With the Kanazawa Marathon 2026 (Sunday, October 25), the Yokohama Marathon 2026 (same day) and the Kobe Marathon 2026 (Sunday, November 15) approaching, how is hotel inventory in the host cities moving toward race eve? When MetroEngines Research’s night-by-night remaining-room data is aligned at the same lead time (LT45 = 45 days before the stay date), Kanazawa City’s occupancy (OTA-listed inventory basis, estimated) for the race-eve night of Saturday, October 24 stands at 91.1%, 10.6pt above the previous Saturday’s 80.5% — and the gap against the following Saturday (76.4%) and race-day night (78.5%) is wider still. The same shape appears in Yokohama’s Naka Ward and Kobe’s Chuo Ward, but how fast rooms fill differs by city. This article quantifies race-eve demand for the three marathons at an identical lead time, back-calculates the rooms required relative to room stock, and sets out in a reverse-planning table when operators should decide on rate changes and minimum length of stay (MLOS).

The Three Races at a Glance — Pre-Race Registration Creates the “Race-Eve Stay”

First, let us align the schedules of the three races, the premise for the comparison, using official materials. The Kanazawa Marathon 2026 starts at 8:30 on Sunday, October 25 (in front of Shiinoki Cultural Complex). Runner registration takes place at the Ishikawa Ongakudo Exchange Hall in front of Kanazawa Station on Friday, October 23 (two days before) from 13:30 to 20:30 and on Saturday, October 24 (the day before) from 9:30 to 19:30. There is no option to shift pre-race registration to race morning, so participants from outside the prefecture structurally have to stay the night before or two nights before. In addition, a free drone show (two performances, at 18:30 and 20:15) is scheduled at Kanazawa Castle Park on the evening of October 24, turning race eve itself into an event.

The Yokohama Marathon 2026 takes place on the same day, Sunday, October 25, and the City of Yokohama’s public bulletin lists it as running from 8:30 to 15:00. The field is 30,200 runners, 1,000 more than the previous year and above 30,000 for the first time. The Kobe Marathon 2026 is on Sunday, November 15; against a full-marathon capacity of 20,000, a total of 61,173 people applied (58,340 domestic and 2,833 overseas), so places were allocated by lottery. Of domestic applicants, 37,899 live outside Hyogo Prefecture — 65% of the domestic total.

Source: Compiled by the HotelBank Editorial Team from each race’s official materials

Why Compare at the Same Lead Time?

Race-eve demand cannot be measured by simply lining up remaining rooms on the day of observation. As of October 11, October 17 is at LT6, October 24 at LT13 and October 31 at LT20 — with different numbers of days until check-in, booking progress also differs. This article therefore extracts the LT45 observation, which exists for every stay date being compared, lines them up, and then overlays the movement from LT45 to LT14 to read “when demand settled”. Observations earlier than LT45 can include states before listings are fully loaded and may not reflect booking progress, so they are not used. Reading this alongside Izumo Ekiden 2026: race-eve OCC 98.8% at 45 days out (published October 2026), which measured the eve of an ekiden relay with the same yardstick, also shows how the shape of race-eve demand differs by type of event.

Kanazawa City — Race-Eve Saturday at 91.1% 45 Days Out, Race-Day Night at 78.5%

In Kanazawa City (137 properties / 9,815 rooms covered, major booking-site basis), race-eve Saturday, October 24 stood at 91.1% occupancy (OTA-listed inventory basis, estimated) at LT45, with 873 rooms remaining. The previous Saturday, October 17, was at 80.5% (1,912 rooms remaining) and the following Saturday, October 31, at 76.4% (2,325 rooms remaining), so race eve runs 10.6–14.7pt above a normal Saturday. Race-day night, Sunday, October 25, on the other hand, stays at 78.5% at LT45, below even the previous Saturday. The structure — only race eve jumps, race-day night does not — comes through clearly.

Rates move in the same direction. The average listed rate of inventory on sale for October 24 is ¥71,100, 1.6x the previous Saturday’s ¥44,500 and more than double race-day night’s ¥32,800. However, this average reflects the price of “remaining inventory” and excludes sold-out properties, so it should be discounted for the skew toward higher price bands on race eve, when inventory is tight. That race-eve rates had already started to separate from a normal Saturday by the summer was confirmed at property level in Japan’s autumn marathons 2026 and hotel supply and demand.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research (LT45 = observation 45 days before each stay date)

What deserves attention is the shape of the curve. October 24 went from 91.1% at LT45 to 92.7% at LT30 and 94.3% at LT14 — a build of only 3.2pt over 45 days. In contrast, the previous Saturday, October 17, grew 9.4pt, from 80.5% at LT45 to 89.9% at LT14. In other words, race-eve demand had largely settled before LT45, and from then on the market entered a phase in which normal weekend demand fills the remaining inventory. Friday, October 23 (two days before) stood at 76.6%, almost level with the previous Friday (October 16) at 77.8%, so a two-nights-before stay does not show up materially in room nights.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research (Kanazawa City, 137 properties / 9,815 rooms covered, LT45 to LT14; LTs not yet reached are left blank)

Ishikawa Prefecture Overall — Business Hotels Absorb the Demand, Ryokan Stay Flat

So which categories are absorbing race-eve demand? Splitting Ishikawa Prefecture (269 properties / 14,840 rooms covered) by category, business hotels (69 properties / 9,212 rooms) reached 87.1% for October 24 at LT45, up 10.6pt from the previous Saturday’s 76.5%, and their average listed rate rose from ¥32,000 to ¥49,800. Ryokan (88 properties / 3,575 rooms), by contrast, were almost flat at 71.6% to 72.5%; the marathon has had little effect on ryokan inventory, including the Kaga Onsen area and Wakura Onsen. Travelling runners choose business hotels around the station and in the city center, while onsen ryokan appear to be moving on normal autumn-weekend demand.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research (Ishikawa Prefecture, at LT45; properties covered as in the table above)

Yokohama — Naka Ward +14.1pt vs. the Previous Week, but Fills “Late”

For Yokohama, we look at Naka Ward (50 properties / 9,927 rooms covered), home to the Minatomirai and Kannai venues, and Nishi Ward (10 properties / 2,284 rooms). Naka Ward’s October 24 stood at 79.4% at LT45, 14.1pt above the previous Saturday’s 65.3% and 26.0pt above the following Saturday’s 53.4%. Race-day night, October 25, was at 64.1% — again, only race eve jumps. Nishi Ward moves in the same direction, from 74.0% to 79.4%, but with only 10 properties the size of the gap should be treated as indicative.

What differs from Kanazawa is how bookings build afterwards. Naka Ward’s October 24 grew from 79.4% at LT45 to 84.7% at LT30, 91.3% at LT20 and 94.3% at LT14 — a 14.9pt build over 45 days. That is more than four times Kanazawa’s 3.2pt, meaning Yokohama’s race eve is still far from full at LT45. Participants from the Tokyo metropolitan area can travel on race day, and there appears to be a thick layer of people who decide late on a race-eve stay. If Kanazawa is framed as “destination type, settles early” and Yokohama as “commuter type, builds late”, the timing of rate changes discussed below naturally differs as well.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research (Yokohama Naka Ward, 50 properties / 9,927 rooms covered, LT45 to LT14)

Kobe — Chuo Ward at 94.2% 45 Days Out, Invisible at Prefecture Level

The Kobe Marathon is three weeks later, on Sunday, November 15, so race eve is Saturday, November 14. Chuo Ward (37 properties / 5,436 rooms covered), which hosts both the start (in front of City Hall) and the finish (Harborland), stood at 94.2% for November 14 at LT45 with 318 rooms remaining, 8.8pt above the previous Saturday, November 7, at 85.4%. Race-day night, November 15, was at 78.5% — the same structure as Kanazawa and Yokohama, where only race eve jumps. The average listed rate also rose 1.38x, from ¥44,800 (11/7) to ¥61,900 (11/14).

Kobe’s following Saturday, November 21, however, is at 91.0%, close to race eve. It is the first day of a three-day weekend including Monday, November 23 (Labor Thanksgiving Day) and coincides with peak autumn foliage, so demand unrelated to the race is layered on top. For Hyogo Prefecture as a whole (573 properties / 19,966 rooms), the figures are 82.9% for November 7, 84.3% for November 14 and 89.2% for November 21 — race eve barely stands out. Kita Ward (40 properties / 1,460 rooms), home to Arima Onsen, is at 90.1% for November 14 and 96.4% for November 21, higher on the long-weekend side, with autumn onsen demand taking the lead. At prefecture level, the marathon’s race-eve demand is buried within foliage and long-weekend demand; its outline only appears once you narrow down to the ward hosting the race.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research (11/13 is not among the prefecture-level comparison dates, hence “—”)

Race-Eve Premiums Across the Three Cities

Summarizing all of this with the same yardstick, the gap between race eve and the previous Saturday is +10.6pt in Kanazawa City, +14.1pt in Yokohama’s Naka Ward and +8.8pt in Kobe’s Chuo Ward, while the gap between race eve and race-day night is +12.6pt, +15.3pt and +15.7pt respectively. The average listed rate of inventory on sale versus the previous Saturday is 1.6x in Kanazawa City, 1.25x in Yokohama’s Naka Ward and 1.38x in Kobe’s Chuo Ward — relative to the occupancy differences, Kanazawa’s price response stands out. This is likely because a race of around 15,000 runners lands on Kanazawa City’s 9,815 rooms, whereas Yokohama’s Naka and Nishi Wards have 12,211 rooms combined, so the same race-eve demand puts a different degree of pressure on supply.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research (at LT45 for each stay date; properties and rooms covered as in each table)

Back-Calculating Required Rooms Against Room Stock

How large is race-eve demand relative to stock? Let us back-calculate from the published breakdown of participants’ place of residence. Of the roughly 15,600 runners in the Kanazawa Marathon 2025, 8,704 came from outside Ishikawa (6,195 from distant prefectures, 1,565 from neighboring prefectures and 944 from overseas), and the companion ratio for out-of-prefecture runners is 0.6 (estimate by Kanazawa Gakuin University). Setting a race-eve stay rate and guests per room to derive required rooms gives the following range against Kanazawa City’s 9,815 rooms.

Source: Kanazawa Gakuin University “On the Economic Ripple Effect of KM2025” (March 21, 2026); HotelBank Editorial Team estimates based on MetroEngines Research

Even the mid scenario equals 60% of Kanazawa City’s rooms and 40% of Ishikawa Prefecture’s. Against this, the actual race-eve uplift observed at LT45 (the difference in remaining rooms versus the previous Saturday) is only 1,039 rooms in Kanazawa City and 1,353 rooms in Ishikawa Prefecture. There are three possible interpretations of the gap. First, tourists who would normally visit Kanazawa on a regular Saturday may shift to another weekend, so marathon demand displaces normal demand. Second, accommodation in neighboring prefectures such as Toyama and Fukui may be absorbing demand. Third, some race-eve stays — with friends or relatives, or sleeping in cars — may not appear in accommodation data. Either way, it is reasonable to conclude that race-eve demand on a scale Kanazawa City alone cannot absorb is spilling into the surrounding area, and Toyama City, 30 minutes away by Hokuriku Shinkansen, and Fukui City, also on the Shinkansen line, have room to capture it depending on how they design rates and inventory. A side-by-side of room stock within 5km of the start for 12 races is shown in 12 marathons from October to February: rooms measured within 5km of the start.

For Yokohama, the organizing committee’s activity report does not show a breakdown by place of residence, so the out-of-prefecture share itself has to be assumed. Taking the 28,770 starters in the 2025 race (all categories) and setting the out-of-prefecture share and race-eve stay rate in a 30–50% range, the rooms required equal 16–45% of the 12,211 rooms in Naka and Nishi Wards combined. Many participants come as day-trippers from the Tokyo area, so race-eve demand is unlikely to exceed half of supply. That Naka Ward and Nishi Ward are both at 79.4% at LT45, less congested than Kanazawa, also reflects this spare capacity.

Source: Yokohama Marathon Organizing Committee “Yokohama Marathon 2025 Activity Report” (January 22, 2026); HotelBank Editorial Team estimates based on MetroEngines Research

Kobe looks different. Of the 20,627 starters in the Kobe Marathon 2025, 12,609 came from outside Hyogo (10,386 domestic and 2,223 overseas), and overseas runners visit in groups averaging 2.9 people including the runner (Hyogo Regional Economic Indicators Study Group). Setting the race-eve stay rate for domestic out-of-prefecture runners at 30–50% and assuming all overseas runners stay on race eve, rooms required reach 85–139% of Chuo Ward’s 5,436 rooms, exceeding supply. That Chuo Ward is already 94.2% full at LT45 is consistent with this scale, and the catchment for race-eve stays extends to Nada Ward, Hyogo Ward, the Shin-Kobe area and even central Osaka. Given that overseas runners’ average accommodation spend was about ¥85,000 per person (2025 race) — a level that implies multi-night stays — this is a market where stay design beyond a single race-eve night pays off.

Source: Hyogo Regional Economic Indicators Study Group “Economic Ripple Effect of the Kobe Marathon 2025” (March 3, 2026); Kobe Marathon Executive Committee; HotelBank Editorial Team estimates based on MetroEngines Research

Checking Other Factors — Not Attributing the Uplift to a Single Cause

Before attributing the race-eve uplift solely to the marathon, we check other factors coinciding with the same weekend. In Kanazawa, the drone show on the evening of October 24 is held in conjunction with the race; it thickens race-eve guests’ stays rather than being an independent draw. The Kenrokuen illumination “Kanazawa Castle and Kenrokuen Four Seasons: Autumn” is scheduled for weekends in November in 2026 and does not fall on October 24–25. Within the range the editorial team checked, there is no major concert on the same days either, and since autumn leaves typically only begin to turn from mid-October, the lift from autumn tourism demand is likely similar to a normal Saturday.

In Yokohama, no major performances at K-Arena, Yokohama Stadium or similar venues could be confirmed for October 24, and Naka Ward’s relatively low 53.4% for the following October 31 indicates that the Halloween weekend is not drawing particular accommodation demand to Kannai and Minatomirai. Kobe’s November 14 is a Saturday in foliage season, and as Kita Ward (Arima Onsen) being at 90.1% also suggests, Chuo Ward’s 94.2% includes some autumn tourism demand. Because it is impossible to separate how much of the 8.8pt gap versus November 7 comes from the marathon, this article limits itself to the statement that it is “at least 8.8pt higher than the previous week”.

Implications for Operators — Rate Changes and MLOS Planned Backward from When Race-Eve Demand Settles

What the three cities’ curves show is that the lead time at which race-eve demand firms up differs by city. In Kanazawa, race eve was already above 90% at LT45 and built only 3.2pt afterwards, so it is fair to say demand settled before LT45. In Yokohama, Naka Ward built 14.9pt from LT45 to LT14, leaving room to move rates as late as LT30–20. Kobe, at 94.2% at LT45 and 96.4% at LT35, is an early-settling type like Kanazawa. Planning race-eve sales backward on this basis gives the following table.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research

Turning to race-day night, Kanazawa City’s Sunday, October 25 is at 78.5% at LT45, Yokohama’s Naka Ward at 64.1% and Kobe’s Chuo Ward at 78.5% — all more than 10pt below race eve. The behavior of participants who head home the same day after finishing a full marathon is hard to change, but among race-eve guests from distant prefectures and overseas, an offer to stay one more night after finishing is likely to be well received. Given that out-of-prefecture runners in Kanazawa bring 0.6 companions and overseas runners 1.9 companions (both for the 2025 race), an extension plan combined with race-day sightseeing for companions is a revenue opportunity that bundles race eve’s high rate and race-day night’s empty rooms into a single stay.

Summary

Viewed through a same-LT comparison, race-eve demand for all three races shared a common shape: “only race eve jumps; race-day night does not”. Kanazawa City was at 91.1% at LT45 (+10.6pt vs. the previous Saturday), Yokohama’s Naka Ward at 79.4% (+14.1pt) and Kobe’s Chuo Ward at 94.2% (+8.8pt), with Kanazawa showing the largest price response. The speed of settlement differs, however: Kanazawa and Kobe are destination-type markets that firm up before LT45, while Yokohama is a commuter-type market that keeps building after LT45. Rooms required, back-calculated from participants’ place of residence, equal 40–70% of rooms in Kanazawa City and exceed supply in Kobe’s Chuo Ward, with the catchment spreading to neighboring prefectures and wards. For operators, the practical path to turning race-eve demand into revenue is to place race-eve rate-change and MLOS decisions at LT60–45 (destination type) or LT30–20 (commuter type), and to bundle race-day-night extensions into offers for race-eve guests.

Frequently Asked Questions

Q. How full are Kanazawa City hotels on the eve of the Kanazawa Marathon 2026?
A. For race eve, Saturday, October 24, 2026, occupancy (OTA-listed inventory basis, estimated) was 91.1% at 45 days out (LT45), with 873 rooms remaining. At LT14 it was 94.3% (Kanazawa City, 137 properties / 9,815 rooms, as of October 11, 2026). This is an estimate based on OTA sales inventory and differs from properties’ actual overall occupancy.

Q. Does race-day night fill the same way as race eve?
A. Compared at the same LT45, race-day night is at 78.5% in Kanazawa City, 64.1% in Yokohama’s Naka Ward and 78.5% in Kobe’s Chuo Ward, 12.6–15.7pt below race eve. In all three cities only race eve jumps, while race-day night moves close to a normal Sunday (observations as of October 11, 2026).

Q. Why is it necessary to compare at the same lead time?
A. If remaining rooms are lined up by observation date, the number of days until check-in differs for each stay date, so booking progress is not aligned. This article extracts the LT45 observation common to every stay date being compared and overlays the movement from LT45 to LT14 to read when demand firmed up.

Q. When should race-eve rate changes and MLOS (minimum length of stay) be decided?
A. In destination-type markets such as Kanazawa and Kobe, race eve is already above 90% at LT45, so LT60–45 is a reasonable anchor for decisions. In commuter-type markets such as Yokohama, bookings keep building after LT45, leaving room to move in two steps at LT30–20. Both are guidelines based on observations for the 2026 races and may vary by race and year.