The Nasdaq Entered a Correction. Our Median Score Moved Two Points.
On Wednesday the Nasdaq closed more than 10% below its June high, which is the technical definition of a correction. The Dow fell over 1,100 points, its worst day since April 2025. Across our own board of 180 assets that day, the median score went from 81 to 81.
That is not a data error. It is the most honest illustration we have of what this model can and cannot see.
The Week On Our Board
- July 24 — median 83, with 114 names at or above the buy line and 49 at Strong Buy.
- July 27 — median 81, 103 above the line, 48 at Strong Buy.
- July 28 — median 81, 98 above the line, 49 at Strong Buy.
- July 29, the correction — median 81, 105 above the line, and Strong Buy readings rose to 54.
- July 30 — median 81, 103 above the line, 40 at Strong Buy.
The single most eventful day of the quarter registered as a slight increase in buy signals.
Why This Happens
Every asset is scored against its own recent average. A market-wide decline pushes almost every price below its own 20-day line at roughly the same time. Individually each name looks cheaper. Collectively, nothing looks unusual, because the model has no view of the market as a whole — only of each asset versus its own history.
Put differently: the score is a relative measure with no absolute anchor. It can tell you a stock is cheap relative to last month. It cannot tell you last month was expensive, or that the reason for the fall is still unfolding.
The Part That Isn't A Limitation
This is also why the model doesn't panic. It didn't downgrade 180 assets on a bad Wednesday, and mechanically refusing to capitulate on a red day is a real feature — that is what stops a mean-reversion system from selling exactly the bottoms it exists to find. Whether the same behaviour is a feature this week depends entirely on whether the decline is finished, which the score does not know and does not claim to.
What We Changed
We now record market breadth every day — median score, the count at each signal threshold, advancing versus declining, and how many names carry the falling-knife flag. Previously the board computed those numbers to render one line of text and then discarded them, which meant the question "what did breadth look like the day the Nasdaq broke?" could only be answered by reconstructing it after the fact, as we have done here.
From this week it is recorded automatically. We also began storing volume, the 20-day and 200-day averages, ATR, and the full score breakdown — so a future version of this article can ask which layer of the model drove a reading, instead of only reporting the number it produced.
What's Actually Actionable Here
Read the score as what it is: a measure of price relative to its own recent average, and nothing more. It is genuinely good at that one job. It will not tell you a correction is underway, because it was never built to look.
That is what the macro panel, the breadth numbers and the falling-knife flag are for — and this week is a reasonable argument for reading them first and the score second.
Board figures are from our own daily record. Index and correction figures are from external reporting. Nothing here is financial advice.
Scores in this piece come from the ZonedInOut model. How the score works →
Get the Sunday Brief
One email a week, before the open: what the board did, which calls worked, which didn't, and what changed. Free. No spam, unsubscribe anytime.