[Overseas Academic Paper Verification] 229,000 people and 72,000,000 tweets data prove: why chasing a "talkative financial influencer" leads to an average monthly loss of -2.3%?
To readers of GogoJungle, where investors and traders gather.
Have you ever felt something odd while watching influencers on social media repeatedly posting about “pre-emptive trading” or “market views”?
“Why can these people tweet so frequently while they should be trading?”
“Why do forecasts from famous accounts with tens of thousands of followers repeatedly go against the market?”
A very important paper that thoroughly tests this intuition using quantitative finance and big data analysis has been published.
A working paper by Ali Kakhbod of UC Berkeley, Seyed Kazempour of Louisiana State University, Dmitry Livdan of UC Berkeley, and Norman Schürhoff of the University of Lausanne and the Swiss Finance Institute (SFI)‘Finfluencers’(December 2024).
In this article, from the perspective of a full-time trader, we interpret the paper’s quantitative data (alpha tested via the Fama-French 5-factor model) and structurally analyze how the follower economy on SNS undermines individual investors’ performance.
1. Data set and validation method of the paper
A hallmark of this study is its enormous data size and rigorous empirical framework.
- Target population: 29,475 financial influencers on the major financial investment platform StockTwits
- Analytical data: 72 million buy/sell recommendations and sentiment tweets
- Evaluation model: Estimate abnormal returns for 20 days after a tweet using the Fama-French 5-factor model. Compute pure “True Alpha” after removing systematic risk (market, size, value, profitability, investment patterns).
2. Empirical results: the majority as “inverse gods” (reverse gods) — a shocking finding
Using a mixture model, the true ability distribution of financial influencers was categorized as follows.
| Classification (Finfluencer Type) | Share | Mean Monthly Alpha |
|---|---|---|
| Skilled Type 2 (true top performers) | 5.1% (0.6) | +8.14%(0.49) |
| Skilled Type 1 (competent) | 23.5% (1.0) | +1.49%(0.10) |
| Unskilled (no skill, market followers) | 16.6% (2.7) | 0.00% |
| Antiskilled Type 1 (inverse gods, losses) | 45.5% (1.8) | -1.19%(0.08) |
| Antiskilled Type 2 (extreme inverse gods) | 9.3% (0.5) | -9.15%(0.33) |
Overall,54.8% (about 55%)are “Antiskilled” generating negative alpha, and the weighted average of monthly abnormal returns reached−2.3%.
In the paper’s 18th page, the authors present the following stark statement.
"For the majority, tweets are informative in the sense of 'do the opposite of what I say'."
Statistically, it has been proven that the signals of most influencers only function as an “anti-trend filter.”
3. Why are the more talkative the most incompetent? A negative correlation between activity and skill
The most insightful analysis in this paper is the quantitative analysis of “Tweet Activity” and the probability of possessing skill (Table 4, p.23).
Change in the probability of possessing skill when tweet volume is increased tenfold (1 log unit) (pp):
- Probability of being Skilled: −3.34 pp (significant at 1% level, negative)
- Probability of being Unskilled: +2.45 pp (significant at 1% level, positive)
- Probability of being Antiskilled: +0.88 pp (significant at 1% level, positive)
On the other hand, the estimated follower growth elasticity (Table 5, p.27) shows a strong positive correlation:“If tweet count increases by 1%, followers increase by 0.72%.”This confirms the positive correlation.
In other words,
- Less skilled individuals who issue frequent market views
- Frequent posting and excessive optimism (hype) are favored by SNS algorithms, rapidly increasing follower counts
- The true skilled traders (e.g., Skilled Type 2 generating monthly +8.14%) focus resources on market observation and disciplined execution, resulting in low posting frequency and slow follower growth
This creates a market where “low-skill individuals dominate as influencers, while high-skill individuals remain obscure”—a complete adverse selection in the market.
4. Practical insights from pro traders
The reality highlighted by this paper aligns 100% with practitioners’ sense of trading.
Originally, a trader’s true edge lies in observing market volatility and liquidity and waiting until the expected value tilts positively. Merely holding a position exposes one to risk, and tweeting about every micro-move of the market with emotional fluctuations is a confession of lack of discipline.
The business model of influencers who chase followers is not “capital gains from trading” but “platform revenue from attracting audiences (salons, affiliates, advertising).” For them, what matters is impressions, not market outcomes.
5. Conclusion: understand algorithmic distortions and cultivate autonomous edge
The lesson from this paper, “Finfluencers,” is clear.
- Follower counts are not a measure of trading skill but merely a measure of marketing activity.
- Chasing others’ pre-emptive information and market views is a reckless move that embeds a negative expected value of −2.3% per month into your portfolio.
- The true tacit knowledge (situational awareness, money management, dynamic adjustment of market views) does not exist on SNS timelines.
The only way to survive in the vast ocean of information is to completely block out others’ behaviors as noise and execute disciplined trades based on your own validated data and statistical edge.