plot with 1m granularity takes forever about vectorbt HOT 5 CLOSED

I've tested it on both: M1 Pro Max 32gb ram machine
And a separated Linux server.

I guess, it's just CPU super intensive and it never ends.
Is there a way to specify the granularity to be used for plots only? And for backtest still use 1m granularity?

from vectorbt.

Plotly is relatively slow when it comes to plotting high-granularity data. In VBT PRO you can resample entire object to a higher timeframe before plotting, but here you need to use more basic plots such as pf.value().vbt.plot() together with plotly-resampler.

from vectorbt.

I see, still trying to figure it out how to combine the two libreries to reduce from 1m to 1h granularity.

What about quantstats?

 pf.qs.html_report(
                download_filename=TEARSHEET_HTML,
                output=TEARSHEET_HTML,
                title=titleReport,
                benchmark=pf.returns_acc.benchmark_rets,
            )

How could I produce a quant-stats report re-sampling it to lower granularity?

from vectorbt.

I'm currently trying as follows:

  resampled_benchmark_rets = pf.returns_acc.benchmark_rets.resample('h').last().ffill()
  vbt.settings.array_wrapper['freq'] = 'h'
  vbt.settings.returns['year_freq'] = '365d'

  pf.get_qs(
      freq='1h',
      benchmark_rets=resampled_benchmark_rets,
  ).html_report(
      download_filename=TEARSHEET_HTML,
      output=TEARSHEET_HTML,
      title=f'{symbol} Report',
      returns=resampled_benchmark_rets,
  )

Receiving the following error:
Error: save_quant_stats error ⚠️ operands could not be broadcast together with remapped shapes [original->remapped]: (54477,) and requested shape (3260371,)

Without resample, with 1m candlesticks it takes ~30min and it finally generates a 250MB html file that is way to heavy to be opened.

Any hint or suggestion is much appreciated.

from vectorbt.

resolved

from vectorbt.