When during the day do Korean stocks rise?
630,000 stock-days of 30-minute bars
Every intraday trader has heard the folklore: the open pops and fades, lunch is dead, and the closing rush lifts prices after half past two. I took 30-minute bars for every Korean stock from July 2025 to August 2026 — 630,660 stock-days — and counted which slots actually went up.
01The session, and the rules
KOSPI is Korea's main board and KOSDAQ the secondary one. The regular session runs 09:00 to 15:30 Korea Standard Time (UTC+9) with no lunch break: continuous trading until 15:20, then a ten-minute closing auction that prints one closing price at 15:30. I split each day into thirteen 30-minute slots from 09:00, plus the closing auction, plus overnight (yesterday's close to today's open).
- Return per slot: each bar's open to its close. This slices the day rather than trading it, so no cost is deducted
- Buy the open, sell at each time: 09:00 open → each slot. 0.22% buy cost deducted
- Buy at each time, sell at the close: each slot's open → the 15:30 close. 0.22% deducted
- Weekday split, intraday versus overnight
Korea's second venue (Nextrade) trades before and after these hours for some stocks; those sessions are excluded because they do not cover every stock. The median is the middle result when everything is lined up from worst to best. For a single 30-minute slot the median is almost always exactly zero (half of all stocks do not move in any given half hour), so the slot table uses averages and the share of slots that rose.
02The day in 30-minute slices
| Slot | Share up | Average | |
|---|---|---|---|
| 09:00–09:30 | 39.9% | −0.221% | worst of the day |
| 09:30–10:00 | 44.7% | +0.039% | |
| 10:00–10:30 | 41.8% | −0.020% | |
| 10:30–11:00 | 42.0% | −0.006% | |
| 11:00–11:30 | 40.2% | −0.020% | |
| 11:30–12:00 | 40.3% | −0.021% | |
| 12:00–12:30 | 41.2% | +0.006% | |
| 12:30–13:00 | 42.5% | +0.038% | |
| 13:00–13:30 | 39.2% | −0.025% | |
| 13:30–14:00 | 41.6% | +0.024% | |
| 14:00–14:30 | 43.7% | +0.050% | |
| 14:30–15:00 | 42.8% | +0.018% | |
| 15:00–15:30 | 50.2% | +0.118% | the only slot up more than half the time |
| 15:30 closing auction | 35.0% | +0.001% | mostly unchanged |
| Overnight (prior close → open) | 44.7% | +0.287% | the whole day's return |
| Intraday (open → close) | 42.1% | −0.233% | |
| Full day (close → close) | 44.7% | +0.044% |
Three things stand out.
The first 30 minutes are the worst of the day. From the 09:00 open to 09:30 the average is −0.22% and only 40% of stocks rise. "The open pops" is true in the sense that the open is usually above yesterday's close — not in the sense that prices rise after the open.
Only the last 30 minutes (15:00–15:30) rise more often than not: 50.2%, average +0.12%. The closing rush is real. The ten slots between 10:00 and 15:00, though, average within ±0.05% of zero — noise.
The day's return is made overnight. Prior close to today's open averages +0.29%; open to close averages −0.23%. Add them and you get the full-day return of +0.04%. For thirteen months, the market lost money while it was open and made it while it was closed.
03Buy the open, sell when?
| Sell at | Win rate | Median | Average |
|---|---|---|---|
| 09:30 | 35.8% | −0.48% | −0.44% |
| 10:30 | 36.6% | −0.56% | −0.46% |
| 12:00 | 35.9% | −0.65% | −0.56% |
| 14:00 | 36.3% | −0.68% | −0.59% |
| 15:00 | 36.9% | −0.67% | −0.57% |
| 15:30 close | 39.2% | −0.55% | −0.45% |
Every exit lost. Buy at the open and you are down 0.48% (cost included) within thirty minutes; it drifts lower to a trough around 14:00 (−0.68%) and recovers a little in the closing rush. Holding to the close is the least bad option, and it is still a median −0.55%.
04Then buy when, and sell at the close?
| Buy at | Win rate | Median | Average |
|---|---|---|---|
| 09:00 | 39.2% | −0.55% | −0.45% |
| 09:30 | 41.9% | −0.30% | −0.23% |
| 11:00 | 41.3% | −0.22% | −0.19% |
| 12:30 | 42.2% | −0.22% | −0.10% |
| 13:30 | 42.5% | −0.16% | −0.08% |
| 14:00 | 41.9% | −0.16% | −0.08% |
| 15:00 | 38.4% | −0.16% | −0.10% |
The best time to buy was 13:30–14:00 — and even then the median was −0.16%, which is +0.06% before the 0.22% cost. No entry time during the session earned enough by the close to cover the cost of getting in. That single table is why day trading is hard: the average stock does not rise during the session.
05Large caps, and weekdays
Restricting to stocks above 100 billion won (about US$70 million) in market cap gives the same shape with a better night: overnight rose 51.9% of the time (all stocks: 44.7%), average +0.34%. The 15:00–15:30 slot was again the only positive one (52.1%). Intraday stayed negative (median −0.37%).
| Day | Intraday share up | Intraday median | Overnight (prior close → open) share up | Overnight average |
|---|---|---|---|---|
| Monday | 42.3% | −0.34% | 42.7% | +0.09% |
| Tuesday | 39.4% | −0.48% | 44.5% | +0.29% |
| Wednesday | 43.7% | −0.22% | 47.1% | +0.35% |
| Thursday | 42.0% | −0.33% | 46.5% | +0.37% |
| Friday | 43.4% | −0.26% | 42.8% | +0.34% |
Tuesday had the worst session (39.4% up) and the Monday morning gap over the weekend was the smallest (+0.09%). With thirteen months of data, treat the weekday split as a curiosity rather than a rule.
06Caveats
Thirteen months. Intraday history is only available for about a year, so the sample starts in July 2025. "Gains overnight, losses intraday" has been documented in US stocks for over two decades, so this is unlikely to be a fluke — but these particular numbers are one year's numbers.
30-minute bars hide what happens inside them. "09:00–09:30 is bad" is the sum of that half hour; whether 09:00–09:05 behaves differently, this data cannot say.
Equal-weighted. A day of Samsung Electronics counts the same as a day of a small cap. A cap-weighted index may look different.
Trade tables deduct only the 0.22% buy-side cost. For intraday trading the bid-ask spread is a large share of the real cost, so reality is worse. Delisted stocks are not included.
07What I take from it
- The first 30 minutes are the worst: −0.22% from the open on average, 40% of stocks up
- 10:00 to 15:00 is noise: averages within ±0.05%
- Only the last 30 minutes rise more often than not (50.2%, +0.12%)
- The day's return is made overnight: night +0.29%, session −0.23%
- Buy the open and every exit lost; buy at any time and sell at the close and nothing covered the cost
"Buy the open, sell the pop" was a structurally losing trade for these thirteen months. If you must buy during the session, the data points to after 13:30; if you must sell during the session, after 15:00. And even combining those, nothing beat buying yesterday's close and selling today's open.
The money was made while the market was closed.
While it was open, on average, the market gave it back.
댓글 쓰기