Reading the market

Why are my stocks falling when the index rises? Reading breadth and concentration

The closing news calls it a rising market, but most stocks on your watchlist are down. That difference does not automatically mean your stock selection failed. An index reflects company size; breadth asks how many stocks moved. This guide explains how to match coverage and dates. The 100-stock examples and calculator are fictional. The separate case for October 21, 2026 is recalculated from the KRX regular-session closing prices supplied by NAVER. It retains the original record that mixed after-market-inclusive final prices and reference prices, explains the correction, and states the remaining verification limits.

TapeFlowPublished / updated: 2026-10-3112 min read

What this guide answers

Index returns and the share of advancing stocks answer different questions.
Match the market, stock/ETF/ETN coverage, observation date and treatment of unchanged names before comparing breadth.
A divergence identifies what to investigate; it does not settle tomorrow's direction.

1. What actually rose when the index rose?

KOSPI uses a market-capitalization-based calculation. Price changes in different companies do not have equal influence. A large index constituent can rise enough to offset declines in many smaller companies. Reading a 1% KOSPI gain as 'most Korean stocks rose 1%' therefore starts with the wrong comparison.

Breadth changes the question to a count. Advancing, declining and unchanged stocks show how widely that day's movement was shared. A small company and a large company each count as one, so breadth can reasonably disagree with a capitalization-weighted index. More advancing stocks alone does not tell you their average return or the size of their gains.

Start with a suitable benchmark. A watchlist concentrated in KOSDAQ is unlikely to be explained fully by KOSPI. Compare the index and breadth for the same market first, then narrow the investigation to the sectors and company sizes you actually follow.

Sources: KRX — KOSPI market and index overview

2. Worked examples: 40 advancers and a 0.8% index gain

The first fictional market contains 100 stocks: 40 advance, 50 decline and 10 are unchanged. Advancers represent 40% of all stocks. Excluding unchanged stocks produces 40 ÷ 90, or approximately 44.44%. Neither result is wrong; the denominators differ. Net advancing breadth is (40 − 50) ÷ 100 = −10%, which is not a −10% index return.

The second example is a separate fictional capitalization-weighted basket. One stock starts with a 30% weight and gains 5%; the remaining 70% has a weighted return of −1%. The basket gains 0.8%. This is not a calculation from the preceding 100-stock distribution. It isolates the weight effect and omits actual index adjustments such as membership and share-count changes during the period.

Advancing share = advances ÷ (advances + declines + unchanged) × 100 Excluding unchanged = advances ÷ (advances + declines) × 100 Fictional basket return = 0.30 × 5% + 0.70 × (−1%) = +0.8%

Two separate fictional calculations, unrelated to current markets
Calculation	Result	Question answered
40 ÷ 100	40%	What share of all stocks advanced?
40 ÷ (40 + 50)	About 44.44%	What share of stocks with a price change advanced?
(40 − 50) ÷ 100	−10%	How large was the net advance-decline count relative to all stocks?
30% × 5% + 70% × (−1%)	+0.8%	What was the return of the separate capitalization-weighted basket?

3. What should a divergence make you investigate?

If the index rises while decliners outnumber advancers, ask whether a few heavily weighted companies supported the index. The concentration screen identifies large weights; check those companies' price moves separately. A high weight establishes potential influence, not a measured contribution to that day's index return.

If the index edges down while more stocks advance, investigate whether weakness in large constituents coexisted with rebounds elsewhere. Neither pattern establishes that the index must soon catch up or that other stocks must soon fall. Sector earnings releases, company-specific news and the chosen period can produce similar-looking divergences through different processes.

The practical result is a better comparison, rather than an automatic trade. Separate the broad market, sector and company-specific explanations when reviewing a holding. A decline on a day when most stocks fell calls for different follow-up from a decline in one company while its sector rose broadly.

Index up, breadth weaker: check the largest weights and the sectors that did not participate.
Index weak, breadth improving: distinguish large-stock influence from the size of rebounds elsewhere.
Index and breadth both rising: record broader participation, leaving persistence as a separate question.

4. Connect four screens after the close

Begin with the KOSPI or KOSDAQ indicator below and record its observation date and daily move. Then check Market breadth's coverage. The current live advance-decline summary and advancing-share denominator use provider-listed product counts that can include ETFs and ETNs. They are not stock-only breadth for the entire KOSPI or KOSDAQ market. To compare stock participation, verify advances, declines and unchanged names in the same dated stock list.

When the two disagree, open KOSPI or KOSDAQ heavyweight concentration to match the market you are comparing and identify its largest weights. Then use Korea Sector Flow to find sectors and individual contributors worth researching. Its estimates multiply foreign and institutional net share counts in selected stocks by closing prices. They are neither sums of actual execution values nor official totals for every stock in a sector.

Move from broad to narrow: 'check index direction → verify participation on a consistent basis → investigate my sector.' Exclude a reading from the conclusion if its coverage or previous prices do not match. The observed case below calculates the 30-stock sample’s daily directions and 20-day positions on the same regular-session price basis, while preserving the original mixed-basis record and its correction.

5. Daily breadth is different from participation above the 20-day average

Daily advancing share measures one day's price direction. The share above the 20-day moving average compares current prices with their recent trend. A stock can rebound today after several declines, and a stock falling today can remain above its 20-day average. Different readings are not necessarily contradictory.

The live 20-day participation and 20-session high-low readings use a sample of large-cap stocks with sufficient price history. That population differs from the daily listed-product count, which can include ETFs and ETNs. Check each stock's latest price date as well as the sample size. The separate observed case fixes its membership and regular-session prices to calculate daily directions and the 20-day mean. It is a separate record from the live screens, and its percentage does not describe the whole market.

Pairing an intraday index with yesterday's closing breadth, or KOSPI prices with KOSDAQ breadth, can create an artificial divergence. Distinguish the page's update time from the underlying observation date. If the dates differ, record the mismatch before interpreting the comparison.

6. A weekly note that remains comparable

At the weekend, place each day's index direction and advancing share side by side for the same market. The average of daily advancing shares summarizes daily participation; it is not the share of stocks that gained over the week. Weekly breadth requires comparing each stock's price at this week's end with its price at the previous week's end.

Use this copyable format: 'Observation date __ / market and product coverage __ / index change __ / consistent membership and previous prices checked __ / advancing __, declining __, unchanged __ / advancing share __ / facts checked about large constituents __ / separate move in my sector __ / explanation still unverified __.' Leave unverified readings blank. The final field keeps a hypothesis visibly separate from an established fact.

Next week, ask not only whether the index gained, but also whether participation widened and reached the sectors you follow. A consistent record turns 'the market is fine but my stocks are not' into a question you can investigate.

A real-data case with fixed dates and membership

2026-10-21: an index gain and different 20-day trend participation

KOSPI changed +3.29% that day. Comparing the same two dates’ regular-session closes in the fixed 30-member sample gives 18 advancing, 1 unchanged and 11 declining members. Although 60.00% rose over the day, only 6 of the 29 trend-eligible members (20.69%) exceeded their 20-day means. Index direction, one-day sample direction and recent trend answer different questions.

This record does not update with live prices. Keep it separate from the fictional examples and calculator. Regular-session prices are supplied by NAVER, not independently certified against exchange records or another provider.

2026-10-31 price-basis correction

The original record compared histories of after-market-inclusive final prices with the listing's previous reference prices, recording 30 prior-price differences and 7 direction differences. The correction uses NAVER's separate KRX regular-session close fields rather than changing values to make them agree. On the case date, final and regular-session prices differed for 25/30 members. They represent different price bases.

Preserved original JSON ↗Corrected data, inputs and checks ↗

KOSPI observations on the two dates: 5,552.74 → 5,735.66; +3.29%; 2026-10-20 → 2026-10-21
Regular-session directions in the fixed sample: 18 / 1 / 11; Advancing / unchanged / declining; 18 / 30 = 60.00%; Same two dates and price basis; unchanged included
Above the 20-day mean / trend-eligible members: 6 / 29 = 20.69%; 1 of the fixed 30 members is held because of a reference-price break. The mean includes 20 regular-session observations through the case date; equality does not count as above.
Price-basis checks repeated: 31/10/2026, 21:42 KST

Recalculate from the same inputs

KOSPI return: (5,735.66 ÷ 5,552.74 − 1) × 100 = +3.29%
Fixed-sample daily advancing share: 18 ÷ 30 × 100 = 60.00%
Each member's 20-day mean: sum of latest 20 regular-session closes, including the case date ÷ 20
20-day participation: 6 ÷ 29 × 100 = 20.69%

Daily direction is the difference between the two regular-session closes. Mean comparisons use unrounded inputs; only displayed results are rounded to two decimals. The daily sample uses 30 members, while the 20-day trend uses 29 members whose reference prices pass the continuity check. Keep the index, sample and periods distinct; do not extend these sample percentages to the whole market.

What was compared, and how far the checks go

We retained the 30 members fixed before inspecting results. ETFs and ETNs are excluded; a preferred share is retained. This is neither the whole KOSPI universe nor a common-share-only sample, and members were not replaced because results changed.
Dated inputs use the KRX regular-session close fields supplied by NAVER. After-market-inclusive final prices, previous reference prices and published changes are retained separately for checking their price bases.
The regular-session closes for 2026-10-20 match the original listing's previous reference prices for 30/30 members. Comparing the two dates on the same basis gives 18 advancing, 1 unchanged and 11 declining members.
Across the retained dates, next-session reference prices match prior regular-session closes in 719/720 comparisons. Re-fetched final prices match archived final prices in 708/720 comparisons. The JSON retains the remaining differences by security and date.
Trend classification is held for 1 member with a reference-price break within the 20-observation window. The fixed sample retains all 30 members; 6 of the 29 eligible members exceed their 20-day means. We did not invent a price adjustment or replace a held member.

Samsung Electronics' archived final price for 2026-10-20 was KRW 209,000, while the listing implied a previous reference price of KRW 211,500. The regular-session close checked for that date is KRW 211,500, matching the reference price. The correction stops treating the archived final price as a regular-session close.

Samsung Biologics (207940) has a regular-session close of KRW 1,429,000 on 2026-10-01, but a reference price of KRW 1,418,000 on 2026-10-02. A KRX notice confirms a paid-in-capital-increase ex-rights adjustment and that reference price. An adjusted regular-session series was not obtained, so its 20-day trend classification is held. The 2026-10-20→2026-10-21 daily comparison passes the continuity check and remains included. Read the ex-rights notice ↗

Price checks compare fields and calculations from the same provider. The KRX notice supports the ex-rights event and reference price above; it does not certify this whole case or NAVER's historical adjustment method. The full reported counts were not rebuilt constituent by constituent, and individual index contributions were not calculated.

Expand the fixed 30-member sample and trend checks

Prices and means are in KRW. Trend held: 삼성바이오로직스. Held members remain in the fixed list and daily count.

1. 삼성전자 (005930): Case-date regular close 223,500 · 20-day mean 252,775.00 · Not above
2. SK하이닉스 (000660): Case-date regular close 1,294,000 · 20-day mean 1,619,100.00 · Not above
3. 삼성전자우 (005935): Case-date regular close 172,200 · 20-day mean 190,375.00 · Not above
4. SK스퀘어 (402340): Case-date regular close 779,000 · 20-day mean 1,004,800.00 · Not above
5. LG에너지솔루션 (373220): Case-date regular close 411,500 · 20-day mean 384,900.00 · Above
6. 삼성전기 (009150): Case-date regular close 1,168,000 · 20-day mean 1,419,250.00 · Not above
7. 삼성바이오로직스 (207940): Case-date regular close 1,417,000 · Trend held: reference-price break
8. 현대차 (005380): Case-date regular close 309,500 · 20-day mean 334,650.00 · Not above
9. KB금융 (105560): Case-date regular close 160,600 · 20-day mean 166,735.00 · Not above
10. 삼성생명 (032830): Case-date regular close 245,000 · 20-day mean 269,975.00 · Not above
11. 삼성물산 (028260): Case-date regular close 290,000 · 20-day mean 324,150.00 · Not above
12. 셀트리온 (068270): Case-date regular close 197,600 · 20-day mean 183,625.00 · Above
13. 삼성SDI (006400): Case-date regular close 554,000 · 20-day mean 545,950.00 · Above
14. 기아 (000270): Case-date regular close 110,500 · 20-day mean 112,870.00 · Not above
15. 신한지주 (055550): Case-date regular close 91,700 · 20-day mean 102,410.00 · Not above
16. 한화에어로스페이스 (012450): Case-date regular close 814,000 · 20-day mean 952,050.00 · Not above
17. 두산에너빌리티 (034020): Case-date regular close 63,500 · 20-day mean 75,840.00 · Not above
18. SK (034730): Case-date regular close 482,000 · 20-day mean 537,750.00 · Not above
19. HD현대중공업 (329180): Case-date regular close 320,000 · 20-day mean 398,100.00 · Not above
20. 하나금융지주 (086790): Case-date regular close 121,800 · 20-day mean 128,020.00 · Not above
21. 현대모비스 (012330): Case-date regular close 346,000 · 20-day mean 364,400.00 · Not above
22. NAVER (035420): Case-date regular close 179,500 · 20-day mean 187,835.00 · Not above
23. LG전자 (066570): Case-date regular close 165,100 · 20-day mean 195,640.00 · Not above
24. 삼성화재 (000810): Case-date regular close 601,000 · 20-day mean 623,450.00 · Not above
25. SK이노베이션 (096770): Case-date regular close 154,300 · 20-day mean 151,885.00 · Above
26. LS ELECTRIC (010120): Case-date regular close 166,100 · 20-day mean 191,335.00 · Not above
27. POSCO홀딩스 (005490): Case-date regular close 301,000 · 20-day mean 307,900.00 · Not above
28. 우리금융지주 (316140): Case-date regular close 31,500 · 20-day mean 33,492.50 · Not above
29. HMM (011200): Case-date regular close 23,400 · 20-day mean 21,902.50 · Above
30. 메리츠금융지주 (138040): Case-date regular close 130,500 · 20-day mean 129,600.00 · Above

Provider counts retained from the original · excluded from the corrected comparison

The NAVER index page showed 372 advancing, 53 unchanged and 490 declining entries when the original was collected. The record is preserved, but we could not establish that the counts cover only the regular session. They are excluded from the corrected calculation and comparison with the regular-session KOSPI index and sample.

372 / 915 = 40.66% · 21/10/2026, 21:51 KST

Read the original JSON record →

Sources and how to check again

Check the KOSPI index values for 2026-10-20 and 2026-10-21 in NAVER Finance's daily-price view. The original advancing/declining counts have no verified session basis, so they are retained with their observation time but excluded from the corrected comparison.
Use the corrected JSON's fixed security codes and dates, keeping regular closes, final prices, reference prices and published changes distinct. Subtracting the signed change from the following session's final price supplies an additional reference-price check. Means use the provider's direct regular-session closes, not reconstructed substitutes.
KRX index material explains definitions and structure; the separate ex-rights notice explains the specific security's reference adjustment. NAVER Finance supplies prices and counts. The corrected JSON includes inputs, calculations, the hold rule and the location of the original record.

Try it: share of rising stocks

The starting values are fictional examples, not current market data. Use matching periods and coverage for your own inputs.

Rising stocks (stocks)

Falling stocks (stocks)

Unchanged stocks (stocks)

Rising share (including unchanged): 40%; Rising ÷ (rising + falling + unchanged) × 100
Rising share (excluding unchanged): 44.44%; Rising ÷ (rising + falling) × 100
Net breadth: -10%; (Rising − falling) ÷ all stocks × 100

Of 100 stocks entered, 90 changed price. Excluding unchanged stocks changes the denominator. Each stock gets one count; these figures do not measure a capitalization-weighted index return or the size of price moves.

Inputs are calculated only on this page and are not sent or saved.

A record to keep when checking real data

Write a short note for each item. An unanswered field is a prompt to investigate before drawing a conclusion.

1
Market
Did you begin with the relevant market and keep KOSPI separate from KOSDAQ?
2
Denominator
Did you check stock/ETF/ETN coverage, treatment of unchanged names and sample membership?
3
Observation date
Do the index, breadth and sector screens refer to the same underlying date?
4
Scope of the conclusion
Did you separate large-stock influence from a forecast of tomorrow's market?

Sources and preparation

An educational explainer prepared with AI assistance using official index definitions and provider data. The 2026-10-21 case uses the KRX regular-session closes supplied by NAVER and publishes the original mixed-basis record, correction and limitations. The earlier illustrative example and the calculator use fictional assumptions. Neither is a validated trading rule or a measure of an individual’s actual return.

KRX — KOSPI market and index overview ↗
Explains KOSPI's capitalization-based calculation. NAVER Finance supplies the observed prices; the fictional basket is not an official KRX contribution calculation.
NAVER Finance — KOSPI daily prices ↗
Open KOSPI → prices to check the October 20 and 21, 2026 dates and closing values. This is the price provider, distinct from the index-definition reference.
NAVER Finance — Korean market-cap rankings ↗
Choose domestic stocks → market capitalization → KRX and KOSPI to inspect coverage. Rankings change; use the case's frozen 30-stock list for recalculation. The sample includes a preferred share.
NAVER Finance — daily stock prices (Samsung Electronics) ↗
Check the dates and price basis in the daily-price view. After-market-inclusive final prices in ordinary daily bars are not inputs to the corrected mean. Recalculate with the regular-session closes retained in the corrected JSON's regularPrices.
NAVER Finance — previous reference and final-price definitions ↗
Explains why the previous regular-session reference price differs from the after-market-inclusive final price. Matching dates do not establish a matching price basis.