Comparer VectorBT à ML4T sur des stratégies réelles

# 18_vectorbt_engine_parity.py


```py
# ---
# jupyter:
#   jupytext:
#     cell_metadata_filter: tags,-all
#     text_representation:
#       extension: .py
#       format_name: percent
#       format_version: '1.3'
#       jupytext_version: 1.19.3
#   kernelspec:
#     display_name: Python 3
#     language: python
#     name: python3
# ---

# %% [markdown]
# # VectorBT Pro and OSS on Current Case-Study Strategies
#
# This notebook reports the VectorBT Pro and VectorBT OSS rows from the current real-strategy audit.
# Required comparisons use ETF, CME futures, USD-quoted foreign-exchange, and US equity-panel target
# streams where the pinned VectorBT edition supports the asset and accounting contract.
#
# **Learning objectives**
#
# - Compare VectorBT Pro and OSS with ML4T on supported real-data workloads
# - Distinguish fill precision from the monetary unit used for account values
# - Understand why VectorBT OSS is not used for the CME futures contract
# - Read engine-only runtime evidence without treating it as a universal ranking
#
# **Book reference**: Chapter 16, Section 16.3

# %% [markdown]
# ## Setup

# %%
"""Current VectorBT parity evidence."""

import json

import polars as pl
from IPython.display import display

from utils.paths import get_chapter_dir

# %% tags=["parameters"]
# Production defaults - Papermill injects overrides after this cell
ROUND_SECONDS = 3

# %%
AUDIT_PATH = get_chapter_dir(16) / "resources" / "framework_parity_audit.json"
audit = json.loads(AUDIT_PATH.read_text(encoding="utf-8"))
FRAMEWORKS = ["vectorbt_pro", "vectorbt_oss"]
FRAMEWORK_NAMES = {
    key: f"{audit['frameworks'][key]['display_name']} {audit['frameworks'][key]['version']}"
    for key in FRAMEWORKS
}
CASE_NAMES = {
    "etfs": "ETF allocation",
    "cme_futures": "CME futures",
    "crypto_perps_funding": "Crypto perpetual funding",
    "fx_pairs": "FX allocation (USD-quoted pairs)",
    "us_equities_panel": "US equity panel",
}

# %% [markdown]
# ## 1. Required comparisons

# %% tags=["results"]
results = (
    pl.DataFrame(audit["real_strategy_records"])
    .filter(pl.col("framework").is_in(FRAMEWORKS))
    .with_columns(
        pl.col("case_study").replace_strict(CASE_NAMES).alias("strategy"),
        pl.col("framework").replace_strict(FRAMEWORK_NAMES).alias("engine"),
    )
    .select(
        "strategy",
        "engine",
        "status",
        "fills",
        "valuations",
        "valuation_timestamps_match",
        "equity_gap",
        "equity_raw_gap",
        "terminal_gap",
        "terminal_raw_gap",
    )
    .sort("strategy", "engine")
)

assert results.height == 7
assert results.filter(pl.col("status") == "pass").height == 7
assert results["valuation_timestamps_match"].all()

display(results)

# %% [markdown]
# Both VectorBT editions participate in the ETF, USD-quoted foreign-exchange, and US equity-panel
# comparisons. VectorBT Pro also supplies contract multipliers and futures-style leverage for the
# CME comparison. Fill prices retain eight-decimal precision, quantities retain five-decimal
# precision, and account monetary values must round to the same cent.

# %% [markdown]
# ## 2. Unsupported asset models

# %% tags=["results"]
unsupported = (
    pl.DataFrame(audit["unsupported_records"])
    .filter(pl.col("framework").is_in(FRAMEWORKS))
    .with_columns(
        pl.col("case_study").replace_strict(CASE_NAMES).alias("strategy"),
        pl.col("framework").replace_strict(FRAMEWORK_NAMES).alias("engine"),
    )
    .select("strategy", "engine", "reason")
    .sort("strategy", "engine")
)
display(unsupported)

# %% [markdown]
# VectorBT OSS does not provide the native multiplier and margin-account model required by the CME
# bundle. Neither edition is used to emulate crypto-perpetual funding and margin accounting.

# %% [markdown]
# ## 3. Engine-only timing
#
# Each correctness-passing VectorBT row has an engine-only timing record.

# %% tags=["results"]
timing = (
    pl.DataFrame(audit["performance_records"])
    .filter(pl.col("framework").is_in(FRAMEWORKS))
    .with_columns(
        pl.col("case_study").replace_strict(CASE_NAMES).alias("strategy"),
        pl.col("framework").replace_strict(FRAMEWORK_NAMES).alias("engine"),
        pl.col("framework_median_seconds").round(ROUND_SECONDS).alias("vectorbt_seconds"),
        pl.col("ml4t_median_seconds").round(ROUND_SECONDS).alias("ml4t_seconds"),
        pl.col("framework_to_ml4t_ratio").round(2).alias("vectorbt_div_ml4t"),
    )
    .select("strategy", "engine", "vectorbt_seconds", "ml4t_seconds", "vectorbt_div_ml4t")
)

assert timing.height == 7
display(timing)

# %% [markdown]
# The measured region excludes data loading, target construction, adapter preparation, and output
# extraction. The observed ratios do not establish the same relationship for other datasets,
# strategy mechanics, or machines.

# %% [markdown]
# ## 4. Synthetic stress evidence

# %% tags=["results"]
stress = (
    pl.DataFrame(audit["synthetic_stress"]["records"])
    .filter(pl.col("framework").is_in(FRAMEWORKS))
    .with_columns(pl.col("framework").replace_strict(FRAMEWORK_NAMES).alias("engine"))
    .select("engine", "intents", "fills", "trades", "terminal_value", "status")
)
display(stress)

# %% [markdown]
# VectorBT Pro and OSS both pass the generated stress comparison against their matching ML4T
# profiles. This establishes scale conformance for the fixed target-order recipe. It does not create
# support for the excluded asset models.

```