This package provides a reference implementation of FIX based trading client connecting to Phoenix Prime. It illustrates how to use the FIX protocol of Phoenix Prime for order execution and to obtain streaming market data.
The project depends on the open sourced Phoenix Prime FIX Foundation Package, which illustrates how to properly set up the different outgoing FIX messages and how to parse the incoming FIX messages. The underlying FIX engine is the open source FIX library and application framework QuickFIX.
Details on the Phoenix Prime Service API and the Phoenix Prime FIX endpoints can be found in the official Phoenix Prime documentation.
The project requires
- Phoenix Prime FIX Foundation Package phx-fix-base >= 1.0
- Python >= 3.11
- QuickFIX >= 1.15.1
- Dependencies as listed in
requirements.txt
Phoenix Prime provides order execution through the FIX protocol. It currently supports FIX version 4.4 with some small custom modifications.
On the left hand side the FIX based order execution path is depicted, which optionally can also deliver market data as FIX messages.
In order to facilitate cross exchange trading, the client can deploy a strategy multiple times in collocation to the desired trading venue.
Here the same stack as for the single exchange trading is deployed multiple times. The client then can use a UDP or TCP connection between the strategies to exchange information and data.
Market data can be obtained through FIX as well. Phoenix Prime also provides a faster and more efficient UDP multicast protocol with Simple Binary Encoding (SBE), which is illustrated on the right hand side in the above diagram.
Client libraries for C++, Java, Go and Python, implementing a fully functional UDP receiver and a SBE decoder:
The Phoenix Prime market data UDP multicast protocol closely follows the CME design. A similar market data service is also offered by Deribit. Lately, Binance also started to experiment with the SBE protocol for market data distribution.
The FIX client needs to provide a session configuration with consists of two parts: First part are the account login credentials
Username=trader
Password=secret
FixAuthenticationMethod=hmac_sha256
Account=A1
The second part are the FIX specific settings, including the FIX server host IP and socket port
SenderCompID=client
TargetCompID=phoenix-prime
SocketConnectPort=1238
SocketConnectHost=127.0.0.1
These configurations are usually passed to the FIX application in a
client specific configuration file client.cfg in the [SESSION] section.
Note that QuickFIX also requires in the [SESSION] configuration section a reference to a
FIX schema file. Phoenix Prime uses the schema provided in phx/fix_spec/FIX44.xml.
The submodule phx.strategy.random contains a fully functional FIX trading application, which
provides useful standard trading functionalities such as
- Get the security list of the exchange
- Subscribe to market data
- Get instrument meta data from the exchange
- Get initial position snapshot and wallet balances from the trading account
- Get initial working orders from the trading account
- Send requests for new order and order canceling
- Parse FIX messages and represent them with Python objects
- Keep track of the order book of selected symbols
It also implements basic OEM services such as tracking the state of orders and positions from FIX execution reports.
The strategy code uses the API classes from phx-fix-base package.
- Clone the repository
git clone [email protected]:mtxpt/phx-fix-base.git. - In
phx-fix-examplesfolder runsource <phx-fix-examples>/opt/conda/bin/activate dev pip install git+file:///<full path to phx-fix-base>- To reinstall a new version of
phx-fix-basepackage runpip uninstall phx-fix-baseand then install command from the previous step.
The strategy can be configured via the <phx-fix-examples>/src/phx/strategy/random/random_strategy.yaml file.
To run the trading application
source <phx-fix-examples>/opt/conda/bin/activate dev
cd <phx-fix-examples>/src/phx/strategy/random
./start.sh
or
python3 main.py
For those who do not know enough about FIX you can learn more here
The online FIX parser is particularly useful to understand FIX message strings quickly.
It is highly recommended to create a new Python environment. The script scripts/setup_all.sh
automates the creation of a conda based environment with all dependencies installed. Optionally
provide the argument clean to remove existing environment and rebuild all.
scripts/setup_all.sh [clean]
Note that setup_all.sh also builds a custom QuickFIX version for arm64 architecture.
Alternatively, a new Python environment can be created and the requirements.txt can
be installed directly as follows
pip3 install -r requirements.txt
Note that requirements.txt does not install QuickFIX for macOS with arm64 architecture
as the current QuickFIX version 1.15.1 has some issues and requires a patch.
This project depends on QuickFIX. Check their license agreement for licensing information.
Building QuickFIX for Apple arm64 requires a patch. The following script
automates the patch and builds QuickFIX for arm64 from source:
scripts/build_QuickFIX_arm64.sh
If you use setup_all.sh you don't have to execute this build step as it is handled by setup_all.sh as well.
The Python QuickFIX bindings also fail to install on Windows. Fortunately, for Windows there are prebuilt wheel packages.
To set up the Python environment using Conda follow these steps:
- Install Conda or Miniconda
- Create a new environment with
conda create --name phx python=3.11 - Activate the environment
- Install all dependencies first
pip install -r requirements.txt - Download the QuickFIX wheel
QuickFIX‑1.15.1‑cp39‑cp39‑win_amd64.whl - Install the wheel
pip install QuickFIX‑1.15.1‑cp39‑cp39‑win_amd64.whl - List packages and check if
QuickFIX 1.15.1shows upconda list
Note that during the execution of pip install -r requirements.txt you should first see
Ignoring QuickFIX: markers 'platform_machine != "arm64" and sys_platform != "win32"' don't match your environment
To conveniently work with PyCharm it must be configured to use the proper interpreter.
Set the Python interpreter managed by the Conda package manager in ./opt/conda/
Lower right corner in PyCharm or in Settings choose Python Interpreter. Then
Add New Interpreter->Add Local Interpreter- Choose
Conda Environmentwith conda executable<path to>/opt/conda/condabin/conda - Click the button
Load Environments, make sure the radio buttonUse existing environmentis selected - Choose
devand give it optionally another name by editing the interpreter configuration
Next navigate to Settings -> Project -> Project Structure and configure the directory src as
source folder and the tests directory as test folder.
We appreciate feedback and contributions. If you have feature requests, questions, or want to contribute code or config files, don't hesitate to use the GitHub issue tracker.
The Phoenix Prime FIX Client Examples Package is released under the BSD 3-Clause License.