Sentiment Trend is the change in news tone, not the tone itself. It compares how positive the last three days of headlines were against how positive the earlier headlines in the seven-day window were. A positive number means the coverage is warming up; a negative number means it is turning against the company.
The news sentiment score tells you the average tone of recent coverage — bearish, neutral or bullish. That is a level. But markets are forward-looking, and a level that everyone already knows about is, in principle, already in the price. A company that has been universally hated for six months carries a deeply negative sentiment level, and that fact is old news.
What is potentially not in the price is the change. A company that was hated and is now merely disliked is doing something different from one that was hated and is now loathed. Sentiment Trend isolates that movement: it deliberately throws away the level and reports only the direction of travel.
This is why the two numbers can point opposite ways, and why that is not a bug. A stock can carry a negative sentiment level and a positive sentiment trend at the same time — the news is still bad, and it is getting less bad. Those are two different facts and the card shows you both.
Every headline in the trailing seven-day window carries a tone score running from −1 (strongly bearish) through 0 (neutral) to +1 (strongly bullish). We split those headlines into two groups at a hard boundary three days ago: "recent" (published within the last three days) and "older" (the rest of the seven-day window).
Sentiment Trend = mean(recent scores) − mean(older scores).
That is the whole computation. It is a difference of two averages, so it can in principle range from −2 to +2, though in practice it clusters near zero. The card colours it green above +0.10, red below −0.10, and leaves it neutral grey in between — that dead band exists because small values are noise and should not be dressed up as a signal.
One rule you must know, because it is invisible on screen: if either group is empty — no headlines in the last three days, or none before that — the trend is reported as exactly 0.00. Not "unknown", not a dash. Zero.
Example (illustrative — invented numbers chosen to show the mechanics, not a live ticker). Over seven days a company produced six headlines. The four older ones scored −0.6, −0.4, −0.5 and −0.5, averaging −0.50. The two from the last three days scored −0.1 and +0.1, averaging 0.00.
Sentiment Trend = 0.00 − (−0.50) = +0.50. The card shows +0.50 in green.
Read that carefully, because it is the most common misreading of this metric. The news about this company is not good. The average tone is still negative. What the green +0.50 says is that the tone has stopped being awful — the recent coverage is merely neutral where the earlier coverage was hostile. That is a real, meaningful change, and it is not the same thing as good news.
Now change one detail: suppose all six headlines had been published four days ago and nothing since. The "recent" group is empty, and the trend reads exactly 0.00 — the same 0.00 you would see if tone were genuinely unchanged. The number cannot tell you which of those two worlds you are in. Check the News Count before you believe a zero.
Small samples make it jumpy. This is a difference between two averages that may each be built from two or three headlines. One unusually strident story can swing the whole number. Always read it next to News Count: a trend of +0.60 computed from twelve headlines is a finding; the same +0.60 from two headlines is a coin flip with a decimal point.
The zero is overloaded. As above, "no headlines on one side of the line" and "tone is genuinely flat" both render as 0.00. On a quiet ticker this is the common case, not an edge case. A 0.00 should be read as "no usable trend information," not as "the news is steady."
The three-day boundary is arbitrary and sharp. A headline published 71 hours ago is "recent" and one published 73 hours ago is "older," and they get sorted into opposite groups despite being two hours apart. On a ticker with few headlines, the trend can flip sign simply because time passed and a story crossed the line. Nothing about the world changed.
And news follows price at least as often as it leads it. Journalists write "shares slide on concerns about X" after the shares have already slid. A deteriorating sentiment trend frequently describes a fall that has already happened rather than predicting one that has not. Treat it as context on what the market is currently being told, not as a signal to act on by itself.
The change in news tone: the average tone score of headlines from the last three days, minus the average tone score of the older headlines in the trailing seven-day window. Positive means coverage is improving, negative means it is deteriorating.
The sentiment score is the level — how positive or negative the coverage is right now. The trend is the change in that level. A stock can have negative sentiment and a positive trend at the same time: the news is bad and getting less bad.
Because it is reported as zero whenever there are no headlines on one side of the three-day boundary — a common situation for quietly covered tickers. A 0.00 means "no usable trend," which is not the same as "tone is steady." Check the News Count alongside it.
Green above +0.10, red below −0.10, and neutral in between. The dead band exists on purpose: small values are noise, and colouring them would present randomness as a signal.
Browse all S&P 500 tickers to see this metric applied to individual companies.
Educational research only — not investment advice.