Monthly Pita-Gyro Price Index: October 2026

Monthly change

Each listing against itself, 26 October 2026 against the regular reading of September 2026: geometric mean of the price ratios +0.32%, median 0.00% with 94.1% of prices unchanged.

Diffusion: of the 1,179 pairs, 67 rose (5.7%), 2 fell (0.2%) and 1,110 kept the same price (94.1%).

Sensitivity check, not the official rate: platform C stopped sending the keys changeInPriceInSfu, envFeeInSfu, friendlyId, tagIds and uid of the field selections, across the whole catalogue and without any product changing. Rule 1 excludes every pair with a changed field. Rule 14, written on 26 October 2026 for the data from November 2026 on, removes such keys before the comparison; here it would give +0.28%, with a 95% interval from +0.20% to +0.36%, on 1,485 pairs (306 from platform C). The index moves with the official change only.

Annual change: from the September 2027 issue, published on 28 September 2027, when the same month of the previous year exists.

Official comparison

Official comparison: Eurostat's Harmonised CPI for Greece, September 2026, the latest month with every value, one month before the bulletin's month. Restaurants, cafés and the like (CP1111): +0.6% on the month, 5.3% on the year; All items (TOTAL, formerly CP00): +1.9% on the month, 5.1% on the year.

The official index measures a different basket with a different method: every restaurant and café, not the pita-gyro on delivery platforms. The comparison shows direction, not the same quantity. Eurostat read on 4 November 2026 (dataset prc_hicp_minr, ECOICOP ver.2), with the SHA-256 of every response in the issue's JSON. Eurostat

Method and denominators

How it is computed: every month three delivery platforms are read on one weekday from day 25 to day 28, never on a holiday. The index is chained: each month multiplies the previous one by the geometric mean of the price ratios of each listing against itself (basic identity, the price the customer pays; on platform B the original price while an offer runs). The 95% interval is a bootstrap with 2,000 replicates and a declared seed, with clusters city × shop name and strata the platform combination, so that the same business on two platforms does not count twice. Below 100 pairs no change is published, and a platform with fewer than 30 gives none of its own. A missed month is bridged as a change over two months, never interpolated.

Denominators: level 1,663 listings, 1,314 businesses, 78 cities, 3 platforms; change 1,179 pairs, 1,055 businesses, 77 cities.

Rules: rules-2026-10-v2, written on 9 October 2026 before the data, SHA-256 04d05cba454663338162b1aa6f4c640895dd274a2708d9ced704eb8e96c048e8. Rules (JSON)

What it does not measure

• Platform prices, not counter prices: the price of the item on the menu of three delivery platforms, which can differ from the price at the shop.
• No seasonal adjustment: that needs at least 3 full years of readings.
• A level is not a rate: a city median also moves when shops enter or leave. The change compares each shop with itself.
• It does not change the annual Pita-Gyro Index. Such a change would be a separate decision, with its own data release and changelog entry.

Data

This issue as data: JSON and CSV, unchanged after publication, with the SHA-256 of each file in the bulletin catalogue.

• Download the issue (JSON)SHA-256 4211ed2fc3902f23150da0e8bfe2ef7e4a4059d4cdb89c364f87274a051778c8
• Download the issue (CSV)SHA-256 09341c9290312495443eaeaaf0e15384a70ae6dc53542b84d2919a59f87ac225
• Every shop of the reading (JSONL)
• API: every issue and the calendar

Revision policy: a published issue never changes. A correction is a new version of the issue, in a new file with its own SHA-256 and a changelog entry; the earlier version stays available.