36 lines
644 B
Markdown
36 lines
644 B
Markdown
|
# Data Directory
|
||
|
|
|
||
|
|
CSV files are not tracked in git (too large).
|
||
|
|
|
||
|
|
## Required files
|
||
|
|
```
|
||
|
|
EURUSD_5M.csv
|
||
|
|
GBPUSD_5M.csv
|
||
|
|
USDJPY_5M.csv
|
||
|
|
```
|
||
|
|
|
||
|
|
## How to obtain data
|
||
|
|
|
||
|
|
1. Download 1-minute data from [HistData.com](https://www.histdata.com/):
|
||
|
|
- EURUSD, GBPUSD, USDJPY
|
||
|
|
- Years 2024, 2025, and 2026 (Jan-Mar)
|
||
|
|
- 1-minute bar format (MetaTrader compatible)
|
||
|
|
|
||
|
|
2. Place `.zip` files in this directory
|
||
|
|
|
||
|
|
3. Run the resampling script:
|
||
|
|
```bash
|
||
|
|
python3 ../scripts/resample_to_5m.py
|
||
|
|
```
|
||
|
|
|
||
|
|
This produces the `*_5M.csv` files needed for training.
|
||
|
|
|
||
|
|
## Data format
|
||
|
|
|
||
|
|
CSV columns:
|
||
|
|
```
|
||
|
|
open,high,low,close,volume
|
||
|
|
1.07451,1.07481,1.07432,1.07465,12345
|
||
|
|
...
|
||
|
|
```
|