NOTICE

Algorithex
=========

This project is a fork of **Jesse** (https://github.com/jesse-ai/jesse), an
open-source crypto trading framework distributed under the MIT License.

Upstream copyright
------------------

    Copyright (c) 2020 Jesse.Trade

All rights reserved by their respective owners.

The MIT License permits use, copying, modification, merging, publication,
distribution, sublicensing and sale, **provided the original copyright notice
and permission notice are retained**. Those notices are preserved verbatim in
the `LICENSE` file at the root of this repository.

We are grateful to Jesse Mir and to every contributor who has worked on the
upstream project. A substantial portion of this codebase — including the
strategy API, the indicator library, the backtest engine, the exchange
integrations and the research tooling — originates in that project and is used
under the terms of its MIT License. Upstream contributors are credited in the
Git history of https://github.com/jesse-ai/jesse.

Modifications made in this fork
-------------------------------

This fork is distributed under the name **Algorithex** and includes the
following changes relative to upstream:

* Rebranding of the Python package from ``jesse`` to ``algorithex``, including
  the import namespace, the Docker image name and the user-facing strings.
* Renaming of the helper-module import alias from ``jh`` to ``ah``
  (``import algorithex.helpers as ah``).
* Removal of the hosted-service dependencies (license tokens and account
  redirects) so the framework runs fully self-hosted.
* The native indicator kernel was renamed and rebuilt from source as
  ``algorithex_kernel``. Sources are in ``native-kernel/``; prebuilt binaries
  for Linux, Windows and macOS (Apple Silicon) are vendored under
  ``algorithex/_native/bin/``. The kernel is built for the stable ABI, so a
  single binary per platform serves CPython 3.10 and later.
* Four new research modules, developed for this fork and not present upstream:
  ``algorithex/research/walkforward`` (purged walk-forward and combinatorial
  cross-validation), ``algorithex/research/regime`` (hidden Markov regime
  detection, change-point detection and regime-aware position sizing),
  ``algorithex/portfolio`` (correlation-aware allocation and risk parity) and
  ``algorithex/research/attribution`` (trade-level and feature-level
  attribution).
* Five further modules, also developed for this fork and not present upstream:
  ``algorithex/execution/costs.py`` and ``book.py`` (slippage and queue-aware
  fill models), ``algorithex/execution/algos.py`` (TWAP, POV, VWAP and
  Almgren-Chriss optimal execution scheduling with implementation-shortfall
  decomposition), ``algorithex/audit/leakage.py`` (static and runtime
  look-ahead detection), ``algorithex/audit/robustness.py`` (adversarial data
  attacks and certified worst-case price search),
  ``algorithex/research/ensemble.py`` (strategy combination, regime-conditional
  arbitration and causal walk-forward reallocation),
  ``algorithex/research/overfitting.py`` (probability of backtest overfitting
  by CSCV, deflated Sharpe ratio and parameter-plateau analysis) and
  ``algorithex/research/certification.py`` (composes the audits into a single
  graded verdict).

The modifications are likewise distributed under the MIT License. If you
intend to redistribute this fork, retain the copyright and permission notices
above.