[str], list[float], list[float], list[float], list[float],: list[float], list[float], dict[str, Any]] (timestamps_us, symbols, open, high, low, close, volume, adj_close, report) for the cleaned rows. report has keys row_count, symbol_count, duplicate_key_count (always 0 after cleaning), gap_interval_count, inferred_interval_us (the inferred bar spacing, None with no positive spacing), ts_min and ts_max (UTC "%Y-%m-%d %H:%M:%S" strings, None when empty) and rows_removed_by_deduplication.

Raises

ValueError: If the column lists differ in length, or a Polars operation fails.

`clean_ohlcv_df`

clean_ohlcv_df(
    pydf: polars.DataFrame,
    dedupe_keep_last: bool,
) -> tuple[polars.DataFrame, Any]

Sort an OHLCV DataFrame by (symbol, ts_us) and drop duplicate keys.

DataFrame form of clean_ohlcv. The frame needs columns symbol (str), ts_us (int64 microseconds since the Unix epoch, UTC) and open, high, low, close, volume, adj_close (float64). A duplicate is a repeated (symbol, ts_us) key. Prices are neither validated nor filled. The report infers the bar spacing as the most common positive spacing between consecutive bars of one symbol. For daily data a gap is a skipped weekday (weekends are not gaps; holidays are); for any other spacing it is a spacing longer than the inferred one.

Parameters

pydf : polars.DataFrame: OHLCV frame with the columns listed above.
dedupe_keep_last : bool: Keep the last occurrence of a duplicated key if True, the first if False.

Returns

tuple[polars.DataFrame, dict[str, Any]]: (cleaned_frame, report). report has keys row_count, symbol_count, duplicate_key_count (always 0 after cleaning), gap_interval_count, inferred_interval_us (the inferred bar spacing, None with no positive spacing), ts_min and ts_max (UTC "%Y-%m-%d %H:%M:%S" strings, None when empty) and rows_removed_by_deduplication.

Raises

ValueError: If a required column is missing or has the wrong dtype, a symbol or ts_us value is null, or a Polars operation fails.

`quality_report`

quality_report(
    timestamps_us: Sequence[int],
    symbols: Sequence[str],
    open: Sequence[float],
    high: Sequence[float],
    low: Sequence[float],
    close: Sequence[float],
    volume: Sequence[float],
    adj_close: Sequence[float],
) -> Any

Data-quality report for OHLCV columns, without modifying them.

Rows are sorted by (symbol, timestamp) before counting. A duplicate is a repeated (symbol, timestamp) key. The report infers the bar spacing as the most common positive spacing between consecutive bars of one symbol. For daily data a gap is a skipped weekday (weekends are not gaps; holidays are); for any other spacing it is a spacing longer than the inferred one.

Parameters

timestamps_us : list[int]: Bar timestamps in microseconds since the Unix epoch (UTC).
symbols : list[str]: Instrument identifier per row.
open : list[float]: Opening prices.
high : list[float]: High prices.
low : list[float]: Low prices.
close : list[float]: Closing prices.
volume : list[float]: Traded volumes.
adj_close : list[float]: Closes adjusted for splits and dividends.

Returns

dict[str, Any]: Keys row_count, symbol_count, duplicate_key_count, gap_interval_count, inferred_interval_us (the inferred bar spacing, None with no positive spacing), ts_min and ts_max (UTC "%Y-%m-%d %H:%M:%S" strings, None when empty) and rows_removed_by_deduplication (always 0 here).

Raises

ValueError: If the column lists differ in length, or a Polars operation fails.

`quality_report_df`

quality_report_df(pydf: polars.DataFrame) -> Any

Data-quality report for an OHLCV DataFrame, without modifying it.

DataFrame form of quality_report. The frame needs columns symbol (str), ts_us (int64 microseconds since the Unix epoch, UTC) and open, high, low, close, volume, adj_close (float64). The report infers the bar spacing as the most common positive spacing between consecutive bars of one symbol. For daily data a gap is a skipped weekday (weekends are not gaps; holidays are); for any other spacing it is a spacing longer than the inferred one.

Parameters

pydf : polars.DataFrame: OHLCV frame with the columns listed above.

Returns

dict[str, Any]: Keys row_count, symbol_count, duplicate_key_count, gap_interval_count, inferred_interval_us (the inferred bar spacing, None with no positive spacing), ts_min and ts_max (UTC "%Y-%m-%d %H:%M:%S" strings, None when empty) and rows_removed_by_deduplication (always 0 here).

Raises

ValueError: If a required column is missing or has the wrong dtype, a symbol or ts_us value is null, or a Polars operation fails.