https://github.com/milliyang/noise
A Backtest or Trading Framework with C++
https://github.com/milliyang/noise
algo-trading algorithmic-trading backtest backtesting backtesting-engine backtesting-frameworks timeseries-analysis trading trading-strategies
Last synced: 30 days ago
JSON representation
A Backtest or Trading Framework with C++
- Host: GitHub
- URL: https://github.com/milliyang/noise
- Owner: milliyang
- Created: 2022-11-17T14:55:40.000Z (about 3 years ago)
- Default Branch: main
- Last Pushed: 2025-10-23T10:28:25.000Z (4 months ago)
- Last Synced: 2025-10-23T12:27:39.647Z (4 months ago)
- Topics: algo-trading, algorithmic-trading, backtest, backtesting, backtesting-engine, backtesting-frameworks, timeseries-analysis, trading, trading-strategies
- Language: C++
- Homepage: https://github.com/milliyang/noise
- Size: 16.2 MB
- Stars: 8
- Watchers: 1
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
Noise
=====
A Backtest or Trading Framework with C++
Why invent the wheel ?
------------
* Inspired by [backtesting.py](https://kernc.github.io/backtesting.py)
* Not invent, just rewrite it in another way
* Need a simpler, faster, ease of replace component framework
Features
--------
* Simple, no-documented API / structure
* Composable Strategies / Broker / Feed / Ploting
* Support different ways of Backtest (SimpleBroker/BrokerMuxer/EventDriven/PreProcessed)
* Cheap Backtest Result
* Cheap Visualization
* Cheap TA-Lib
Requirements
------------
- OpenMP [Optional] maximize 48+ Cores to run
- Qt [Optional] for GUI ploting
- GCC
- CMAKE
- 3rd in: noise/3rd
- [fmt](https://github.com/fmtlib/fmt)
- [spdlog](https://github.com/gabime/spdlog)
- [args](https://github.com/Taywee/args)
- [inih](https://github.com/benhoyt/inih)
- [Armadillo](https://arma.sourceforge.net)
- [glob](https://github.com/p-ranav/glob)
- [hdf5](https://github.com/HDFGroup/hdf5) & [HighFive](https://github.com/BlueBrain/HighFive)
Build on Linux
--------------
```sh
cd noise
mkdir build && cd build
cmake ..
make && ./bt
```
Usage
-----
```sh
make && ./bt
```
Results in:
```text
[Stat] summary:
Start [YYYMMDD] 20180102
End [YYYMMDD] 20180213
Duration [days] 112
Exposure Time [%] 99.11
Equity Final [$] +6277.18
Equity Peak [$] +10109.52
PNL [$] -3722.82
Return [%] -37.23
Buy & Hold Return [%] -8.54
Return (Ann.) [%] 0.00 TBD
Volatility (Ann.) [%] 0.00 TBD
Sharpe Ratio 0.00 TBD
Max. Drawdown [%] 0.00 TBD
Avg. Drawdown [%] 0.00 TBD
Max. Drawdown Duration [days] 0.00 TBD
Avg. Drawdown Duration [days] 0.00 TBD
Trades 57
Win Rate [%] 35.09
Avg. Trade [%] -3.99
Best. Trade [%] 11.11
Wors. Trade [%] -17.50
Best. Trade on PNL [$] +40.85
Wors. Trade on PNL [$] -246.01
Max. Trade Duration [days] 108.00
Avg. Trade Duration [days] 51.63
SQN -5.53
[main] time elapsed: 12.594399 ms
```
Find more usage examples in dir noise/sample/
Build on Windows with GUI Ploting
--------------
```sh
cd noise
mkdir build && cd build
cmake ../sample
MSBuild.exe noise.sln # require VisualStudio
.\Debug\cd.exe
```
Contibutes
----------
- Always welcomed ( Feature or Bugfix or Unitest )
- Platform Independent
- Simple is the most important thing. No heavy / fancy code. Not much 3rd library I Hope...
- Keep less code
Final
-----
* No Lisence, or [UnLisence](https://unlicense.org/)
* Thanks to [Github](https://github.com/), [kernc](https://github.com/kernc)
* Again, Inspired by [backtesting.py](https://kernc.github.io/backtesting.py)
Additional Remarks
------------------
* I did try not to include too much third-party library. But as project went on, they grows ...
* Some are just tiny cpp header-only:
- [fmt](https://github.com/fmtlib/fmt)
- [spdlog](https://github.com/gabime/spdlog)
- [args](https://github.com/Taywee/args)
- [inih](https://github.com/benhoyt/inih)
* Some made it easier for me to do unitest, math calculation
- [Armadillo](https://arma.sourceforge.net)
- [Eigen](https://gitlab.com/libeigen/eigen)
- [glob](https://github.com/p-ranav/glob) & [filesystem](https://github.com/gulrak/filesystem)
* Some help to speed up backtesting for large dataset:
- [hdf5](https://github.com/HDFGroup/hdf5) & [HighFive](https://github.com/BlueBrain/HighFive)
* They may have diff Lisences ...