jesse-research

Analyze Jesse backtest results and indicators with clean TradingView charts in jupyter.

This project is a continuation of these two projects: Lightweight Charts for Jesse : https://github.com/Gabri/jupyterlab-lightweight-charts-jesse and Jesse Trades Info : https://github.com/nick-dolan/jesse-trades-info

It leverages the power of lightweight charts TradingView library imported to jupyter to visualize indicators and backtest results in a straightforward and modulable way for improved research.

Main Features:

1. Print costumizable candle and indicator series in TV clean charts.
2. Feed candles series from jesse database and calculate indicators directly within jupyter or import indicator series from .json file.
3. Read backtest json files and display an interactive table of trades that expands to display orders table and show them on chart.

Installation process:

1. Copy this repo into the root of your Jesse directory in your server.
2. Run Jesse and Jupyter Lab.
3. Import candles in Jesse.
4. Edit your strategy file like the example to select the indicators you want to save in the self.after method and dump the json files in the self.terminate method.
5. Run your backtest.
6. Go to jupyter, open the notebook, uncomment and run the first cell to install dependencies you may not have.
7. Select your backtest by editing the name file ; select the timeframe you want to look at ; warmup candles and initial balance.
8. Select and customize your indicators to plot on the chart in the last notebook cell.
9. Run the notebook.

Tips:

1. Read the notebook and strategy example to understand how to plot your own stuff nicely.
2. Include the methods used in the strategy file in a separate python file or in a new class inherited from Jesse Strategy.
3. Read the Lightweight chart documentation to see how to apply styles to your charts: https://tradingview.github.io/lightweight-charts/
4. Use the dump functions in the self.after method to analyze running live strategies.

Known issues:

1. "Javascript: require is not defined" may pop, just rerun the notebook.
2. Candlestick markers on orders may have an offset when using different timeframes than the backtest.
3. Jesse natively stores incomplete information on orders such as fees. Expect deviations between backtest results and live.