from the timeseries realease example: http://pandas.pydata.org/pandas-docs/version/0.19.0/whatsnew.html#merge-asof-for-asof-style-time-series-joining making ticker as category:

trades.ticker = trades.ticker.astype('category') 
quotes.ticker = quotes.ticker.astype('category') 

merge = pd.merge_asof(trades, quotes,
                  on='time',
                  by='ticker')

Expected Output

merge.dtypes Out[31]: time datetime64[ns] ticker category price float64 quantity int64 bid float64 ask float64

geting : time datetime64[ns] ticker object price float64 quantity int64 bid float64 ask float64

Comment From: chris-b1

Yes, this would be nice to support and is essentially a special case of #10409.

Comment From: jreback

xref this to the duplicate issue.