QuantConnect Community

Hey Nevo Raz,

I played around a little bit and wasnt able to Plot a chart with the consolidated 45 second bars. Tbh I'm not sure if w're even allowed to print so much data into a chart because we have the 4000 limit cap.

Either way, if you try to plot buy and sell signal into candle chart this is one way to do it:

#Initialize

self.candles = Series('Second', SeriesType.Candle)

self.chart = Chart('Candles')

self.chart.AddSeries(self.candles)

self.chart.AddSeries(Series('Longs', SeriesType.Scatter, "", Color.Green, ScatterMarkerSymbol.Triangle))

self.chart.AddSeries(Series('Shorts', SeriesType.Scatter, "", Color.Red, ScatterMarkerSymbol.TriangleDown))

self.AddChart(self.chart)



#After buy order

self.Plot( 'Candles', 'Longs', self.Securities["SPY"].Price - 10)

#After sell order

self.Plot( 'Candles', 'Shorts', self.Securities["SPY"].Price - 10)

You can adjust the position of the triangle by changing the value that you substract by the price.

You will see the buy signals when you clone the algo.

Hope it helps ;)

Nico Xenox Thanks for your explanation.

About the 4000 data points limit: 

If we add to the chart just one day of data (6.5 hours = 23,400 seconds of trading data) it gives us ~519 candles of 45 seconds each, it's still <4000. Even if each candle is made of 4 data points (OHLC) its 2076 points per graph, still <4000. In addition we have the buy/sell data points, obviously depends on the algorithm but for one day were limited to 1924 extra data points on the graph. 

Bottom line the 400 data points shouldn't limit us if I'm right.

I hope someone else figured it out,

Hi Nevo,

The charting software on the backtest results page automatically groups the datapoints to create the candlesticks. We can't enforce the candlesticks to span a specific period of time. For more information about candlestick charts in backtests, see Supported Chart Types. As an alternative, if you save the OHLC values of the consolidated bars into the ObjectStore, you can load them into the Research Environment to create a candlestick plot with any of these third-party libraries.

Best,
Derek Melchin

Want to invest in QuantConnect as we build the Linux of quant finance? Checkout our Wefunder campaign to join the revolution.

Hey Nevo Raz, I followed Derek Melchin  advice and build this small piece. I think by doing it this way, the only limitation you will have when wanting to plot a chart will be the 50 mb limit of the objectstore. (Chart is in Notebook)

One thing I wasnt able to do because I have not a lot of time was adding the buy / sell signals.

This can be archieved by saving the date values of the buy / sell order into a list. I already did that and saved it into the objectstore. The only problem I have is adding the triangles into the chart with the time values.

If there's a better way let me know. Whole thing is done with plotly 😊

Hope it helps ;)  

Hey again,

I added this line of code for the buy signals:

fig = go.Figure(data=[candlestick], layout=layout)

fig.add_trace(go.Scatter(mode="markers", x=buy_list, y=df['low'] - 0.15, marker_symbol="triangle-up",

                           marker_line_color="green", marker_color="green",

                           marker_line_width=0.01, marker_size=7))

The only problem is that the y-axis of the buy signal is not located 0.15 below the lowest point. 

Also Derek Melchin the bokeh library cant be importe, there's an error.

Do you guys know what to change to make it work? Or add a code to the scatter that it is not allowed to touch the candlesticks.

Hi Nico,

To be notified when the bokeh library is supported again, subscribe to GitHub Issue #6639.

The triangle markers aren't in the correct position because buy_list and df['low'] aren't aligned. In this case, we can fix the issue by replacing df['low'] with df.loc[buy_list]['low']. See the attached notebook for reference.

Best,
Derek Melchin

Want to invest in QuantConnect as we build the Linux of quant finance? Checkout our Wefunder campaign to join the revolution.

Thanks a lot Nico Xenox for sharing this code, it's exactly what I was looking for, I'll work on it and share here my additions.

Thanks Derek Melchin for your support, using the ObjectStore can be very useful whenever we need to think/use outside the box(=QunatConnect platform limitations).