What Market Data Showed Before the Last Correction
Price holds. The data already knows.
Market corrections are rarely invisible in the data before they appear in price. The characteristic pattern, documented across multiple correction episodes, is a period of divergence: price continuing to hold near highs or advance modestly while sentiment data, breadth indicators, and volatility dynamics are already signalling a deteriorating underlying environment. The divergence does not guarantee a correction. It identifies a condition in which the visible price action and the data infrastructure supporting it have stopped moving together.
A market correction is defined, conventionally, as a decline of 10% or more in a major index from a recent peak, typically occurring over weeks to months rather than the compressed timeframe of a crash. Corrections are more common than crashes and are frequently misidentified in real time: in the early stages, they are routinely characterised by financial media as temporary pullbacks or buying opportunities, because the narrative that built the preceding advance is still dominant. The data tends to be further along in its assessment.
This article uses the structure of what sentiment data typically shows before a major correction to illustrate the systematic reading of a category of market event that emotional responses consistently mischaracterise.
The divergence pattern: what the data shows while the narrative holds
In the weeks before a significant correction, NLP-driven sentiment analysis of financial news and corporate communications typically shows a pattern of subtle deterioration relative to the positive narrative still dominating price action. This does not manifest as dramatically negative headlines. It manifests as a gradual reduction in the proportion of positive sentiment readings, an increase in cautionary language in earnings transcripts, and a shift in the semantic content of central bank communications toward acknowledgement of risk.
The 2022 correction in global equity markets, which saw the S&P 500 decline approximately 25% from its January peak to its October trough, was preceded by a period in which sentiment data around rate sensitivity, inflation persistence, and earnings quality was becoming systematically more cautious while equity valuations were still embedding assumptions that the data environment was beginning to contradict. The Sentiment Layer was tracking this shift in the information environment in real time. The consensus was still pricing a soft-landing narrative.
This is the divergence that matters analytically. Not a single negative headline but a directional shift in the aggregate sentiment signal, occurring while price is still sending a different message. The gap between the two is the data's most informative reading.
Breadth data confirms what sentiment suggests: the advance is narrowing
A second pre-correction signal that quantitative systems consistently identify is the narrowing of market breadth: the shrinking proportion of instruments participating in the index-level advance. Major indices can continue to make marginal new highs while the median constituent is already declining, because the largest-capitalisation components have sufficient weight to sustain the aggregate.
In the period before the 2022 correction, the number of S&P 500 constituents trading above their 200-day moving average had been declining for several months before the index-level peak. The advance was increasingly concentrated in a small number of large-capitalisation names. The median stock had already entered correction territory before the headline index confirmed the move.
A Market Regime model incorporating breadth data alongside price and sentiment inputs was moving toward a transitional classification in this period. Not signalling a correction with certainty, which no systematic model can do, but reflecting the genuine deterioration in the breadth of the advance that the index-level chart was not yet showing. This is the practical value of regime classification over index-level price monitoring alone.
The emotional response: why the narrative outlasted the data
The emotional response to the pre-correction environment in late 2021 and early 2022 was shaped by several years of market experience that had rewarded buying dips and dismissing risk signals. Every prior volatility episode since 2012, with the exception of the brief 2020 crash, had resolved with a rapid recovery that punished those who had reduced exposure. The experiential prior of the investor population was strongly biased toward the conclusion that negative signals would resolve benignly.
This is availability bias operating at a generational scale: the most recent and most vivid experiences shaped probability assessments more strongly than the longer-term historical record would warrant. When the data was signalling genuine deterioration, the emotional framework around that data was filtering it through an assumption of imminent recovery.
The Opes Borsa Sentiment Layer does not carry this prior. It classifies the current information environment against its trained parameters, without the experiential weighting that a human analyst's recent history introduces. When the signal environment was becoming cautious in late 2021, the systematic reading was cautious. The emotional reading was still constructive.
The durable lesson: divergence between sentiment and price is the data's most consistent pre-correction signal
The pattern of sentiment divergence from price ahead of corrections is sufficiently consistent across episodes that it constitutes a genuine systematic input rather than a retrospective narrative. It does not produce a precise timing signal. It produces a reading of the current market environment that is more complete than price alone.
The Opes Borsa platform applies the Sentiment Layer alongside trend and regime data continuously across its coverage universe. The combination, expressed through the Signal Stack, provides exactly the multi-dimensional reading that pre-correction environments require: not a single indicator but the interaction between sentiment trend, breadth, regime, and price behaviour simultaneously. Explore the current reading at opesborsa.com.
Sentiment data showed the deterioration before it appeared in price. The systematic investor does not need to predict a correction to benefit from this. They need a framework that reads the environment accurately as it evolves.
Key Terms:
Market Correction: A decline of 10% or more in a major market index from a recent peak, typically occurring over weeks to months. Distinguished from a crash by its pace and from a bear market by its magnitude.
Sentiment Divergence: The condition in which NLP-driven sentiment data is trending in a different direction from prevailing price action, creating a gap between the information environment and the market valuation that reflects it.
Breadth Deterioration: The narrowing of the proportion of constituent instruments participating in an index-level advance. A historically consistent leading indicator of correction risk when it persists for several weeks.
Market Regime: The prevailing structural character of a market as classified by quantitative indicators. A regime model incorporating breadth and sentiment alongside price can move toward transitional classifications before index-level prices confirm the shift.
Availability Bias: The cognitive tendency to assign higher probability to outcomes that are more easily recalled, typically the most vivid and recent experiences. In investing, it causes systematic overestimation of the probability that past market patterns will persist.
