Brain Sentiment Data › Part 4: Brain Language Metrics on Earnings Call Transcripts

Brain Language Metrics on Earnings Call Transcripts¶

In this notebook, we examine the Brain Language Metrics on Earnings Call Transcripts dataset (BLMECT).

Start by setting the bundle for this notebook:

In [1]:

from zipline.research import use_bundle
use_bundle("usstock-1d-bundle")

We create a Pipeline that contains the MD_SENTIMENT field, which measures the sentiment of Management Discussion section of the earnings call transcript.

In [2]:

from zipline.pipeline import Pipeline, brain, master
from zipline.pipeline.factors import AverageDollarVolume

# limit analysis to stocks
universe = master.SecuritiesMaster.SecType.latest.eq("STK")

avg_dollar_volume = AverageDollarVolume(window_length=90)

md_sentiment =  brain.BLMECT.MD_SENTIMENT.latest

pipeline = Pipeline(
    columns={
        "md_sentiment": md_sentiment,
    },
    initial_universe=universe,
    screen=(
        avg_dollar_volume.top(1000) 
        & md_sentiment.notnull()
    )
)

Next, we run the Pipeline and generate the tear sheet. Stocks with highly positive sentiment (quantile 5) outperform stocks with highly negative sentiment (quantile 1).

In [3]:

import alphalens as al

factor_data = al.from_pipeline(
    pipeline,
    start_date="2012-01-02", # start of dataset
    end_date="2024-03-31",
    factor="md_sentiment",
    periods=[1, 21],
    quantiles=5,
    segment="Y"
)

Factor Distribution
	min	max	mean	std	count	avg daily count	count %
Md Sentiment Quantile
1	-1.000	0.647	0.217	0.302	472,631	153.5	20.2%
2	0.250	0.793	0.580	0.092	468,211	152.0	20.0%
3	0.433	0.871	0.722	0.070	468,607	152.1	20.0%
4	0.586	0.935	0.833	0.053	467,643	151.8	20.0%
5	0.750	1.000	0.949	0.047	466,845	151.6	19.9%

Returns Analysis
	1D	21D
Ann. alpha	0.018	0.018
beta	-0.030	-0.039
Mean Relative Return Top Quantile (bps)	0.544	0.467
Mean Relative Return Bottom Quantile (bps)	-0.744	-0.662
Mean Spread (bps)	1.287	1.123

Information Analysis
	1D	21D
IC Mean	0.008	0.013
IC Std.	0.098	0.096
Risk-Adjusted IC	0.080	0.140
t-stat(IC)	4.448	7.744
p-value(IC)	0.000	0.000
IC Skew	-0.174	-0.489
IC Kurtosis	0.134	0.115

Turnover Analysis
	1D	21D
Quantile 1 Mean Turnover	0.013	0.183
Quantile 2 Mean Turnover	0.021	0.273
Quantile 3 Mean Turnover	0.023	0.292
Quantile 4 Mean Turnover	0.022	0.277
Quantile 5 Mean Turnover	0.014	0.197

	1D	21D
Mean Factor Rank Autocorrelation	0.994	0.873

Back to Introduction