Understanding Sentiment Percentiles
A raw sentiment score of 72 tells you almost nothing without knowing where that sits in historical context. Stonkly normalizes news, analyst, and social signals into percentile rankings that let you compare across stocks, sectors, and time periods. Here's exactly how it works.
The raw signal problem
News sentiment scores measure positive versus negative language in headlines and transcripts. Analyst sentiment tracks rating changes, price target revisions, and EPS estimate trends. Social sentiment aggregates volume and directionality of retail discussion. Each has different baselines, different volatility, and different lead/lag relationships to price. Combining them without normalization produces a number that reflects measurement scale more than actual market sentiment.
How Stonkly's normalization works
Stonkly runs a rolling 52-week percentile calculation on each component signal, then weights the components by their historical predictive power for each sector. A large-cap technology stock's social sentiment component is weighted lower than its analyst sentiment component — retail discussion is louder and less predictive in that context. A small-cap biotech inverts those weights because retail and news sentiment lead analyst revisions by more than the other way around.
Reading the percentile display
The score shown on the company profile is the weighted composite, expressed as a 0–100 percentile versus all stocks in the same sector and market-cap tier. A score of 80 means the stock is in the top 20% of its peer group for sentiment right now. The sparkline shows where that score has been over the last 90 days — the direction of the trend is as informative as the current level.
Practical use cases
Sentiment percentiles are most useful as a confirmation or contradiction layer on top of a technical or fundamental setup — not as a standalone signal.
- Momentum confirmation: enter a technical setup only when sentiment is above 60th percentile and rising
- Contrarian setup: price at 52-week high but sentiment dropped from 85 to 55 — divergence often precedes a pullback
- Earnings preview: sentiment that falls sharply in the week before earnings often signals negative whisper numbers
Key Takeaway
Don't use sentiment as a standalone signal. Use it to add or subtract conviction from a setup you identified on technical or fundamental grounds. The percentile is most useful when it's moving in a direction that surprises you given what price is doing.
More in Market Education