Where the 2027 Holiday Data Comes From and How the Optimizer Works

The short answer: 2027, 10 countries, 118 days

The optimizer covers one year only: 2027. It holds 10 countries and 118 national public holidays, counted country by country from the data built into the tool. The counts add up as follows: 11 plus 8 is 19, plus 9 is 28, plus 11 is 39, plus 17 is 56, plus 18 is 74, plus 6 is 80, plus 10 is 90, plus 11 is 101, plus 17 is 118.

Each holiday entry carries a date, an English name, and a local name. The data is built into the page itself: the tool page states that it runs entirely in the browser, that the holiday data is built into the page for 2027, and that nothing is fetched or uploaded. There is no account and no server-side lookup behind the plan.

One coverage limit belongs up front, not in a footnote. Mainland China is not in version 1 of the tool. The tool notes say it will be added after the State Council publishes the 2027 holiday schedule, so anyone planning around mainland China holidays cannot use this tool for that yet.

Where the 118 days come from

The source note inside the tool says the data comes from Nager.Date and was checked against official calendars on October 26, 2026. That note is the single stated provenance for the whole set, and it is the reason this guide can describe the data as checked in general while still grading the verification country by country below. A general checking note is not the same as opening each official calendar again, and this guide does not pretend otherwise.

Because nothing is fetched when the page loads, the data in the page is a snapshot, not a live feed. If an official calendar were corrected after the snapshot was built, the page would not pick up the correction on its own. That is the practical consequence of browser-local data: it is private and fast, and it is frozen at build time. Rechecking the official calendar before submitting leave is therefore part of using the tool well, not an optional extra.

Verification, graded honestly: not every country was checked the same way

This round of verification fell into three tiers. Two countries were checked against their official publications in this round. One country matched an official wording in substance, but its original source page was not directly opened. Seven countries were not reopened against their official originals in this round and rest on the tool checking note. The table states the tier for every country, and the paragraphs below it give the detail.

United Kingdom: verified date by date

The United Kingdom set was checked against the GOV.UK bank holidays page, which publishes separate tables for England and Wales, Scotland, and Northern Ireland. The England and Wales table for 2027 lists January 1, March 26, March 29, May 3, May 31, August 30, December 27, and December 28, and all 8 match the tool entries one by one. The page address is https://www.gov.uk/bank-holidays . Only the England and Wales table is the tool set; the Scotland and Northern Ireland tables are different lists and are not used.

Singapore: verified, with one difference you should know about

Singapore was checked against the Ministry of Manpower annual gazetted public holiday table, published as a press release. Its 11 gazetted holidays match the tool 11 in substance, but one listing differs in presentation. The Ministry table names Sunday, February 7 as the day itself and notes Monday, February 8 as the day in lieu. The tool lists only Monday, February 8, and does not list February 7. The practical result is the same Monday off, but anyone comparing the two lists side by side will see February 7 in one and not the other. The press release address is https://www.mom.gov.sg/newsroom/press-releases/2026/0618-public-holidays-for-2027 .

United States: wording consistent, original page not opened

The United States set rests on two elements: the federal holiday statute, 5 U.S.C. 6103, and the annual observed dates published by the Office of Personnel Management, which shift a holiday that falls on a weekend to a nearby weekday. A retrieved page restating the Office of Personnel Management wording was compared with the tool 11 entries one by one, and all 11 agreed, including the three observed dates rather than the calendar dates: June 18 as the observed Juneteenth day, July 5 as the observed Independence Day, and December 24 as the observed Christmas Day. But the original opm.gov page itself was not directly opened in this round, so the honest wording is that the tool wording is consistent with the official wording — not that the original page was verified. No official address is claimed for a page that was not opened.

The other seven: not reopened in this round

Germany, France, Japan, South Korea, Australia, Canada, and Hong Kong were not each opened against their official originals in this round. For these seven, this guide reports only what the tool itself states: the data comes from Nager.Date and was checked against official calendars on October 26, 2026. Their scope rules are described in the next section from the tool notes, and no specific official address or agency name is invented for them. Before citing a specific date from any of these seven in a leave request, check the official calendar of that country once more.

Scope rules that change the answer

A holiday count is meaningless until the set behind it is stated, because several countries publish more than one list. The tool makes a choice for each country, and that choice changes which bridges exist:

• United Kingdom: the England and Wales set only. Scotland and Northern Ireland have their own bank holiday tables, which the tool does not use.
• Canada: the federal set only. Provincial holidays are not included.
• Germany and France: no make-up rule. A holiday that falls on a weekend is simply lost; no weekday is substituted for it.
• Japan: substitute holidays are included. One example in the tool data is March 22, 2027, the substitute day for the Vernal Equinox holiday.
• South Korea: alternative public holidays are included. One example in the tool data is December 27, 2027, the substitute day for Christmas.
• Australia: the national set holds 6 days, and the Anzac Day make-up day is not in it. Anzac Day itself falls on Sunday, April 25, 2027, and the Monday make-up on April 26 applies in most states but not in Victoria, South Australia, or Tasmania. The tool tells the user to add April 26 as a company day where it applies, which is the correct workaround inside a national set.
• Hong Kong: 17 days, including the first three days of the Lunar New Year, the Birthday of the Buddha, the Dragon Boat Festival, the day after the Mid-Autumn Festival, and Chung Yeung Festival.

If your own holidays follow a different set — a Scottish calendar, a Canadian provincial calendar, or an Australian state calendar — the tool plan will miss bridges that exist for you and may show bridges that do not. The company-day feature exists partly for this: a day your calendar observes but the national set omits can be added by hand, as the Anzac Day example shows.

From holidays to a plan: non-working days and bridges

The plan starts from five inputs. The country selects one of the 10 holiday sets. The leave balance is entered as a number of days, from 1 to 30. The weekend is either Saturday-Sunday, which is the default, or Friday-Saturday. The strategy is one of the three described in the next section. And company days are extra days off added by clicking any working day on the calendar; public holidays and weekend days cannot be clicked, and a Clear company days button removes them all.

From those inputs the tool builds one set: non-working days, defined as public holidays plus company days plus weekends. Everything else in the year is a working day. The tool then cuts the 2027 calendar into alternating runs: stretches of consecutive non-working days, separated by gaps of consecutive working days.

Only a gap that sits between two non-working runs can become a bridge. Take the whole gap as leave, and the break runs from the start of the left run through the gap to the end of the right run. The efficiency of a bridge is its ratio: consecutive days off divided by leave days used, shown to one decimal place. A bridge that turns 4 leave days into 12 days off has a ratio of 3.0, which is read as three days off for every leave day spent.

Two neighbouring gaps can also be chained into one longer break when they share a middle non-working run: take both gaps as leave, and the break spans the left run, both gaps, and the middle and right runs. Version 1 does not chain three or more gaps. That is a stated boundary of the algorithm, not an oversight to work around: a break that would need three separate working gaps bridged together simply does not appear as a candidate, however attractive it might look on a calendar.

The three strategies, and what their headline numbers mean

Longest single break: one break, the longest within budget

The longest strategy looks only at candidates whose leave cost fits the budget, takes the one with the most consecutive days off, and breaks ties in favour of the higher ratio. It returns exactly one break, or nothing. When nothing fits, the tool says: No bridge fits within your leave budget — try more leave days or add company days. That message is a property of the budget, not an error: adding a company day creates new non-working runs, which can create new gaps and new candidates.

Spread evenly: several breaks, totalled

The balanced strategy works in two passes. First it anchors one break per quarter, taking the highest-ratio candidate in each quarter whose starting month falls in that quarter, provided it fits the budget and does not overlap a break already chosen. Then it fills the remaining budget with a weighted interval scheduling pass: the best set of non-overlapping candidates that maximises total consecutive days off under the remaining budget. The result is sorted by start date, and the headline number is the total across all chosen breaks. Comparing that total with the longest strategy single break is therefore comparing a sum with a single value, which is why the balanced total is usually larger while no single balanced break is the longest available.

All opportunities: a ranked table, not a plan

The all-opportunities view does not choose at all. It lists every candidate in a table with five columns — Break, Days off, Leave used, Ratio, and Leave days — sorted by ratio descending, then by consecutive days descending. Crucially, the table is not filtered by the leave budget. Rows that cost more leave than the entered budget appear in it alongside rows that fit, which is why it must never be read as a recommended plan. Its use is comparison: seeing the highest-ratio breaks in the year, including ones a different budget might unlock.

One input, three different answers

Japan with 4 leave days, Saturday-Sunday weekends, and no company days shows why the strategy matters more than the headline. The longest strategy gives 4 days of leave turning into 12 days off at 3.0 times, one break from April 24 to May 5, 2027, using leave on April 26, April 27, April 28, and April 30. The balanced strategy on the same input gives 4 days of leave turning into 19 days off at 4.8 times, but as four breaks totalling 19 days: February 11 to February 14, April 29 to May 5, September 23 to September 26, and November 20 to November 23. The all-opportunities table on the same input contains both of those breaks and also a 16-day break from May 1 to May 16, 2027, which costs 7 leave days — beyond the 4-day budget entered. That row is the proof that the table ignores the budget: it is displayed because it exists as a candidate, not because the entered budget can pay for it.

What this verification does not cover

• The official originals for Germany, France, Japan, South Korea, Australia, Canada, and Hong Kong were not reopened one by one in this round, as stated above.
• The calendar export was checked as a file, not as a rendered calendar. The download is named leave-2027.ics, uses the standard calendar format, and creates one all-day event per leave day, but how that file displays after import into Google, Outlook, or Apple calendars was not tested in this round.
• Only the standalone tool itself was tested in this round. How it behaves inside the embedded BeadCalc page was not part of this verification round.

None of these limits changes the arithmetic in this guide, but each one marks a place where a reader should verify before acting: the official calendar for the seven unreopened countries, the imported calendar for the export, and the embedded page for the final experience.

How to double-check a date before you book

1. United Kingdom: open the GOV.UK bank holidays page at https://www.gov.uk/bank-holidays and read the England and Wales table, not the Scotland or Northern Ireland tables.
2. Singapore: open the Ministry of Manpower gazetted holiday table at https://www.mom.gov.sg/newsroom/press-releases/2026/0618-public-holidays-for-2027 and remember the February presentation difference: the table names February 7 with February 8 in lieu, while the tool lists only February 8.
3. United States: check whether the date you see is the calendar holiday or the observed weekday. The three observed dates in the tool are June 18, July 5, and December 24, each standing in for a holiday that falls on a weekend.
4. Germany, France, Japan, South Korea, Australia, Canada, and Hong Kong: open the official calendar of that country and confirm the specific dates you plan to use, since this round did not reopen those originals.
5. Every country: confirm the plan against your employer rules before submitting it. The optimizer knows dates, not company policy, which is the subject of the companion guide 7 leave-planning mistakes.

Done that way, the tool does what it is built to do: it takes a stated snapshot of 2027 holidays, applies stated scope rules, finds the bridges the arithmetic allows, and reports them under the strategy you chose. The snapshot and the rules are now on the table, tier by tier, so the plan can be checked rather than merely trusted.