← Back to dashboard·

Claude

Anthropic
· 2026-07-10
Anthropic's Claude Sonnet. Tends toward balanced, well-calibrated analysis with moderate assumptions.
Calibration and accuracy describe observed behavior, not predictive ability. Experimental research — not investment advice. Terms of Use
Most Consistent1.3%/day
Claude is a moderate model with an average valuation gap of -5.8%. Most stable day-to-day (1.30%/day). Wins: Most Consistent. High validity rate (98.1%).

Model Positioning

Calibration score vs estimate consistency
Calibrated & ConsistentConsistent but UncalibratedCalibrated but VolatileLow ScoresCalibrationConsistencyGPTCLAUDEGEMINIDEEPSEEKGROK
Compare:GPTGeminiDeepSeekGrok

Key Metrics

-5.8%
Avg Gap
67/100
Calibration
16%
Correction Rate
35%
Positive Bias %
Consistent
Volatility
23
Days Tracked

Bias History

Model bias over time vs other models

Prediction Accuracy

Directional accuracy vs actual price changes
Directional Accuracy
45.5%
Mean Absolute Error
33.3%
Prediction Bias
-5.8%
Hit Rate (±5%)
6.1%
v80%25%50%75%100%10.6.16.6.22.6.26.6.2.7.9.7.DeepSeekClaudeGeminiGrokGPT
N = 506 · 1 Day

Current Estimates

Largest Positive Gap
+85.0%
+55.8%
+37.8%
+29.6%
+21.9%
Largest Negative Gap
-39.7%
-39.4%
-29.3%
-28.7%
-26.6%

Assumption Profile

7.5%
σ = 5.82
Revenue CAGR 5Y
9.3%
σ = 1.34
WACC
2.34%
σ = 0.44
Terminal Growth
24.0%
EBIT Margin Target

Validity

98.1%
2126 / 2168 validity rate

Cost & Performance

Avg Latency: 28.6s
Cost per Run: $0.0393
Total Cost: $85.23
~6629 + 1337 tokens

Calibration Detail

67
/100
Correction Rate: 16.4%
Uncapped Deviation:
Total Valuations: 530
Want these insights weekly?
Subscribe to AI Signals →
AI Investor Barometer · Claude← Back to dashboard
All content is generated by AI models and may contain errors. This is an experimental tool — not investment advice, research, or recommendation. Terms of Use · Privacy Policy