QuantRocket logo
Disclaimer


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
 minmaxmeanstdcountavg daily countcount %
Md Sentiment Quantile       
1-1.0000.6470.2170.302472,631153.520.2%
20.2500.7930.5800.092468,211152.020.0%
30.4330.8710.7220.070468,607152.120.0%
40.5860.9350.8330.053467,643151.820.0%
50.7501.0000.9490.047466,845151.619.9%
Returns Analysis
 1D21D
Ann. alpha0.0180.018
beta-0.030-0.039
Mean Relative Return Top Quantile (bps)0.5440.467
Mean Relative Return Bottom Quantile (bps)-0.744-0.662
Mean Spread (bps)1.2871.123
Information Analysis
 1D21D
IC Mean0.0080.013
IC Std.0.0980.096
Risk-Adjusted IC0.0800.140
t-stat(IC)4.4487.744
p-value(IC)0.0000.000
IC Skew-0.174-0.489
IC Kurtosis0.1340.115
Turnover Analysis
 1D21D
Quantile 1 Mean Turnover0.0130.183
Quantile 2 Mean Turnover0.0210.273
Quantile 3 Mean Turnover0.0230.292
Quantile 4 Mean Turnover0.0220.277
Quantile 5 Mean Turnover0.0140.197
1D21D
Mean Factor Rank Autocorrelation0.9940.873

Back to Introduction