Skip to content

profiles.md

Latest commit

 

History

History
172 lines (124 loc) · 7.2 KB

profiles.md

File metadata and controls

172 lines (124 loc) · 7.2 KB
Raw

Profiles

Profiles are pre-configured BacktestConfig settings for framework-specific execution semantics. The retained evidence in this page states which pinned framework scenarios currently match exactly and which remain release-blocking.

Available Profiles

Core Profiles

Profile Description
default Sensible defaults for general use with integer-share execution
fast Zero-cost, integer-share execution -- fastest possible execution
backtrader Match Backtrader's default behavior
vectorbt Match VectorBT's default behavior (including fractional shares)
zipline Match Zipline Reloaded's default behavior
lean Match QuantConnect LEAN's default behavior
realistic Conservative settings for production

Broker Presets

Profile Description
ibkr_us_stocks_fixed Interactive Brokers US stocks fixed commission schedule

The broker preset also supports a modular alias:

Alias Resolves To
ibkr:us:stocks:fixed ibkr_us_stocks_fixed

Strict Profiles

Strict variants tune additional knobs (cash validation, settlement, short policies) for maximum parity on large-scale comparisons:

Profile Base Additional Tuning
backtrader_strict backtrader Submission precheck, simple cash check
vectorbt_strict vectorbt Lock notional for shorts, FIFO ordering
zipline_strict zipline Skip cash validation, allow shorts

Aliases

Alias Resolves To
vectorbt_pro vectorbt
vectorbt_oss vectorbt
quantconnect lean
ibkr:us:stocks:fixed ibkr_us_stocks_fixed

Usage

from ml4t.backtest import BacktestConfig

# Load a profile
config = BacktestConfig.from_preset("backtrader")

# Or build from structured broker assumptions
config = BacktestConfig.from_assumptions(
    broker="ibkr",
    region="us",
    asset_class="stocks",
    plan="fixed",
)

# Use with run_backtest
from ml4t.backtest import run_backtest
result = run_backtest(prices, strategy, config="zipline")

# Override specific settings
config = BacktestConfig.from_preset("backtrader")
config.commission_rate = 0.002
config.initial_cash = 500_000

Profiles define behavioral defaults. Quote-aware feeds layer on top of them: you can start from a preset, then override execution_price, mark_price, and the feed's price_col / quote columns without changing the rest of the profile.

Profile Comparison

Execution

Setting default backtrader vectorbt zipline lean realistic
Execution mode next_bar next_bar same_bar next_bar same_bar next_bar
Execution price open open close open close open

Stops

Setting default backtrader vectorbt zipline lean realistic
Fill mode stop_price stop_price stop_price stop_price stop_price next_bar_open
Level basis fill_price signal_price fill_price fill_price fill_price fill_price
Trail HWM close close bar_extreme close close close
Trail timing lagged lagged intrabar lagged lagged lagged

Account

Setting default backtrader vectorbt zipline lean realistic
Short selling No Yes (margin) Yes No Yes No
Leverage No Yes (50%) No No No No
Share type integer integer fractional integer integer integer

Costs

Setting default backtrader vectorbt zipline lean realistic
Commission none 0.1% none $0.005/share $0.005/share 0.2%
Slippage none 0.1% none 10% volume 0.1% 0.2%
Stop slippage 0 0 0 0 0 0.1%
Cash buffer 0% 0% 0% 0% 0% 2%

Order Processing

Setting default backtrader vectorbt zipline lean realistic
Fill ordering exit_first fifo exit_first exit_first exit_first exit_first
Reject insuff. yes yes no yes yes yes
Partial fills no no yes yes no no
Rebalance mode incremental snapshot hybrid snapshot snapshot incremental

Parity Validation

Framework profiles can be validated at two levels:

  1. Scenario-level (16 scenarios per framework): Exact trade-by-trade matching on synthetic data covering entries, exits, stops, trailing stops, brackets, and multi-asset strategies.

  2. Large-scale: Trade-by-trade comparison on a retained real-data workload. A claim is omitted when no passing artifact is retained for that framework.

Scenario claims use the retained release-candidate matrix. "Exact" appears only when every required scenario has zero canonical gap.

Profile Pinned framework Required scenarios Evidence
vectorbt_strict VectorBT Pro 2025.12.31 16/16 exact scenario evidence
vectorbt VectorBT OSS 0.28.2 15/15 exact scenario evidence
backtrader_strict Backtrader 1.9.78.123 16/16 exact scenario evidence
zipline_strict Zipline Reloaded 3.1.1 15/15 exact scenario evidence

Large-scale claims are published only when a retained workload has zero canonical gap.

Profile Pinned framework Compared Trade gap Terminal value Evidence
vectorbt_strict VectorBT Pro 2025.12.31 (1305a1e19743) 225,844 trades 0 685179.007330 large-scale evidence

No large-scale claim is published for Backtrader, Zipline, VectorBT OSS, or LEAN without a passing retained artifact.

Performance

The stable release does not publish cross-framework speed ratios. Release performance evidence is limited to the deterministic workloads and measurement boundaries in validation/performance_baselines.json; framework parity uses correctness evidence instead.

Listing Profiles

from ml4t.backtest.profiles import list_profiles

print(list_profiles())
# ['backtrader', 'default', 'lean', 'realistic', 'vectorbt', 'zipline']

Next Steps