limen.backtest

Canonical docs

In the research projects

The Python project constructs execution inputs and policy directly. The Compiled project binds them through its Manifest. Their execution scenarios retain native ledgers, declared clocks, product rules and recorded funding.

What this package owns

Owns Limen's research execution, inventory/economics ledger and snapshot assumptions. Signal generation and operational account management belong to their respective callers.

Key entry points

Entry point	Use case	Notes
`backtest_snapshot`	Vectorized evaluation across rounds	`from limen.backtest import backtest_snapshot`
`long_flat_strategy`	Default execution model, or a template for a new strategy	Returns an `ExecutionResult`; `from limen.backtest import long_flat_strategy`
`short_flat_strategy`	Explicit binary short/flat snapshot approximation	Import from `limen.backtest.short_flat_strategy`; contract
`limen.backtest.funding`	Validated funding observations and event cash flows	Uses the declared funding clocks and product
`limen.backtest.observed_costs`	Recorded retrospective cost calibration and stress comparisons	Python Research; requires the pinned calibration and benchmark
`trade_execution`	Signed exposure with actual fill costs and persistent episodes	Accepts `TradeInputs` and `TradePolicy` with a `ProductSpec`

The package root exports backtest_snapshot, long_flat_strategy, ExecutionResult, trade_execution, TradeInputs, TradePolicy and ProductSpec.

Adjacent modules

limen.log uses backtest_snapshot() to summarize experiment permutations.
limen.experiment and limen.sfd sit upstream by producing the predictions that backtests consume.

Quick orientation

Module or family	Responsibility
`backtest_snapshot.py`	Vectorized snapshot evaluator (price validation + ledger)
`long_flat_strategy.py`	Default long-only execution model
`short_flat_strategy.py`	Explicit constant-notional short/flat snapshot model
`trade_contract.py` / `trade_execution.py`	Product, policy, event fills and inventory ledger
`funding.py` / `funding_presets.py`	Funding records and declared venue conventions
`observed_costs.py`	Cost calibration, dispositions and stress reports

text · README.md

1backtest/2├── README.md3├── __init__.py4├── _cost_comparison.py5├── _cost_history.py6├── _cost_identity.py7├── _cost_statistics.py8├── _event_clock.py9├── _event_plan.py10├── _exit_evidence.py11├── _explicit_exit_strategy.py12├── _long_flat_tp_sl.py13├── _rule_exit_state.py14├── _snapshot_execution.py15├── _snapshot_ledger.py16├── _snapshot_signals.py17├── backtest_snapshot.py18├── execution_events.py19├── funding_adapter.py20├── funding_presets.py21├── long_flat_strategy.py22├── observed_costs.py23├── short_flat_strategy.py24├── trade_contract.py25└── trade_execution.py

Merged source PR · 62d79072Package layout · Limen 6.0.1 · current package validation · package source

Things to know

Configure tunable fixed TP/SL on the manifest; Backtest defines exits, OHLC validation, and replay requirements.
backtest_snapshot() is the standard analysis path for vectorized experiment-sweep evaluation.
Explicit class_mapping or label_index interprets multiclass/multilabel snapshot inputs; Backtest defines the contract.
Snapshot return and cost outputs are basis-point scaled (*_bps columns), matching the basis-point fee and slippage inputs.