Data & Analytics
Elon Musk Tweets & Tesla Stock — Predictive Modeling
Sentiment, lag, ARIMA/ARIMAX, and causal-impact analysis of Elon Musk tweets (2010–2025) against Tesla prices, finding a statistically detectable but weak relationship (r ≈ 0.16).
Role: Analyst
Situation
Markets often treat high-profile social-media activity as an alternative-data signal. The question is whether tweet sentiment actually moves Tesla's stock once other drivers are acknowledged.
Decision question
Sentiment, lag, ARIMA/ARIMAX, and causal-impact analysis of Elon Musk tweets (2010–2025) against Tesla prices, finding a statistically detectable but weak relationship (r ≈ 0.16).
Data and constraints
See method and results below for sample, coverage, and documented limits.
Method
Merged Musk tweets (Kaggle, 2010–March 2025) with Tesla prices via quantmod. Scored daily sentiment with the Bing lexicon, then estimated contemporaneous and lagged correlations, regression of price on sentiment and tweet volume, ARIMA vs. ARIMAX with sentiment as an exogenous predictor, and CausalImpact counterfactuals around specific tweets.
Tools
- R
- tidytext (Bing lexicon)
- quantmod
- ARIMA / ARIMAX
- CausalImpact
- ggplot2
Deliverables
- 26-page research paper
- Correlation, lag, regression, and time-series results
Outcome / decision value
- Daily tweet sentiment and Tesla adjusted price were weakly positively correlated (r ≈ 0.156 contemporaneous; 0.156 at 1-day lag; 0.155 at 2-day lag).
- The relationship is statistically detectable but limited; market conditions and fundamentals likely dominate price moves.
- Investor implication: treat influencer tweets as noisy signals rather than a standalone trading rule.