optstrategy() and plotting

holicst

Hi Everybody, dear @backtrader,

When I try to use optstrategy() with a Pandas data source, I cannot plot my results to png files. As I understand, this should be possible, according to the documentation:

Cerebro.plot()
savefig: set to True to save to a file rather than plot
figfilename: name of the file. Use {j} in the name for the strategy index to which the figure corresponds and use `{i}`` to insert figure number if multiple figures are being used per strategy plot

I'd like to plot my results after each run has finished.

This is my code:

           if g_save_plot==True:
               plot_file_path = log_file_path.replace('.txt','_{j}_{i}.png')
               cerebro.plot(savefig=True, figfilename=plot_file_path) #numfigs=len(g_instruments), 
               cerebro.plot()

However, when I call this, I get an error:

Traceback (most recent call last):
 File "C:\Baixinho\train_keras_model.py", line 962, in <module>
   cancelled = bt_backtest.backtest(g_data_source, g_backtest_mode, g_resolution, loaded_training_df, training_fold_ranges_df, model, g_db, test_df, g_log_file_path)                
 File "C:\Baixinho\simulate_backtrader.py", line 995, in backtest
   cerebro.plot(savefig=True, figfilename=plot_file_path) #numfigs=len(g_instruments), 
 File "C:\Users\peter\Anaconda3\lib\site-packages\backtrader\cerebro.py", line 927, in plot
   start=start, end=end)
 File "C:\Users\peter\Anaconda3\lib\site-packages\backtrader\plot\plot.py", line 119, in plot
   if not strategy.datas:
AttributeError: 'OptReturn' object has no attribute 'datas'

And this is line 119 and 120 in plot.py (I've just upgraded backtrader to the latest version):

       if not strategy.datas:
           return

I don’t really get why “strategy” is an OptReturn object, but I’m not a Python expert.

Do I misunderstand the usage of cerebro.plot() when optimizing strategies?

Alternatively, it would be great to be able to save the plots separately, when a strategy finished (method stop()). Is that possible?

Cheers,
Tamás

backtrader

OptReturn is a simplified return instance which mimics a strategy, to reduce the overhead of message passing from worker processes to the main process.

See: Docs - Optimization Improvements

Plotting has never been tested with optimization (not an expected use case)