Jane Street Electronic Trading Competition (Algorithmic Strategies)

This repository contains a collection of algorithmic trading strategies and a client to interact with a Jane Street–style electronic exchange. Strategies include pennying/market-making, a micro MACD signal, ETF basket arbitrage, ADR conversion arbitrage, and a simple bond market maker.

Key Features

• Pennying market-making around the NBBO
• Micro-MACD momentum signal per symbol (12/20 EMA spread)
• ETF basket arbitrage scaffolding for XLF vs components (GS, MS, WFC, BOND)
• ADR conversion logic (VALBZ/VALE) with convert orders
• Bond market-making around a fair value
• Simple client harness and test harnesses

Repository Structure

├─ client.py           # Main live client (prod by default)
├─ test.py             # Minimal client focusing on micro MACD (test exchange)
├─ test_etf.py         # ETF conversion experiment client (test exchange)
├─ bondtrade.py        # Bond market-making logic
├─ pennying.py         # Pennying strategy for most symbols
├─ micromacd.py        # Micro MACD helper (12/20 EMA logic)
├─ adrconversion.py    # ADR arbitrage logic (VALBZ/VALE convert)
├─ etf-convert.py      # ETF basket conversion utils/experiments
├─ oldbond.py          # Legacy bond logic
├─ client_test.py      # Helper utilities for testing clients
├─ RNN-DJIA/           # Experimental DJIA RNN notebooks and data (unrelated to client)
├─ LICENSE
└─ README.md

Requirements

• Python 2.7+ or 3.x (code uses print_function; sockets/JSON only)
• numpy

pip install numpy

Configuration

Edit the CONFIGURATION section inside the desired client script.

• team_name: your exchange team identifier
• test_mode: set True for test exchange, False for prod
• testexchangeindex: choose 0/1/2 for different test environments

# client.py (prod by default)
test_mode = False
test.py (test by default)
test_mode = True

Running

Make the client executable and run in a loop (recommended by the competition scaffold):

chmod +x client.py
while true; do ./client.py; sleep 1; done

Or simply run with Python:

python client.py

For the test clients:

python test.py
python test_etf.py

Strategy Modules

• pennying.py: Places quotes slightly inside the inside market, with risk checks based on current outstanding orders and holdings.
• micromacd.py (and inline function in clients): Maintains rolling 12-EMA and 20-EMA on mid-prices; issues small buy/sell signals on crossovers.
• bondtrade.py: Quotes around the bond’s fair value and manages inventory; can cancel, add, and track best buy.
• adrconversion.py: Arbitrage between VALBZ and VALE via convert orders; sequences buy/convert/sell when spreads appear.
• etf-convert.py: Experiments/utilities for XLF basket arbitrage versus components (GS/MS/WFC/BOND). Marked experimental in test_etf.py.

Notes

• Network protocol: JSON over TCP. The client sends hello, add, cancel, convert and handles book, fill, ack, reject, error.
• Rate control: basic TIMEOUT sleeps; adjust for your environment.
• Risk: Holdings and outstanding orders are tracked per symbol to limit exposure.

Testing/Experiments

• Use test.py to observe micro-MACD behavior on the test exchange (prints order/ack/fill counts).
• Use test_etf.py to explore ETF conversion logic; the code is scaffolded and partially experimental.
• client_test.py includes helper routines for interacting in tests.

License

This project is licensed under the MIT License. See LICENSE for details.