By the time Tokyo’s cash market opens at 9:00 a.m., most of the night’s news is already in the price. Nikkei 225 futures trade nearly around the clock across Osaka, Singapore, and Chicago, and the cash index typically opens close to the level the overnight futures imply once you adjust for fair-value basis. If you can read three things — the overnight futures level, the fair-value basis, and the quarterly roll calendar — almost every “mysterious” opening gap in Tokyo turns into simple arithmetic.
Interest in this plumbing spikes in weeks like mid-September 2026. The Bank of Japan’s policy meeting lands on September 17–18, just days away; the Nikkei Volatility Index printed 30.68 on September 15, around its 84th percentile of the prior three-plus years; and the official Nikkei 225 close was 63,484.10 on September 15, little changed on the session, while the S&P 500 fell 0.45% at that evening’s U.S. close. Each of those inputs reaches the Tokyo open through the same relay of futures contracts. This guide explains the relay, and it will read the same in any week.
Three venues, one index
The Nikkei 225 — the Nikkei heikin (“Nikkei average”), a price-weighted index calculated by Nikkei Inc. — is computed during Tokyo hours only. Futures on it trade on three exchanges in three time zones:
| Venue | Contracts | Multiplier per index point | Currency | Typical role |
|---|---|---|---|---|
| Osaka Exchange (OSE, part of JPX) | Large / Mini / Micro | ¥1,000 / ¥100 / ¥10 | JPY | Home market; deepest book; contracts settle to the official SQ |
| Singapore Exchange (SGX) | Nikkei 225 futures | ¥500 (a US$5 contract also lists) | JPY / USD | Asian-hours alternative; popular with regional books for margin offsets |
| CME (Chicago) | Yen-denominated and dollar-denominated Nikkei | ¥500 / US$5 | JPY / USD | Price discovery through U.S. hours |
Three practical consequences follow. First, the contracts are fungible in price but not in size: a one-point move is worth ¥1,000 on the OSE large contract but ¥500 on the SGX and CME yen contracts, so quoted volumes across venues are not directly comparable. Second, the CME dollar contract is a quanto — profit and loss accrue in dollars at a fixed $5 per point with no currency conversion — so its price can sit a few points away from the yen contracts, and that gap widens when the equity–FX correlation is unstable. Third, arbitrage desks keep every open venue within a handful of points of the others, which is why practitioners speak of “the future” as a single global price.
The 24-hour relay into the Tokyo open
The OSE day session runs 8:45–15:45 JST. The night session — naito sesshon — reopens at 17:00 and runs to 6:00 the next morning, spanning the entire European day and the full U.S. cash session. SGX covers Asian and overnight hours as well (historically opening a few minutes before Osaka in the morning), and CME’s Globex session runs nearly continuously from Sunday evening to Friday afternoon Chicago time. Japan risk is therefore priced almost continuously; only the cash index sleeps.
That last point is the single most common reading error. The cash Nikkei level on your screen at 7:00 a.m. Tokyo time is yesterday’s close — a stale number. The live price of Japanese equity risk is the futures print. When news breaks after Tokyo’s close, the night session and CME absorb it in real time. At 9:00 there is no single opening auction for the index itself: each of the 225 constituent stocks holds its own opening itayose auction, in which the resting orders in that name are matched at one clearing price, and not every stock necessarily establishes its opening price at the same moment. The index’s opening value is calculated from the constituents’ resulting opening prices, with substituted quotes for names that have yet to trade. The practical effect is that the cash index typically converges quickly toward the futures-implied, basis-adjusted level — a strong tendency, not point-for-point matching of the raw overnight print. A “gap up at the open” is rarely new information at 9:00; it is the cash index catching up to what futures already did while Tokyo slept.
The same discipline applies to foreign inputs. At the September 15, 2026 U.S. close, the S&P 500 had fallen 0.45% to 7,585.73 — yet in the overnight session that followed, S&P futures traded around 7,660, up roughly half a percent. A trader reading only the U.S. cash close would have positioned for a weak Tokyo lead; the live futures said the opposite. Rule of thumb: before the Tokyo open, check the most recently traded prices — Nikkei futures and U.S. index futures — and distrust any “close” more than a few hours old.
Basis: what the futures–cash gap actually means
The basis — saya in Japanese trading slang — is the difference between the futures price and the cash index. It is mostly carry, not sentiment. Cost-of-carry fair value is:
Fair futures price = Spot × (1 + (funding rate − dividend yield) × time to expiry)
A future should trade above cash when funding costs exceed the dividends you forgo by holding the future instead of the stocks, and below cash when dividends exceed funding.
Worked example: fair value with September 2026 figures
- Spot. The official Nikkei 225 close was 63,484.10 on September 15, 2026 (Nikkei Inc.).
- Funding rate. The proper input is a short-term yen funding or repo rate matching the contract’s remaining life — think three-month TONA-linked or GC repo money. Lacking a term funding quote here, we fall back on the 2-year JGB yield, 1.861% as of September 15, 2026 (MOF), as a rough and admittedly imperfect proxy: a two-year government yield can sit meaningfully away from three-month funding, and that mismatch alone can distort computed fair value. Call funding roughly 1.9%, with that caveat attached.
- Dividend yield. Historically around 2% for the index — an approximation, since exact forecast dividends require a constituent-level model.
- Time. In mid-September, just after the September expiry, the front quarterly contract is December: about three months, so t ≈ 0.25.
- Compute. (1.9% − 2.0%) × 0.25 ≈ −0.03% of spot, or roughly −15 to −20 index points. Fair value ≈ 63,470 — essentially flat to cash.
Reading: a December future printing 15–20 points under cash in that environment is not bearish positioning; it is carry math. The instructive contrast is with the zero-rate era. With funding near 0% and dividends around 2%, a three-month contract carried a structural discount of roughly half a percent — over 300 points at these index levels — which routinely fooled newcomers into believing the futures market “predicted” a decline every quarter. As Japanese short rates rose toward the dividend yield through the mid-2020s, that structural discount collapsed toward zero. When the basis moves, first ask what rates and dividends did; only what remains is sentiment or flow.
Two refinements. Japanese dividends are lumpy: record dates cluster at fiscal half-year and year ends, in late September and late March, so the effective dividend yield is concentrated around those dates and the basis dips mechanically just before them. And never compute “basis” across time zones — subtracting yesterday’s Tokyo cash close from this morning’s CME print measures overnight news, not carry.
Rollover: the quarterly migration — and why it is not SQ
OSE large contracts expire quarterly, in March, June, September, and December; minis add monthly expiries. Each expiring contract settles to the SQ — tokubetsu seisan shisū, the Special Quotation — calculated from the opening auction prices of all 225 constituents on the second Friday of the contract month. The SQ is a settlement event, with its own well-known open-auction dynamics, and we cover it separately. The rollover is what positions do before that event, and conflating the two is a common mistake.
Rolling means closing the expiring contract and re-establishing the position in the next quarterly, usually via a calendar spread executed as a single trade. Mechanics worth knowing:
- Timing. Open interest migrates over roughly the week before the second Friday. Once the bulk has moved, quotes in the expiring contract thin out quickly — do not plan size in it during its final days unless you intend to hold into settlement.
- Roll cost. The calendar spread is the deferred price minus the near price — Fdeferred − Fnear — and its fair value is the carry between the two expiries: (funding − dividend yield) applied over the roughly three months separating the contract months, at the prevailing index level. It is not the deferred contract’s entire fair-value basis versus spot; the near leg’s remaining basis to expiry nets out of the spread. A spread persistently richer or cheaper than that inter-expiry carry signals one-sided positioning pressure — hedgers or index-replicating longs all needing the same side of the roll.
- Volatility regime. Execution is costlier when volatility is elevated — for scale, the Nikkei Volatility Index printed 30.68 on September 15, 2026, around its 84th percentile of the prior three-plus years — so spreads widen earlier in roll week.
- Charting trap. A continuous futures chart that simply splices contracts shows the roll gap as a phantom price move. Use back-adjusted series, or reconcile against the cash index around each expiry.
FAQ
Why do OSE, SGX, and CME futures show slightly different prices at the same moment?
Multiplier and currency differences, exchange latency, and the quanto adjustment on the CME dollar contract each contribute a few points. Arbitrage keeps the venues tightly linked whenever two are open, so persistent gaps beyond a handful of points are rare and short-lived. For most purposes, treat the most liquid open venue as the reference price.
Is a futures discount to the cash Nikkei bearish?
Usually not. Compare the observed basis to computed fair value, not to zero. In the zero-rate era the index future carried a structural discount purely from dividends exceeding funding; with short rates near the dividend yield (the 2-year JGB — itself only a rough stand-in for short-term funding — stood at 1.861% as of September 15, 2026), fair basis sits close to flat. Only a basis meaningfully away from fair value — after accounting for dividend seasonality — says anything about positioning.
Which number should I check before the Tokyo open?
The latest traded Nikkei future — CME or the OSE night session — plus live U.S. index futures. The prior cash close, whether Tokyo’s or New York’s, is stale by definition. The constituent opening auctions at 9:00 will usually pull the cash index toward the futures-implied level once fair-value basis is accounted for, so the futures print at 8:55 is the best single anchor for the cash open — a convergence target, not a guarantee of a point-for-point match.
Sources
- JPX / Osaka Exchange statistics and derivatives specifications — www.jpx.co.jp/english
- Nikkei Indexes (official Nikkei 225 and Nikkei Volatility Index data) — indexes.nikkei.co.jp/en
- Bank of Japan (policy rate and money market data) — www.boj.or.jp/en
- Ministry of Finance Japan (JGB yield data) — www.mof.go.jp/english
Disclaimer: This is an information and analysis publication, not investment advice. See our Methodology for data sources, standards, and our corrections policy.