When the index rises but fewer stocks go up than down

When the market closes and the news says the Nasdaq rose 1.57%, most people assume it was a decent day. Yet on the same day, of the major overseas stocks Jupick was tracking, 55 (27.5%) rose and 143 (71.5%) fell. The index went up and most stocks went down. This actually happened on August 28, 2026.

Neither number is wrong. The two numbers count different things to begin with. The index counts how much money moved, and the advance-decline count counts how many stocks went up. When the counting rules differ, it is not strange to get different answers about the same day.

The sections below cover where that difference comes from, how to read the market temperature item on Jupick's screen, and which questions those numbers cannot answer. They are not about where the market goes the next day.

August 25 through 28, 2026 were consecutive trading days. Pulling the index changes and the advance-decline counts side by side from the records Jupick saved each day gives the following.

Start with overseas. On the 25th, the Nasdaq fell 0.76%, but 121 stocks rose and 77 fell. On the 26th, the Nasdaq rose 0.66%, but 78 rose and 119 fell. On the 27th, the Nasdaq fell 0.08% with 106 up and 90 down, and on the 28th, the Nasdaq rose 1.57% with 55 up and 143 down.

On four days running, the direction of the index and the direction of the stock count were opposite. Four out of four. Four days is a short stretch, so this ratio cannot be taken as a general probability. What it does confirm is that a split in direction is not a once-in-a-while accident.

Korea looked different. On the 25th, the KOSPI, Korea's main market index, rose 0.68% with 205 of 300 stocks up and 86 down; on the 26th, it rose 0.97% with 183 up and 107 down; and on the 28th, it fell 1.79% with 94 up and 194 down, so the directions matched. Only the 27th diverged: the KOSPI rose 1.05%, but 138 stocks rose and 157 fell.

On August 28, the biggest gainer on the overseas list was Salesforce at 22.58%, followed by Synopsys at 13.39% and Palo Alto Networks at 12.83%. On the other side, the biggest decliners were GE Aerospace at 3.29%, Boston Scientific at 3.11% and Booking Holdings at 3.03%.

The ranges differ. The top of the gainers is in double digits, while the top of the decliners stops in the 3% range. While 143 stocks each slipped a little, a few rose sharply, and if those few are large companies, the money-weighted total can easily end up positive.

The index is weighted by market cap. A bigger company gets more votes. The advance-decline count gives each stock one vote. Two companies whose market caps differ a hundredfold each get exactly one vote. With calculation rules this different, it would be stranger if the results always matched.

So rather than arguing over which one is the real market, it is easier to separate them by the question each one answers. The index is closer to how much all the money in that market moved. The advance-decline count is closer to the share of stocks that would have gone up if one were chosen at random from the list. Which one is closer to what you actually feel depends on whether your holdings are concentrated in a few large caps or spread across many stocks.

Over the four days, Korea matched in direction three times while the overseas side diverged all four times. With a sample of only four days, nothing should be concluded, but the difference in character between the two lists is worth noting.

During that period, the Korean list Jupick tracked was the top 300 by market cap and the overseas list was 200 stocks. The Korean 300, combining the KOSPI and KOSDAQ, Korea's second, growth-oriented market, reaches fairly far down. The overseas 200, by contrast, is just the top 200 out of thousands of US-listed stocks. The more a list is cut from the top only, the more heavily it leans toward mega caps even within itself, and the more room there is for the index and the stock count to split.

August 27, the day Korea diverged, has the same shape. The KOSPI rose 1.05%, but 157 of 300 stocks fell. That day's biggest gainers in Korea were Gaon Cable at 20.86%, Simmtech at 12.24% and Sanil Electric at 10.97%. But companies of that size cannot lift the KOSPI by 1%. The force that pushed the index up lay somewhere other than the top-gainers list, and since Jupick does not calculate each stock's contribution to the index, this goes no further.

One thing should be stated plainly. The advance-decline counts above are not counts of all Nasdaq constituents. They are counts of the list of major overseas stocks that Jupick updates daily. The Korean figures likewise cover a list of the largest companies by market cap, not every stock on the KOSPI and KOSDAQ.

So the sentence “the index rose but the market fell” is not accurate. Stated precisely, it becomes “the Nasdaq index rose, but among the major overseas stocks Jupick tracks, more fell than rose.” The index is not wrong; it counts something different.

The list size is not fixed either. The Korean list was adjusted from 300 to 200 stocks right after these four days of records. So comparing raw advance-decline counts from different periods leads astray. The 183 gainers on August 26 and 183 gainers on some day months later have different denominators. Comparing ratios rather than counts is safer.

These limits are spelled out not to say the numbers should not be trusted, but to make clear the range of questions they can answer. Numbers without a stated range tend to be used later more broadly than they were meant.

The market temperature item on Jupick's home screen shows the number of stocks that rose and fell, for Korea and overseas separately. The advance ratio is the number of rising stocks divided by the total of rising and falling stocks; for overseas on August 28, that is 55 divided by 198, or 28%.

The same area shows a fear and greed score. The score is an equal-weighted average of three things: the advance ratio just described, the average price change of the whole list capped at 3% in either direction, and the average of where each stock sits between its 52-week high and low. From August 25 to 28, the Korean score came down through 62, 55, 47 and 38, while the overseas score was 59, 49, 56 and 46.

Looking only at the index, overseas on August 28 was a good day, with the Nasdaq up 1.57%. By temperature, it was 46, below neutral. Putting the two numbers together reveals the shape of a rise pulled up by a few stocks at the top. The point is not which one is right, but that looking at only one hides something.

That said, this contrast cannot be said to indicate the next day's direction. Jupick does not make such calculations and does not intend to. What follows a rise with weak breadth is not the subject here, and it is not a question four days of records can answer.

The index and the advance-decline count measure different things. The index counts money weighted by market cap, and the advance-decline count gives each stock one vote.

So it is not rare for them to point in opposite directions. Over the four days from August 25 to 28, 2026, the direction of the Nasdaq and the direction of the advance-decline count on Jupick's overseas list were opposite all four times.

On the days they diverged, a handful of large caps were likely doing the pulling. On August 28 overseas, the biggest gains were in double digits while the biggest losses were in the 3% range.

Jupick's advance-decline count covers a list of the largest companies by market cap, not the whole market. When the list size changes, the counts change with it, so they should be read as ratios rather than counts.

This content is for information only and is not investment advice. Jupick does not suggest trading any particular stock and does not forecast the future direction of share prices. The specific stocks and indexes cited serve only to explain how index and advance-decline calculations differ and are not an evaluation or endorsement of any kind. The figures cited are values recorded on those dates and may be corrected later. Investment decisions and responsibility for them rest with the user.