|
33 | 33 |
|
34 | 34 | <br> |
35 | 35 |
|
36 | | -[](https://github.com/Y-Research-SBU/QuantAgent/stargazers) |
37 | | -[](https://github.com/Y-Research-SBU/QuantAgent/forks) |
38 | | -[](https://github.com/Y-Research-SBU/QuantAgent/pulls) |
39 | | -[](https://github.com/Y-Research-SBU/QuantAgent/issues) |
40 | | -[](https://github.com/your-username/QuantAgents/graphs/contributors) |
41 | | -[](https://github.com/Y-Research-SBU/QuantAgent/blob/main/LICENSE) |
| 36 | +[](https://github.com/Y-Research-SBU/QuantAgent/stargazers) |
| 37 | +[](https://github.com/Y-Research-SBU/QuantAgent/forks) |
| 38 | +[](https://github.com/Y-Research-SBU/QuantAgent/pulls) |
| 39 | + |
| 40 | +[](https://github.com/Y-Research-SBU/QuantAgent/issues) |
| 41 | +[](https://github.com/Y-Research-SBU/QuantAgent/graphs/contributors) |
| 42 | +[](https://github.com/Y-Research-SBU/QuantAgent/blob/main/LICENSE) |
| 43 | +[](assets/wechat.jpg) |
42 | 44 |
|
43 | 45 | </div> |
44 | 46 |
|
@@ -123,7 +125,6 @@ export OPENAI_API_KEY="your_api_key_here" |
123 | 125 |
|
124 | 126 | ## 🔧 Implementation Details |
125 | 127 |
|
126 | | -We built QuantAgents with LangGraph to ensure flexibility and modularity. We utilize gpt-4o and gpt-4o-mini as our deep thinking and fast thinking LLMs for our experiments. However, for testing purposes, we recommend you use gpt-4o-mini to save on costs as our framework makes lots of API calls. |
127 | 128 |
|
128 | 129 | **Important Note**: Our model requires an LLM that can take images as input, as our agents generate and analyze visual charts for pattern recognition and trend analysis. |
129 | 130 |
|
@@ -191,23 +192,7 @@ The system supports the following configuration parameters: |
191 | 192 |
|
192 | 193 | You can view the full list of configurations in `default_config.py`. |
193 | 194 |
|
194 | | -## 📊 Benchmark |
195 | | - |
196 | | -The `benchmark/` folder contains evaluation datasets for testing the QuantAgents system across multiple assets. For each asset, we collect 5000 historical bars via public trading APIs such as yfinance. |
197 | | - |
198 | | -From this data, we randomly sample 100 evaluation segments per asset. Each segment is composed of a sequence of 100 consecutive candlesticks, and the final three candlesticks are withheld from the input to prevent exposing the verified market outcome within the test-time prompt. The system operates in a zero-shot setting—without any supervised fine-tuning—by generating structured trade reports that include a directional decision (LONG or SHORT), a concise textual justification, and a predicted risk-reward ratio. |
199 | | - |
200 | | -### Available Assets |
201 | | -- **BTC**: Bitcoin (100 CSV files) |
202 | | -- **CL**: Crude Oil (100 CSV files) |
203 | | -- **DJI**: Dow Jones Industrial Average (100 CSV files) |
204 | | -- **ES**: E-mini S&P 500 (100 CSV files) |
205 | | -- **GC**: Gold Futures (100 CSV files) |
206 | | -- **NQ**: Nasdaq Futures (100 CSV files) |
207 | | -- **QQQ**: Invesco QQQ Trust (100 CSV files) |
208 | | -- **SPX**: S&P 500 (100 CSV files) |
209 | 195 |
|
210 | | -Each CSV file contains 4-hour candlestick data with OHLCV (Open, High, Low, Close, Volume) information for evaluation and backtesting purposes. |
211 | 196 |
|
212 | 197 | ## 🚀 Usage |
213 | 198 |
|
@@ -276,3 +261,8 @@ If you encounter any issues, please: |
276 | 261 | For questions, feedback, or collaboration opportunities, please contact: |
277 | 262 |
|
278 | 263 | |
| 264 | + |
| 265 | + |
| 266 | +## Star History |
| 267 | + |
| 268 | +[](https://www.star-history.com/#Y-Research-SBU/QuantAgent&Date) |
0 commit comments