Introduction

Analysts suffer from confirmation bias: they selectively emphasize information supporting their conclusions. Detecting confirmation bias in reports helps investors identify analysts with biased views vs objective analysts. Machine learning text analysis measures confirmation bias by comparing report language to underlying data.

Confirmation Bias Detection

For each analyst report on a stock, extract language about positive/negative factors. Compare word emphasis (what percentage of words discuss positive vs negative factors) to actual factor balance (what percentage of recent data suggest positive vs negative). High discrepancy suggests confirmation bias. NLP extracts sentiment alignment with conclusions.

Analyst Credibility Assessment

Identify analysts with low confirmation bias (objective) vs high bias (selective). Track recommendation performance: do objective analysts outperform biased analysts? Results: low-bias analysts have better recommendation accuracy. Investors can weight low-bias analysts more heavily.

Conclusion

Detecting confirmation bias in analyst language enables more credible analyst selection.