News Sentiment vs. Price Action
Which Is the Better Predictor?
News sentiment and price action are not competing analytical frameworks. They are two different ways of observing the same market, from different vantage points, with different lead times and different failure modes. The question of which is the better predictor misses the more interesting question: under what conditions does each carry more information than the other, and how do the two interact when they disagree?
The answer has significant practical consequences. An investor who relies exclusively on price action is analysing the market's response to information after that response has begun. An investor who relies exclusively on news sentiment is analysing the information environment without knowing how the market will price it. Each approach alone is incomplete. Understanding how to combine them, and what their divergence signals, is the more useful analytical question.
What news sentiment and price action each measure
News sentiment, in a quantitative context, is a structured signal derived from the NLP processing of financial news, earnings communications, central bank statements, and market commentary. A Sentiment Layer classifies each incoming item as positive, negative, or neutral at the instrument level, and aggregates those classifications over a rolling time window to produce a composite directional reading of the information environment surrounding a given instrument. Sentiment measures what is being said about an instrument, with what directional weight, and at what velocity.
Price action is the observable record of how an instrument has traded: direction, velocity, volume, and structural characteristics such as trend consistency and support or resistance levels. A Trend Signal derived from price action captures the market's aggregate revealed behaviour, the sum of all buying and selling decisions by all market participants over the measurement period. Price measures what the market is doing in response to all available information, including information that is not captured in public news.
One measures the information environment. The other measures the market's response to all information. These are different things.
The lead-lag relationship between sentiment and price
The most important empirical finding in the academic literature on news sentiment and price action is that the relationship is asymmetric across time. Sentiment tends to lead price over short time horizons because news is processed by sophisticated participants before it is fully reflected in price. Price tends to lead sentiment over longer horizons because price reflects all information, including private information and positioning data that does not appear in public news.
This means sentiment is most informative as a near-term leading indicator: a shift in the news environment for an instrument often precedes the price move that reflects that shift by hours to days. It is least informative as a standalone signal over longer periods, because the market's price-based information processing eventually incorporates everything that public news contains and more.
Macro Signal Lag is a related concept: the measurable delay between a macroeconomic event and its full propagation into both sentiment data and price data. Sentiment typically responds faster than price to macro announcements, because NLP systems process the text in near real time while the price impact takes time to propagate through different asset classes, maturities, and geographies.
Where the two signals agree, and where they diverge
When news sentiment and price action are aligned, the signal from their combination is stronger than either alone. Strongly positive sentiment combined with a confirmed positive Trend Signal and rising volume is a more robust directional picture than any of these inputs independently. The Emotionless Edge operates most cleanly when the Signal Stack is coherent: multiple independent data sources pointing in the same direction reduces the probability that any single source is producing a noise artefact.
When sentiment and price diverge, the divergence is the signal. Several distinct patterns are analytically meaningful.
Positive sentiment against a failing price trend suggests the market has information or is weighing factors that the public news environment has not yet reflected. Private information, flow-driven selling, or a structural regime shift not yet visible in news coverage are all candidate explanations. This divergence warrants investigation rather than dismissal.
Negative sentiment against a resilient price trend suggests the market is discounting the negative news flow for a reason: either the news is being correctly assessed as less market-impactful than the headlines imply, or the broader structural trend is strong enough to absorb the negative sentiment. This divergence is associated with capitulation dynamics in declining markets and with the terminal phase of negative narratives before price turns.
The Regret Loop can be triggered by sentiment divergence: investors who act on negative sentiment against a resilient trend, exit early, and then watch price continue higher are the classic expression of reactive decision-making driven by information rather than by the market's integration of that information.
Which is more predictive, precisely defined
Over very short time horizons, high-velocity news sentiment is the stronger predictor of near-term price reaction. Over medium time horizons (weeks to months), price-based trend signals, particularly those that integrate momentum and regime classification, are more predictive of sustained directional movement. Over longer horizons, fundamental factors not fully captured in either sentiment or price action dominate.
Opes Borsa integrates both. The Sentiment Layer processes news in real time and feeds classifications into the broader signal framework alongside price, volume, and regime data. The resulting Trend Signal reflects the coherent view from all inputs, with the Signal Confidence Score adjusted for the degree of alignment across them. When sentiment and price agree, confidence is higher. When they diverge, confidence is calibrated to reflect the genuine uncertainty that the divergence represents. You can see this integration operating across instruments at opesborsa.com.
A framework for using both together
Use sentiment as a near-term attention signal: shifts in sentiment for a specific instrument, particularly when sentiment velocity is high, are worth examining against the price structure. Use price action as the structural arbiter: the trend and regime context from price data determines whether a sentiment signal is operating in a supportive or contrary structural environment. Treat divergence between the two as a genuine analytical question rather than a reason to act on either signal independently.
Neither is universally better. Both are more useful together than alone.
Key Terms:
Sentiment Layer: The NLP-driven component that classifies incoming financial news as positive, negative, or neutral at the instrument level, producing a structured directional signal from unstructured text.
Price Action: The observable record of an instrument's price, volume, and structural characteristics. Reflects the market's aggregate response to all available information, including private information not captured in public news.
Macro Signal Lag: The measurable delay between a macroeconomic event and its full propagation through sentiment data and price data across affected instruments. Sentiment typically responds faster; price adjustment takes longer to complete.
Trend Signal: In the Opes Borsa framework, the probabilistic directional assessment generated from the integrated Signal Stack of price, volume, sentiment, and regime data. Distinct from any single-input signal.
Signal Stack: The combined layers of analytical inputs feeding into a composite signal output. A Signal Stack integrating both sentiment and price data produces more contextually grounded directional assessments than either input alone.
