LLM Investment Bias Leaderboard
LLM Investment
Bias Leaderboard
The leaderboard is based on recently published paper, the details of which are provided in the github.
The leaderboard is based on recently published paper,
the details of which are provided in the github.
The leaderboard is based on recently published paper, the details of which are provided in the github.
LLMs are widely used for financial analysis but may exhibit intrinsic biases.
We developed a framework to evaluate these biases across models,
presented as a public leaderboard for transparent comparison.
This enables clearer understanding and more informed model selection in financial decision-making.
This enables clearer understanding and
more informed model selection in financial decision-making.
This enables clearer understanding and more informed model selection
in financial decision-making.
Acknowledgements
Acknowledgements
Acknowledgements
GPT-5.2 (none)GPT-4.1
Qwen3 235BGPT-5.1 (none)GPT-5.1 (high)
Grok 4 FastGPT-5GPT-5.1 (low)GPT-5.1 (medium)GPT-5.2 (high)
DeepSeek V3GPT-5.2 (low)GPT-5.2 (medium)GPT-5.2 (none)GPT-4.1Qwen3 235BGPT-5.1 (none)GPT-5.1 (high)Grok 4 FastGPT-5GPT-5.1 (low)GPT-5.1 (medium)GPT-5.2 (high)DeepSeek V3GPT-5.2 (low)GPT-5.2 (medium)Bias Index
Bias Index
Measures bias magnitude and consistency. High values indicate stronger,
less consistent bias; lower values reflect neutrality and stability.
Measures bias magnitude and consistency.
High values indicate stronger, less consistent bias;
lower values reflect neutrality and stability.
Cost
Cost
The total cost incurred to calculate the bias score.
The total cost incurred to calculate the bias score.
Latency
Latency
The total processing time required to calculate the bias score.
The total processing time required to calculate the bias score.
Experimental Design
This study evaluates whether LLMs exhibit intrinsic investment bias
under controlled conditions. To ensure reliability, the experiment minimizes hallucination by focusing on well-represented companies
from the S&P 500, encouraging decisions based on learned knowledge rather than speculation.
This study evaluates whether LLMs exhibit intrinsic investment bias under controlled conditions. To ensure reliability, the experiment minimizes hallucination by focusing on
well-represented companies from the S&P 500, encouraging decisions based on learned knowledge rather than speculation.
This study evaluates whether LLMs exhibit intrinsic investment bias under controlled conditions. To ensure reliability, the experiment minimizes hallucination by focusing on well-represented companies from the S&P 500, encouraging decisions based on learned knowledge rather than speculation.
A balanced prompt structure presents equal buy and sell arguments,
with each model making repeated decisions across identical inputs.
The results are aggregated into a bias score, capturing both direction
and magnitude of preference. This framework enables a consistent
and measurable comparison of how LLMs behave in financial
decision-making contexts.
A balanced prompt structure presents equal
buy and sell arguments,
with each model making repeated decisions across identical inputs.
The results are aggregated into a bias score, capturing both direction and magnitude of preference. This framework enables a consistent and measurable comparison of how LLMs
behave in financial decision-making contexts.
A balanced prompt structure presents equal buy and sell arguments, with each model making repeated decisions across identical inputs. The results are aggregated into a bias score, capturing both direction and magnitude of preference. This framework enables
a consistent and measurable comparison of how LLMs behave in financial decision-making contexts.
Bias Induction and
Measurement Procedure
Bias Induction and
Measurement Procedure
To evaluate the design,
we use a structured four-step process.
To evaluate the design,
we use a structured four-step process.
Balanced Prompt Input
Each stock is presented through a balanced prompt containing an equal number of buy and sell arguments. This ensures that the model receives neutral input conditions from the start.
Balanced Prompt Input
Each stock is presented through a balanced prompt containing an equal number of buy and sell arguments. This ensures that the model receives neutral input conditions from the start.
Repeated Evaluation
Each model is asked to make repeated decisions on the same stock under identical conditions. This helps capture whether its choices remain stable or shift across runs.
Repeated Evaluation
Each model is asked to make repeated decisions on the same stock under identical conditions. This helps capture whether its choices remain stable or shift across runs.
Decision Recording
For every stock, buy and sell outcomes are recorded across all trials. These results reveal the model’s overall directional tendency.
Decision Recording
For every stock, buy and sell outcomes are recorded across all trials. These results reveal the model’s overall directional tendency.
Bias Scoring
The recorded decisions are aggregated into a bias score ranging from -100 to 100. Higher values indicate a stronger buy bias, while lower values indicate a stronger sell bias.Variation across repeated runs is also incorporated to reflect inconsistency.
Sector Bias
100
100
Buy
Buy
50
50
0
0
-50
-50
-100
-100
Sell
Sell
Bias Score
Bias Score
Sector Bias &
Portfolio Implications
Sector Bias
Models exhibit statistically significant sector bias, with higher bias scores in Technology and Energy, and lower scores in Financials and Consumer Defensive.
Sector Bias
Models exhibit statistically significant sector bias, with higher bias scores in Technology and Energy, and lower scores in Financials and Consumer Defensive.
Preference Pattern
A persistent preference for certain sectors—particularly Technology—suggests that evaluations may be influenced more by sector affiliation than by underlying fundamentals or market conditions.
Preference Pattern
A persistent preference for certain sectors—particularly Technology—suggests that evaluations may be influenced more by sector affiliation than by underlying fundamentals or market conditions.
Investment Risk
This bias introduces risks of portfolio over-concentration, reduced diversification, and missed opportunities in underrepresented sectors.
Investment Risk
This bias introduces risks of portfolio over-concentration, reduced diversification, and missed opportunities in underrepresented sectors.
Size Bias
100
Buy
50
0
-50
-100
Sell
Bias Score
100
Buy
50
0
-50
-100
Sell
Bias Score
Size Bias &
Investment Implications
Market-Cap Framework
Market-cap quartiles are defined by five-year average capitalization, with Q1 representing large-cap and Q4 small-cap.
Market-Cap Framework
Market-cap quartiles are defined by five-year average capitalization, with Q1 representing large-cap and Q4 small-cap.
Size Bias Pattern
Models show statistically significant size bias, with higher bias scores in Q1 that decline toward Q4.
Size Bias Pattern
Models show statistically significant size bias, with higher bias scores in Q1 that decline toward Q4.
Investment Risk
A preference for large-cap stocks may distort evaluations, leading to overlooked growth opportunities and skewed recommendations toward dominant companies.
Investment Risk
A preference for large-cap stocks may distort evaluations, leading to overlooked growth opportunities and skewed recommendations toward dominant companies.
Momentum Bias
Momentum Bias &
Investment Implications
Measurement Framework
Momentum bias is measured by comparing opposing perspectives (e.g., momentum vs. contrarian) and calculating win rates across repeated decisions.
Measurement Framework
Momentum bias is measured by comparing opposing perspectives (e.g., momentum vs. contrarian) and calculating win rates across repeated decisions.
Preference Pattern
Most models show a statistically significant preference for the contrarian perspective, while some models exhibit momentum-oriented bias.
Preference Pattern
Most models show a statistically significant preference for the contrarian perspective, while some models exhibit momentum-oriented bias.
Investment Risk
Consistent bias toward a specific strategy may distort decisions, favoring one perspective even when opposing signals are stronger.
Investment Risk
Consistent bias toward a specific strategy may distort decisions, favoring one perspective even when opposing signals are stronger.
Where Global Data Becomes
Investment Conviction
Copyright © 2026 LinqAlpha Inc. All rights reserved.
Where Global Data Becomes
Investment Conviction
Copyright © 2026 LinqAlpha Inc. All rights reserved.
Where Global Data Becomes
Investment Conviction
Copyright © 2026 LinqAlpha Inc. All rights reserved.
