paperswithbacktest /awesome-systematic-trading

A curated list of awesome libraries, packages, strategies, books, blogs, tutorials for systematic trading.

★ 15,377⑂ 1,868👁 193⑃ 107● 29▤ 0
PythonCreated 2022-02-05paperswithbacktest.com
+18
Daily
+224
Weekly
+724
Monthly
15,377
stars
1,868
forks
193
watchers
107
prs
29
issues
0
releases

README

Awesome Systematic Trading



希望阅读中文版?点我
日本語版はこちら

We are collecting a list of resources papers, softwares, books, articles for finding, developing, and running systematic trading (quantitative trading) strategies.


 Run your first paper with Papers With Backtest

Start with a published strategy, its Python code and its data. Clone it into your
workspace and run a backtest in your browser.

Try the value-and-size example →

No card or local setup required. A free account includes one full strategy
unlock, one clone and $1 of research-agent credit, once per account. The strategy
you unlock stays readable; the allowance does not reset.

Choose one starting point for your free unlock:

| Research question | Open the example |
|---|---|
| How do value and size affect equity returns? | Value and Size Effect |
| Can correlation improve commodity momentum? | Commodity Momentum |
| How do trend and mean reversion behave in Bitcoin? | Bitcoin Seasonality, Trend and Mean Reversion |


Walk through your first backtest

1. Open an example, create a free account and unlock that strategy's code and results.
2. Clone it into your workspace. Its backtest starts in the strategy chat.
3. Inspect the result and the code. Use the available agent credit to ask which
   trading costs the backtest includes and how they affect the result.

Further research uses the remaining agent credit or a top-up.
Backtester includes 100 strategy unlocks per month, market datasets, the course,
$10/month of browser-agent credit and catalog access from your own Claude, Codex
or Cursor over MCP.
Compare free and paid access ·
Connect your own assistant




 What will you find here?

- 136 libraries and packages for research and live trading, with dead and dormant projects flagged
- Strategies from published papers, with the Sharpe ratio each one produced when it was coded and run
- 55 books for beginners and professionals
- 22 videos and interviews
- And also some blogs and courses


 What the replication record looks like

We have coded and run 4,843 of these papers over their own full history. Some numbers worth
knowing before you pick one to implement:

- The median replication returns a Sharpe ratio of 0.37, and 48% clear a t-statistic of 1.96.
  Half the published record cannot be distinguished from zero on its own sample.
- Median test window: 34 years. A strategy needs roughly (1.96 / Sharpe)² years to prove
  itself, so a Sharpe of 0.4 needs about 24 of them.
- The median strategy carries a beta of +0.17 to the S&P 500. Removing it takes the median
  information ratio down to 0.21, so a meaningful slice of the published edge is index
  exposure rather than skill.
- Across 2,838 papers with a record on both sides of their publication date, we could find no
  measurable decay after publication once the market period is controlled for, to within a fifth
  of a percentage point a year.

Method and caveats are written up on the wiki.




Click here to see the full table of content

Top issues

Discussions

Snapshot: 2026-11-16T09:23:14Z · Data license Apache-2.0 · JSON