Table 4.

Summary statistics for daily bullish sentiment

Panel A: Twitter
	$Q_{1\%}$	$Q_{10\%}$	$Q_{25\%}$	Median	Mean	$Q_{75\%}$	$Q_{90\%}$	$Q_{99\%}$
Harvard-IV	−1.61	−0.69	0.00	0.56	0.51	1.10	1.61	2.56
LM	−2.14	−1.10	−0.41	0.00	0.05	0.69	1.10	2.14
L1	−1.59	−0.69	0.00	0.51	0.49	1.10	1.56	2.48
L2	−1.79	−0.69	0.00	0.41	0.41	1.10	1.61	2.56
VADER	−1.25	0.00	0.26	0.86	0.88	1.42	1.98	2.94
Naive Bayes	−1.20	−0.19	0.41	0.88	0.85	1.39	1.83	2.74
Max. entropy	−1.39	−0.41	0.12	0.69	0.62	1.10	1.61	2.48
Deep-MLSA	−1.95	−0.69	0.00	0.00	0.15	0.69	1.10	2.30
DeepMoji	−1.10	0.00	0.58	1.10	1.06	1.61	2.08	3.00

Panel B: StockTwits
	$Q_{1\%}$	$Q_{10\%}$	$Q_{25\%}$	Median	Mean	$Q_{75\%}$	$Q_{90\%}$	$Q_{99\%}$
Harvard-IV	−1.39	-0.69	0.00	0.00	0.24	0.69	1.10	1.79
LM	−1.39	−0.69	0.00	0.00	0.02	0.00	0.69	1.39
L1	−1.39	−0.69	0.00	0.41	0.38	0.77	1.39	2.20
L2	−1.39	−0.69	0.00	0.29	0.32	0.69	1.25	2.20
VADER	−1.10	−0.29	0.00	0.41	0.44	0.69	1.39	2.08
Naive Bayes	−1.10	−0.69	0.00	0.69	0.59	1.10	1.39	2.30
Max. entropy	−1.39	−0.69	0.00	0.69	0.49	1.10	1.39	2.20
Deep-MLSA	−1.20	−0.59	0.00	0.00	0.06	0.00	0.69	1.39
DeepMoji	−1.10	−0.69	0.13	0.69	0.63	1.10	1.61	2.40

Note: This table reports summary statistics for daily bullish sentiment constructed from Twitter (Panel A) and StockTwits (Panel B) data. More precisely, for each of the nine sentiment estimation approaches, the table reports the 1, 10, 25, 50, 75, 90, and 99%-quantiles as well as the mean daily bullish sentiment