Portfolio grouped stats on every parameters

[eromoe]

Since I haven't found a way to calculate grouped stats on vbt.Portfolio, I wrote below code and share here.

def grouped_stats(pf, metric_cols=None):
    if metric_cols:
        metric_cols = set(pf.metrics.keys()) & set(metric_cols)
        extra_metric  = set(metric_cols) - set(pf.metrics.keys())
    else:
        metric_cols = ['start', 'end', 'total_trades', 'max_dd', 'win_rate', 'total_return', 'benchmark_return', 'sharpe_ratio', 'calmar_ratio', ]
        extra_metric = ['annualized_return', 'annualized_volatility']

    a = [pf.loc[ind].stats(metrics=metric_cols) for ind in pf.wrapper.columns]
    stats = pd.concat(a,axis=1).T
    stats.columns = metric_cols
    
    if extra_metric:
        b = [[getattr(pf.loc[ind], c)() for c in extra_metric] for ind in pf.wrapper.columns]
        extra_stats = pd.DataFrame(data=b, columns=extra_metric, index=pf.wrapper.columns)
        stats = pd.concat((stats, extra_stats), axis=1)

    return stats

s = grouped_stats(pf)

And now you can use idxmax on more metric .

print(s.annualized_return.idxmax())

print(s.win_rate.astype(np.float).idxmax())  # don't know why  win_rate's dtype is object here , need convert first .

Now I am working on rolling stats, trying to understand how to reuse the vbt api.