Accessing indicator and model prediction data for plotting

[SW4T400]

Hi @edtechre @albertandking & pybroker community!

Question and Background

I was wondering if there is a way to access the indicator & model prediction data / columns that the strategy class uses. Similar to the following providing a TestResult class, I would like to also access the underlying data that produced the results.

result = strategy.backtest()

I am asking because I would like to plot

1. the DataSource alongside
2. the indicators and
3. the models prediction values
4. as well as all the trades (this I get from the TestResult Class without problem)

My idea is to allow plotting for all 4 stages with only 3 and 4 requiring a backtest. For example I might want to plot only the basic data with the indicators, just to develop an indicator. Later in the development process I would like to run a backtest to develop trading rules and/or ML models. At this point I would like to not only see the trades but also the models prediction values that caused the trades.

Currently I only know how to achieve 1. & 2. - can anyone tell me of a better way and/or tell me how to achieve 3. & 4.?

indicator_set = IndicatorSet()
indicator_set.add(cmma_20, hhv_5)
df_indicators=indicator_set(df)

all_in_one_df=pd.merge(df, df_indicators, on=['date', 'symbol'])

This gives me a simple dataframe to plot. By filtering for a specific symbol I could plot its price and some indicators.

For 3. I assume I have to start from the following lines in the tutorial:

model_slr = pybroker.model('slr', train_slr, indicators=[cmma_20])

config = StrategyConfig(bootstrap_sample_size=100)
strategy = Strategy(YFinance(), '3/1/2017', '3/1/2022', config)
strategy.add_execution(None, ['NVDA', 'AMD'], models=model_slr)

strategy.backtest(train_size=0.5) 

The last line in the code above ↑ seems to perform the training and testing, but obviously does not give any trades (which I dont need for stage 3. anyways as I just want to look at prediction values alongside the price). Could I extract the prediction values (for all symbols) from this line (somehow) or am I on the wrong track?

What would be the best way to get the DataSource columns, indicator & prediction values (plus the TradeResults) all together ? - where do I start looking?

Plotting

I am wondering if any of you have already developed plotting solutions for pybroker?

Here´s what I am trying to do..

I am using 1 minute bars and the only thing I found capable of plotting millions up to hundreds of millions of points is plotly-resampler. This works only for scatter plots, so I can show candlesticks only when the user has zoomed in enough for regular plotly to handle it. I am currenlty writing a dash app in order to display the data from 1. - 4. in my browser (but as mentioned I dont know how to access this data)

My hope is that I will manage to get visualizations per symbol (showing price, indicators, prediction values & trades) as well as portfolio views.

I find the plotting of backtesting.py and autotrader quite pleasing and am aiming for something similar.

• indicator view
• trading view

What are your thoughts on this? Im new to both coding & algotrading and would appreciate any help and/or comments.

[edtechre]

Note, you can use the return_signals option to return bar data, indicator data, and prediction values.