ACADEMY / Markets and trading sessions

Mini-TAIEX night session: hours, liquidity and how to backtest it

The Mini-TAIEX night session trades 15:00 to 05:00 Taiwan time and counts toward the next trading day. Backtest it as its own segment, with wider slippage than the day session, before you combine it with day trading.

By Soren Huang · Published 2026-10-26 · updated 2026-10-31 · Editorial policy

Contents

The Mini-TAIEX night session, which TAIFEX calls the after-hours session, trades from 15:00 to 05:00 the next morning, Taiwan time. The regular session runs from 08:45 to 13:45. In a backtest, assign night bars to the correct session, use wider slippage than in the day session, and test day and night as separate segments before you combine them.

When does the mini-TAIEX night session trade?

The Mini-TAIEX night session trades from 15:00 to 05:00 the next morning, Taiwan time, after a regular day session from 08:45 to 13:45. The night session (夜盤) is the after-hours trading session that TAIFEX runs from the afternoon until early the next morning. Traders often call it 小台夜盤 or 小台指期夜盤. Taiwan does not use daylight saving time, so these local times stay the same all year.

Period (Taiwan time, UTC+8)	What happens
08:45 – 13:45	Regular (day) session
13:45 – 15:00	Break, no trading
15:00 – 05:00 next day	After-hours (night) session, crosses midnight
05:00 – 08:45	Break, no trading

The Friday night session runs until 05:00 on Saturday. There is no trading during the rest of the weekend. TAIFEX added the after-hours session in 2017, so older price history contains day-session bars only.

One Mini-TAIEX point is worth NT$50, and the minimum price move is one point. The free futures market hours tool shows these sessions in your own time zone, and the futures contract calculator turns points into money per contract. Exchanges can change session hours and contract terms, so check the TAIFEX contract page for the current values before you set up a test.

One Mini-TAIEX trading day starts with the night session at 15:00 and ends with the next day session at 13:45. Only the night block crosses midnight.

How do you date night-session bars that cross midnight?

Group bars by session, not by calendar date, because one night session covers two calendar dates and belongs to a third concept, the trading day. A trading day is the business day the exchange uses for settlement and margin. TAIFEX assigns trades in the after-hours session to the next regular session's trading day. A trade at 22:00 on Tuesday and a trade at 02:00 on Wednesday both count toward Wednesday's trading day.

This matters in a backtest in three places:

Session filters. A rule such as "trade only between 15:00 and 05:00" must wrap past midnight. A simple start-time-before-end-time check fails, because 05:00 is earlier on the clock than 15:00.
Daily statistics. If your software groups results by calendar date, one night session is split across two days. Daily loss limits and "one trade per day" rules then behave differently from what you intended.
Weekends. Bars from 00:00 to 05:00 on Saturday are dated Saturday but belong to Monday's trading day. A filter that drops all Saturday bars cuts every Friday night session in half.

Also check how your data source stamps its bars. Some vendors label a bar with its start time and others with its end time. With 15-minute bars, the first night bar may be labelled 15:00 or 15:15. The last night bar may be labelled 04:45 or 05:00. Look at a few days of raw data around 15:00 and 05:00 before you write any session rule. If your data builds daily bars, find out whether each daily bar includes the previous night session.

Why is slippage wider in the night session?

Night-session trading volume is usually much lower than day-session volume, so each order moves the price more and fills tend to be worse. Slippage is the difference between the fill your test assumed and the one the market gives you. In a thin market, the bid and offer sit further apart and hold fewer contracts. A stop order can then fill several points past its trigger.

Liquidity also changes within the night. Activity tends to pick up around US economic releases and the US stock market open. The New York open at 09:30 is 21:30 Taiwan time while the US uses daylight saving time, and 22:30 Taiwan time the rest of the year. Many US data releases come out at 08:30 New York time, which is 20:30 or 21:30 in Taiwan. In the early morning hours, when local traders are asleep and US markets are quiet, the book can be thin.

The practical rule: give the night session its own slippage assumption, and make it larger than the day-session one. The general method for adding commission, tax and slippage is covered in trading costs in backtests. The night session is where an optimistic cost setting hurts most.

Hypothetical example

A Mini-TAIEX strategy has a gross average trade of 8 points (NT$400). Commission and tax together are assumed at 2 points per round trip. In the day session you assume 1 point of slippage on each side, so the round trip costs 4 points and the net average trade is 4 points. At night you assume 2 points on each side, so the round trip costs 6 points and the net average trade is 2 points. The same signals keep half as much per trade at night.

If a night strategy stays positive only with day-session slippage, it depends on fills the night market may not give you. Rerun it with double the night slippage. If the net average trade turns negative at double slippage, treat the night version as cost-sensitive and keep testing it before you trade it.

How do gaps between sessions change a backtest?

A gap at a session open can fill your stops and targets at prices far from where you placed them, and the data you choose decides whether your backtest sees that gap correctly. An overnight gap is a jump between the last price of one session and the first price of the next. Mini-TAIEX has two breaks a day, from 05:00 to 08:45 and from 13:45 to 15:00, so it has two places where gaps form.

With day-only bars, the whole night is hidden. The night's move appears as one jump at 08:45. A stop that would have filled at 21:40 in the full market shows up in the backtest as a fill at the next morning's open. The fill price is different, and so is the time. With full-session bars, the 05:00 to 08:45 break is still there, but the gap is usually smaller, because the night session has already absorbed most of the news.

Hypothetical example

You hold a long position at the day close of 20,000 with a stop at 19,950. During the night the market falls and trades 19,948 at 21:40. On full-session data, the stop fills near 19,948, a loss of about 52 points. On day-only data, the next bar opens at 19,880 at 08:45, and the stop fills there, a loss of 120 points. Same rule, same market, two different results.

The same stop in the hypothetical example: day-only data hides the night move and fills at the morning gap, while full-session data fills it during the night.

Decide early whether your strategy holds positions through the breaks. A strategy that always exits before 13:45 and before 05:00 never carries a gap. A strategy that holds overnight carries every gap, and its worst trades will often be gap trades. The article on day-only versus full-session bars in night-session backtests covers this data choice for other futures markets as well.

How do you test day and night as separate segments?

Test the day session, the night session and the full session as three separate backtests, and combine them only when each part holds up on its own. A session segment is a fixed block of trading hours that you backtest as if it were its own market. Day and night Mini-TAIEX trading differ in who trades, how much volume there is and what drives the price. A rule that works in one segment often fails in the other, and a full-session result can hide that. Filing the day, night and full-session runs of one rule in the strategy databank of AlphaPrime (Valynox) lets you sort them and revalidate each segment on its own.

Segment	Session hours	Slippage assumption	What it tells you
Day only	08:45 – 13:45	Day setting	Whether the rule works in the liquid session
Night only (夜盤策略)	15:00 – 05:00	Wider night setting	Whether the rule still pays after thinner fills
Full session	Both, with breaks	Per segment	Whether combining helps, and how gaps affect positions held across the breaks

A workable order of steps:

Check the history. Your data has no night bars from before TAIFEX added the after-hours session in 2017. Start full-session and night tests at the first date your data contains night bars. Otherwise the early years test a different market.
Fix the session definitions. Set the bar timestamp convention, the midnight wrap and the trading-day assignment once, and use them in every run.
Run day only with day-session costs.
Run night only with the wider night slippage.
Validate each segment separately. Use out-of-sample data and walk-forward analysis, so each segment has to prove itself on data it was not tuned on.
Combine last. Run the full session only for strategies that passed in their own segment. Then check the combined drawdown and the trades held across a break.

Count every variant you try. If you test day, night and full sessions with several parameter sets and keep the best one, you have run many trials. The best result among many trials is inflated by luck. Harvey and Liu (2015) describe how to account for multiple testing when you judge a backtest.

Test the two sessions as separate segments with their own costs, then test the combination last.

Do you need live night-session quotes for research?

No. Strategy research runs on historical bars, so a live feed (小台指夜盤即時) adds nothing to a backtest. A backtest replays past bars and decides at each bar's close. What it needs is clean history: complete night bars, correct timestamps, correct contract rolls and no missing sessions. A live quote screen shows the price now. It cannot tell you whether a rule worked over the past several years.

Live data matters later, when you run a finished strategy on a trading platform. Even then, compare the platform's night-session bars with the bars you researched on. Look at the first bar after 15:00, the bars around midnight and the last bar before 05:00. If the platform builds those bars differently, the signals can differ.

What this means for AlphaPrime users: AlphaPrime runs on your own historical bar data, including TAIEX futures. You can search and validate day-only and night-only Mini-TAIEX segments separately. Each run uses its own trading session setting and its own fees and slippage, and AlphaPrime applies next-bar execution in both. For the validation step, AlphaPrime's strategy validation includes walk-forward analysis and custom thresholds. AlphaPrime's backtesting lets you set the data resolution, trading sessions, fees and slippage before a search starts.

Frequently asked questions

What time does the mini-TAIEX night session open and close?

The Mini-TAIEX after-hours session opens at 15:00 and closes at 05:00 the next morning, Taiwan time. The regular day session runs from 08:45 to 13:45. Taiwan does not use daylight saving time, so the local hours do not shift during the year. Check the TAIFEX contract page for any change.

Does the night session count toward the next trading day?

Yes. TAIFEX assigns trades in the after-hours session to the trading day of the next regular session. A Friday night session, which ends at 05:00 on Saturday, belongs to Monday's trading day. Your backtest should group bars the same way, or daily statistics will split each night in two.

Is the Mini-TAIEX night session less liquid than the day session?

Usually yes. Fewer contracts trade at night, so the market is thinner and fills tend to be worse, especially in the early morning hours. Activity tends to pick up around US data releases and the US stock market open. Use a wider slippage assumption for night trades than for day trades.

How do I backtest a night-session-only strategy (夜盤策略)?

Set the session filter to 15:00–05:00 and make sure it wraps correctly past midnight. Start the history at the first date your data contains night bars, since TAIFEX added the after-hours session in 2017, and use night-specific slippage. Then validate the strategy on out-of-sample data before combining it with any day-session rules.

Do I need real-time night-session quotes to research a strategy?

No. Research uses historical bars, and what matters is that those bars are complete and correctly timestamped. Live quotes matter only when you run the finished strategy on a trading platform.

References

Harvey, C. R., & Liu, Y. (2015). Backtesting. The Journal of Portfolio Management, 42(1), 13–28.