Backtest¶

class BacktestDataConfig¶

Bases: object

Represents the data configuration for one specific backtest run.

bar_spec¶

bar_types¶

catalog_fs_protocol¶

catalog_fs_rust_storage_option_keys¶

catalog_fs_storage_option_keys¶

catalog_path¶

client_id¶

data_type¶

end_time¶

filter_expr¶

instrument_id¶

instrument_ids¶

metadata¶

optimize_file_loading¶

start_time¶

class BacktestEngine¶

Bases: object

PyO3 wrapper around [BacktestEngine].

Exposes the backtest engine to Python as BacktestEngine. Uses unsendable because the inner engine holds Rc<RefCell<…>>.

add_actor(actor)¶

Adds an actor from a constructed Python instance.

The actor ID and logging flags are sourced from the instance’s config.

add_actor_from_config(config)¶: Adds an actor from an importable config.

add_actors(actors)¶: Adds multiple actors from constructed Python instances. Stops at the first error.

add_actors_from_configs(configs)¶: Adds multiple actors from importable configs. Stops at the first error.

add_builtin_actor(type_name, config)¶

Adds a built-in example actor from its type name and config.

This method exists only to single-source bundled example actor code across Rust and Python tests/examples. It is not a first-class extension path for adding native actors.

add_builtin_strategy(type_name, config)¶

Adds a built-in example strategy from its type name and config.

This method exists only to single-source bundled example strategy code across Rust and Python tests/examples. It is not a first-class extension path for adding native strategies.

add_data(data, client_id=None, validate=True, sort=True)¶: Adds data to the engine.

add_defi_data(data, client_id=None, sort=True)¶: Adds DeFi data to the engine.

add_exec_algorithm(exec_algorithm)¶

Adds an execution algorithm from a constructed Python instance.

The execution algorithm ID and logging flags are sourced from the instance’s config.

add_exec_algorithm_from_config(config)¶: Adds an execution algorithm from an importable config.

add_exec_algorithms(exec_algorithms)¶: Adds multiple execution algorithms from constructed Python instances. Stops at the first error.

add_exec_algorithms_from_configs(configs)¶: Adds multiple execution algorithms from importable configs. Stops at the first error.

add_instrument(instrument)¶: Adds an instrument to the engine.

add_native_exec_algorithm(type_name, config)¶

Adds a compiled-in native Rust execution algorithm from its type name and config.

The type name determines which built-in execution algorithm is constructed. All execution happens in Rust; Python is the configuration layer.

add_strategies(strategies)¶: Adds multiple strategies from constructed Python instances. Stops at the first error.

add_strategies_from_configs(configs)¶: Adds multiple strategies from importable configs. Stops at the first error.

add_strategy(strategy)¶

Adds a strategy from a constructed Python instance.

The strategy ID, order ID tag, and logging flags are sourced from the instance’s config.

add_strategy_from_config(config)¶: Adds a strategy from an importable config.

add_venue(venue, oms_type, account_type, starting_balances, base_currency=None, default_leverage=None, leverages=None, margin_model=None, fill_model=None, fee_model=None, latency_model=None, modules=None, book_type=Ellipsis, routing=False, reject_stop_orders=True, support_gtd_orders=True, support_contingent_orders=True, use_position_ids=True, use_random_ids=False, use_reduce_only=True, use_message_queue=True, use_market_order_acks=False, bar_execution=True, bar_adaptive_high_low_ordering=False, trade_execution=True, liquidity_consumption=False, queue_position=False, allow_cash_borrowing=False, frozen_account=False, oto_trigger_mode=Ellipsis, price_protection_points=None, settlement_prices=None, liquidation_enabled=False, liquidation_trigger_ratio=None, liquidation_cancel_open_orders=True)¶

Adds a simulated exchange with the given parameters to the engine.

# Liquidation parameters

liquidation_enabled (bool, default False): if margin liquidation should be triggered when the account’s equity falls to or below the maintenance margin threshold scaled by liquidation_trigger_ratio.
liquidation_trigger_ratio (float, optional, default 1.0): the ratio of maintenance margin used as the liquidation threshold. A value of 1.0 liquidates when equity <= maintenance margin; higher values trigger earlier.
liquidation_cancel_open_orders (bool, default True): if open resting orders for the venue should be cancelled before synthetic close-out fills are emitted for open positions.

backtest_end¶: Returns the last backtest range end, in nanoseconds since the UNIX epoch.

backtest_start¶: Returns the last backtest range start, in nanoseconds since the UNIX epoch.

cache¶: Returns the cache shared with the kernel and registered components.

change_fill_model(venue, fill_model)¶: Changes the fill model for a venue.

clear_actors()¶: Clears all actors from the engine.

clear_data()¶: Clears all data from the engine.

clear_exec_algorithms()¶: Clears all execution algorithms from the engine.

clear_strategies()¶: Clears all strategies from the engine.

dispose()¶: Disposes of the engine, releasing all resources.

end()¶: Ends the backtest run, finalizing results.

generate_account_report(venue=None, account_id=None)¶

Generates an account report as a pandas DataFrame.

At least one of venue or account_id must be provided.

# Errors

Returns an error if neither venue nor account_id is provided, or if the Python ReportProvider import or call fails.

generate_fills_report()¶

Generates a fills report as a pandas DataFrame.

# Errors

Returns an error if the Python ReportProvider import or call fails.

generate_order_fills_report()¶

Generates an order fills report as a pandas DataFrame.

# Errors

Returns an error if the Python ReportProvider import or call fails.

generate_orders_report()¶

Generates an orders report as a pandas DataFrame.

# Errors

Returns an error if the Python ReportProvider import or call fails.

generate_positions_report()¶

Generates a positions report as a pandas DataFrame.

# Errors

Returns an error if the Python ReportProvider import or call fails.

get_result()¶: Returns the backtest result from the last run.

instance_id¶: Returns the unique instance ID for this engine.

iteration¶: Returns the current iteration count.

list_venues()¶: Returns the list of registered venue identifiers.

machine_id¶: Returns the machine ID for this engine.

portfolio¶: Returns the portfolio shared with the kernel and registered components.

reset()¶: Resets the engine state for a new run.

run(start=None, end=None, run_config_id=None, streaming=False)¶: Runs the backtest engine.

run_config_id¶: Returns the last run config ID, if any.

run_finished¶: Returns when the last run finished, in nanoseconds since the UNIX epoch.

run_id¶: Returns the last run ID, if any.

run_started¶: Returns when the last run started, in nanoseconds since the UNIX epoch.

sort_data()¶: Sorts the engine’s internal data stream by timestamp.

trader_id¶: Returns the trader ID for this engine.

class BacktestEngineConfig¶

Bases: object

Configuration for BacktestEngine instances.

bypass_logging¶

cache¶

controller¶

data_engine¶

delay_post_stop¶

exec_engine¶

instance_id¶

load_state¶

logging¶

msgbus¶

portfolio¶

risk_engine¶

run_analysis¶

save_state¶

shutdown_on_error¶

timeout_connection¶

timeout_disconnection¶

timeout_portfolio¶

timeout_reconciliation¶

timeout_shutdown¶

trader_id¶

class BacktestNode¶

Bases: object

Orchestrates catalog-driven backtests from run configurations.

BacktestNode connects the [ParquetDataCatalog] with [BacktestEngine] to load historical data and run backtests. Supports both oneshot and streaming modes.

add_actor_from_config(run_config_id, config)¶

add_builtin_strategy(run_config_id, type_name, config)¶

Adds a built-in example strategy to the engine for the given run config.

This method exists only to single-source bundled example strategy code across Rust and Python tests/examples. It is not a first-class extension path for adding native strategies.

add_strategy_from_config(run_config_id, config)¶

build()¶

Builds backtest engines from the run configurations.

For each config, creates a BacktestEngine, adds venues, and loads instruments from the catalog. If building a config fails with BacktestRunConfig.raise_exception disabled, logs the error and skips that config; successful return does not guarantee an engine for every config.

# Errors

Returns an error if building an engine from a config fails and BacktestRunConfig.raise_exception is enabled for that config.

configs¶: Returns the run configurations.

dispose()¶: Disposes all engines and releases resources.

generate_account_report(run_config_id, venue=None, account_id=None)¶

Generates an account report for the given run config engine.

At least one of venue or account_id must be provided.

# Errors

Returns an error if no engine exists, neither selector is provided, or report generation fails.

generate_fills_report(run_config_id)¶

Generates a fills report for the given run config engine.

# Errors

Returns an error if no engine exists or report generation fails.

generate_order_fills_report(run_config_id)¶

Generates an order fills report for the given run config engine.

# Errors

Returns an error if no engine exists or report generation fails.

generate_orders_report(run_config_id)¶

Generates an orders report for the given run config engine.

# Errors

Returns an error if no engine exists or report generation fails.

generate_positions_report(run_config_id)¶

Generates a positions report for the given run config engine.

# Errors

Returns an error if no engine exists or report generation fails.

get_engine_cache(run_config_id)¶

Returns the cache for the given run config engine.

# Errors

Returns an error if no engine exists for the run config ID.

get_engine_portfolio(run_config_id)¶

Returns the portfolio for the given run config engine.

# Errors

Returns an error if no engine exists for the run config ID.

run()¶

Runs all configured backtests and returns results.

Automatically calls build() if engines have not been created yet. For each run config, loads data from the catalog and runs the engine. Supports both oneshot (chunk_size = None) and streaming modes. Configs without a built engine are skipped. If a run fails with BacktestRunConfig.raise_exception disabled, logs the error, clears its loaded data, leaves the engine undisposed, and omits its result.

# Errors

Returns an error if building, data loading, or engine execution fails and BacktestRunConfig.raise_exception is enabled for the run config.

class BacktestResult¶

Bases: object

Results from a completed backtest run.

backtest_end¶

backtest_start¶

elapsed_time_secs¶

instance_id¶

iterations¶

machine_id¶

returns_series¶

run_config_id¶

run_finished¶

run_id¶

run_started¶

stats_general¶

stats_pnls¶

stats_returns¶

summary¶

total_events¶

total_orders¶

total_positions¶

trader_id¶

class BacktestRunConfig¶

Bases: object

Represents the configuration for one specific backtest run. This includes a backtest engine with its actors and strategies, with the external inputs of venues and data.

chunk_size¶

data¶

dispose_on_completion¶

end¶

engine¶

id¶

raise_exception¶

start¶

venues¶

class BacktestVenueConfig¶

Bases: object

Represents a venue configuration for one specific backtest engine.

account_type¶

allow_cash_borrowing¶

bar_adaptive_high_low_ordering¶

bar_execution¶

base_currency¶

book_type¶

default_leverage¶

fee_model¶

fill_model¶

frozen_account¶

latency_model¶

leverages¶

liquidation_cancel_open_orders¶

liquidation_enabled¶

liquidation_trigger_ratio¶

liquidity_consumption¶

margin_model¶

modules¶

name¶

oms_type¶

oto_trigger_mode¶

price_protection_points¶

queue_position¶

reject_stop_orders¶

routing¶

settlement_prices¶

starting_balances¶

support_contingent_orders¶

support_gtd_orders¶

trade_execution¶

use_market_order_acks¶

use_position_ids¶

use_random_ids¶

use_reduce_only¶

class FXRolloverInterestModule¶

Bases: object

Simulates FX rollover (swap) interest applied at 5 PM US/Eastern daily.

When holding FX positions overnight, the interest rate differential between the two currencies is credited or debited. Wednesday and Friday rollovers are tripled (Wednesday for T+2 settlement, Friday for the weekend).

class InterestRateRecord¶

Bases: object

A single interest rate data entry.