Counting Distribution Days: Reading Overall Market Health

There is a recurring pattern in the autopsies of major market tops. When traders review the weeks preceding a significant correction, they almost always find the same thing: the index was still making new highs, sentiment was still bullish, leading stocks were still breaking out — and yet, underneath the surface, something had quietly shifted. Volume had started behaving differently. Gains were being given back in the afternoon. Small down days kept arriving on expanding volume, while up days came on lighter participation. By the time the tape "broke," the damage had already been done. The top was not a single event; it was a process, and that process left evidence.

That evidence has a name. It is called distribution — the institutional act of selling into strength, of handing stock back to retail buyers at or near the highs. A single session of distribution is meaningless. Stocks sell off. Funds rebalance. Noise happens. But when distribution accumulates — four sessions, five sessions, six — on a broad-market index within a short window, it stops being noise and becomes signal. The market is telling you that the largest, best-informed participants are no longer net buyers. They are stepping aside, or worse, they are stepping out.

The framework for counting and interpreting these sessions was developed by William J. O'Neil and institutionalized through Investor's Business Daily, and it remains one of the few market-health indicators that has earned the respect of both trend-followers and academic researchers. What follows is an honest working tour of the methodology — what qualifies, what doesn't, how to track it, what the counts mean, and why the same machinery also generates the cleanest signal for when a new uptrend has begun.

The Anatomy of a Distribution Day

The formal definition is deceptively simple. A distribution day is a session in which a major market index — the S&P 500, the NASDAQ Composite, or the Dow — closes down at least 0.2 percent on higher volume than the previous session. Two conditions, both mandatory: a decline of a specified minimum, and volume that expanded relative to the prior day.

The decline threshold is worth pausing on. O'Neil's original publications used 0.2 percent as the qualifying cut-off, and that number has attracted criticism over the years for being too permissive — surely a 0.2 percent decline is market noise, not institutional selling. The criticism misses the point. The threshold is deliberately low because the volume condition is doing the heavy lifting. On any given day, an index can close down a tenth of a percent simply because mid-caps traded slightly negative. But an index that closes down even a quarter of a percent on expanded volume is telling you something structurally different: more shares changed hands on a down day than on the preceding session. That is the quiet signature of supply outpacing demand.

Why volume is non-negotiable. Institutions move size. When a pension fund trims a position, or a mutual fund complex rotates out of a sector, the volume shows up in the tape because the transaction is simply too large to be absorbed invisibly. Retail selling, by contrast, is distributed across thousands of small orders and tends to register as average or below-average volume. A down day on lighter volume is often healthy — it means the decline lacked conviction, that supply wasn't motivated. A down day on expanded volume inverts that story entirely: supply was motivated enough to drive prices lower and the selling was large enough to expand the session's participation.

Why the index, not individual stocks. A single stock can distribute for a hundred idiosyncratic reasons — a downgrade, a lockup expiry, an insider sale, a weak earnings whisper. A broad index cannot. When the S&P 500 posts a distribution day, the weight of many simultaneous institutional decisions is required to move it. That is why distribution day counts are read off the indices rather than off individual names, and why the NASDAQ Composite — the most growth-heavy of the majors — tends to register distribution earliest and most sharply when risk appetite contracts.

Stalling Days — Distribution in Disguise

The clean version of distribution is the down-volume down-close. But institutions are not always willing to be that obvious. When markets are still in uptrends and sentiment is still buoyant, large sellers often use a subtler technique: they distribute into strength. The session opens firm, runs to a new high intraday, and then grinds sideways or reverses, closing flat or barely positive on volume that is conspicuously larger than the previous day's. The chart shows green. The tape shows distribution.

O'Neil formalized this as a stalling day — a session in an uptrend that rises but fails to make meaningful progress despite heavy volume. The canonical version closes in the lower half of the day's range, near or below the midpoint, with volume equal to or exceeding the prior session, and a gain that is noticeably smaller than the volume expansion would warrant. A 0.1 percent advance on volume twenty percent above the prior session is not strength — it is demand meeting supply and losing the tug of war.

Why stalling matters more than raw declines. A clean distribution day is easy to see and therefore easy to discount — "it was just a bad session." A stalling day is camouflaged as a good day. Traders who count only the red candles miss the churn entirely, and churn is where the informed seller does the most efficient work. Many experienced practitioners treat stalling days with the same weight as distribution days, adding them into the running count when the intraday profile makes institutional supply obvious. Some treat them as half-weight. Either approach is defensible; refusing to count them at all is not.

The 25-Session Rolling Window

Distribution days do not accumulate indefinitely. The count is measured over a 25-trading-session rolling window — roughly five calendar weeks — and days age out of the count as they fall outside that window. A distribution day that occurred twenty-six sessions ago no longer counts. A fresh one registered today becomes the newest entry. The count is always a snapshot of recent institutional behavior, not a cumulative tally stretching back forever.

Why twenty-five sessions. The number is not arbitrary, though it is not sacred either. Five weeks is long enough for a genuine distribution pattern to develop — institutions rarely complete meaningful selling in three or four sessions — and short enough that stale information doesn't contaminate the read. A distribution day from seven weeks ago happened in a different market regime; rolling it into today's count would be like using last month's weather to dress for tomorrow. Twenty-five sessions preserves the rolling, adaptive quality that makes the indicator responsive.

The rally-reset rule. There is one additional mechanism that can remove days from the count before the 25-session expiration. When the index advances a substantial amount — typically on the order of five percent or more from the lowest point within the count window — some of the older, less damaging distribution days can be considered "erased." The logic is that if buyers have been strong enough to push the index meaningfully higher, the supply registered weeks ago has been absorbed. The threshold is not codified rigidly — some practitioners use 4 percent, others 6 percent, some use a percentage based on the prior distribution days' combined decline — but the principle is that strong rallies partially cleanse the count, and weak, sideways grinds do not.

The combined effect is that the 25-day window behaves like a short-term memory with a reset button. Days age out naturally after five weeks; genuinely strong buying accelerates the aging of older, smaller distribution events. The count that matters is always the one you have today.

The Count Thresholds — From Noise to Warning

A running count of one or two distribution days within a five-week window is background noise. Markets do not rise in a straight line, and isolated sessions of institutional profit-taking are part of how healthy uptrends breathe. The count begins to carry informational weight as it climbs. The table below summarizes how practitioners who follow O'Neil's framework typically interpret the running total, paired with the kind of market behavior that tends to accompany each zone. The percentages are illustrative composite tendencies observed across multiple cycles, not hard statistics — the point is the directional relationship between count and outcome, which has been remarkably stable across decades of market history.

Count	Regime	Typical Behavior	Trend Continuation
1 – 2 days	Normal	Routine profit-taking. Uptrend intact. Leaders still setting up.	High
3 days	Watch	First structural caution. Raise vigilance on breakouts.	Moderate
4 days	Caution	Historical threshold where correction odds rise materially.	Mixed
5 days	Warning	Reduce new-position aggression. Tighten existing stops.	Low
6+ days	Correction	Broad market under distribution. Step aside or go defensive.	Poor

The four-to-five distribution day threshold is the one that matters most. This is where the historical evidence shifts from "uptrend with friction" to "uptrend at risk of failure." It is not a mechanical sell signal — the count has flashed warning and then corrected itself plenty of times — but it is the point at which a disciplined trader begins to take defensive action. That action rarely means liquidating the portfolio; it means pausing new long entries, honoring stops more rigorously, avoiding fresh breakouts that would normally qualify, and accepting that the environment has become arithmetic-unfriendly for long-only momentum.

Six and beyond. Once the count reaches six or seven sessions within the rolling window, the historical record is unambiguous: the probability of a meaningful correction within the following weeks rises sharply. The market can still bounce, still rip shorts, still produce false dawns — but the expected outcome over the subsequent two to six weeks tilts decisively toward drawdown. Traders who continue to press long exposure at this count are not being brave; they are betting against the base rate.

The Follow-Through Day — Confirmation of a New Uptrend

The distribution day framework has a natural counterpart, and it is arguably the more valuable of the two. Once the market has corrected — whether mildly or severely — the question every trader faces is the same: when is it safe to re-engage? Markets lie about their bottoms. They produce oversold bounces that fail, snap rallies that reverse in two days, V-shaped reversals that aren't. Re-entering too early can produce a drawdown inside a drawdown; re-entering too late surrenders the most explosive leg of the new advance.

O'Neil's solution is the follow-through day, and it is structurally elegant. The setup requires two phases. First, the market must be attempting a rally — typically defined as an index that has put in a short-term low and is making higher closes. That rally attempt begins on day one. Then, somewhere between day four and day seven of the attempt, the index must post a session that advances 1.25 percent to 1.7 percent or more on volume higher than the previous day. That session is the follow-through.

Why early days don't qualify. A rally attempt on day one, two, or three of the bounce is too reflexive to carry information. Oversold markets bounce. Short-covering produces mechanical rallies. The first few sessions of a bottom attempt are indistinguishable from the first few sessions of a failed dead-cat bounce, because the flow driving them is identical in both cases: forced covering, reflexive dip buying, and tactical mean-reversion. By waiting until at least day four, the trader allows the easy, mechanical flow to exhaust itself. What remains after that exhaustion is conviction buying — and conviction buying on expanding volume is what the follow-through captures.

Why volume expansion is mandatory. A 2 percent up day on lighter volume than the prior session is not a follow-through. It is a vacuum rally — price moved because sellers stepped away, not because buyers stepped in. The indicator specifically demands that more shares traded on the up-volume session than on the previous one, because institutional accumulation, like institutional distribution, shows up in volume. Without the volume expansion, the rally has no buyer-sponsorship, and buyer-sponsorship is the entire point.

Tracking the Count — A Practical System

The mechanics of tracking distribution days and follow-through days do not require sophisticated infrastructure. A simple spreadsheet, updated daily after the close, is sufficient. The essential columns are the date, the index being tracked, the close, the day-over-day percent change, the volume, and the volume change. A boolean column flags qualifying distribution days; another flags stalling candidates based on close-within-range and volume-expansion criteria. A running count column totals qualifying sessions within the trailing 25 rows.

Which indices to track. For most momentum traders, the primary read comes from the NASDAQ Composite and the S&P 500. The NASDAQ is more growth-heavy and therefore more sensitive to risk-appetite shifts; it often registers the first distribution days of a correction and also the first follow-throughs of a new uptrend. The S&P provides confirmation and a broader read. When the two indices diverge — NASDAQ distributing while the S&P holds firm, for example — the divergence itself is informational and usually means growth leadership is thinning while defensive sectors hold bids.

The weekly review. Once a week, the running count should be interpreted alongside two complementary reads: the behavior of leading stocks (are they breaking out cleanly, or are breakouts failing within days?), and the condition of market breadth (are new highs expanding or contracting?). A distribution count of five combined with failing breakouts and contracting breadth is a far stronger defensive signal than the same count read in isolation. The count is a lens, not an oracle.

The Broader Principle

The distribution day framework is a specific answer to a more general question every active trader must address: how do I know when the regime has changed? It is tempting to treat regime assessment as either gut feel ("it feels toppy") or permanent structural bearishness ("a crash is always three weeks away"). Neither approach is actionable; neither survives contact with a real P&L curve. What O'Neil's method provides is a falsifiable, countable, repeatable mechanism for converting the vague feeling that "something is off" into a specific number that either crosses a threshold or doesn't.

The deeper insight is that institutional behavior is observable. It does not announce itself with press releases or television appearances. It shows up in the tape, in volume, in the subtle asymmetries between up sessions and down sessions, in the repeated appearance of stalling days inside a still-rising chart. Learning to read those signatures is not a form of market timing — at least not the kind of market timing that fails. It is a form of market listening. The signals have been there in every major correction of the last seventy years. They will be there in the next one. The only question is whether the trader is counting.