ssh.autos

Recipes

Recipe · Apify Store · runs in a terminal

The square-metre trap in Buenos Aires apartment listings

At a US$3,500 per m² ceiling, 63 ZonaProp listings passed using total area. Only 42 of the same listings passed using covered area.

COSTFree to reproduce from archived data
DATAArchived Apify datasets
TIMECollected November 14 and 17, 2026 UTC
YOU NEEDPython standard library

Say I’m shortlisting apartments in Palermo and Belgrano and want to pay at most US$3,500 per covered m². If I calculate the rate with each listing’s total area, 63 apartments in my ZonaProp sample make the shortlist. Use each ad’s covered area and 42 make it. All 21 changes come from the denominator. The asking prices are identical in both calculations.

Price the same ads twice

I used a ZonaProp archive of Palermo and Belgrano apartment sales collected on November 17, 2026 UTC. The run returned 120 rows. I excluded 20 development ads and five other ads without a usable area pair. The remaining 95 had a positive USD asking price and a positive covered area no larger than the total area. I used the neighborhood names from each listing’s location hierarchy, including sub-neighborhoods under their parent. automatic/zonaprop-ar-scraper

Neighborhood Returned Used in both calculations Median USD per total m² Median USD per covered m² At or below $3,500 using total At or below $3,500 using covered
Palermo 60 50 3,408 3,898 27 18
Belgrano 60 45 2,712 3,333 36 24
Both 120 95 63 42

For each listing I divided the same price by each area field, then took the median of the resulting rates. Thresholds use the unrounded rates. The median listing-level increase from switching to covered area was 9% in Palermo and 14% in Belgrano. Covered area and total area capture different things. A terrace can make an apartment more valuable to you, even when you set your price ceiling using covered area.

Check what a second portal can support

I checked a Mercado Libre archive of Palermo apartment sales collected on November 14, 2026 UTC. Of 96 Mercado Libre ads, 58 were classified as developments, 15 had unknown listing type and 23 qualified as matched individual units with a positive USD price and a single covered area. A development card can show an area range alongside a starting price. A ratio built from those fields may combine figures from different units. automatic/ml-inmuebles-scraper

Palermo sample Individual ads used Median USD per covered m² At or below $3,500 per covered m²
ZonaProp, paired-area ads 50 3,898 18
Mercado Libre, covered-area ads 23 2,714 16

The Mercado Libre units supplied no total-area values, so the paired calculation ends with ZonaProp. The portals returned different sets of ads on different days. The gap between their medians reflects those samples and cannot establish a portal effect or a citywide price. For a cross-portal shortlist, filter for individual units with covered area, then keep total area as a separate column for the ads that report it.

These are asking prices from ranked search results. Each archive has unique listing keys within its own portal. Multiple listings can still advertise the same apartment, and the analysis leaves the portal samples unmerged.

Reproduce the tables

Save the archive as apartments.json and the code below as compare.py, then run python3 compare.py apartments.json. The download preserves all returned rows from both runs and includes only the fields used in the analysis. Download the archived analysis fields

import json
import sys
from collections import Counter
from pathlib import Path
from statistics import median

zone, mercado = json.loads(Path(sys.argv[1]).read_text(encoding="utf-8"))["runs"]
ceiling = 3500

def count_at_or_below(rates):
    return sum(rate <= ceiling for rate in rates)

def paired(row):
    return (
        row["currency"] == "USD"
        and not row["isDevelopment"]
        and not row["isStartingPrice"]
        and (row["price"] or 0) > 0
        and 0 < (row["coveredAreaM2"] or 0) <= (row["totalAreaM2"] or 0)
    )

zone_rows = zone["items"]
assert len(zone_rows) == len({r["listingKey"] for r in zone_rows})
print("Development exclusions:", sum(r["isDevelopment"] for r in zone_rows))
print("Other exclusions:", sum(not r["isDevelopment"] and not paired(r)
                               for r in zone_rows))
for neighborhood in ("Palermo", "Belgrano"):
    raw = [r for r in zone_rows if r["neighborhood"] == neighborhood]
    rows = [r for r in raw if paired(r)]
    total = [r["price"] / r["totalAreaM2"] for r in rows]
    covered = [r["price"] / r["coveredAreaM2"] for r in rows]
    lift = [c / t - 1 for c, t in zip(covered, total)]
    print(neighborhood, len(raw), len(rows), f"{median(total):,.0f}",
          f"{median(covered):,.0f}", count_at_or_below(total), count_at_or_below(covered),
          f"{median(lift):.0%}")

rows = [r for r in zone_rows if paired(r)]
print("Both", len(zone_rows), len(rows),
      count_at_or_below(r["price"] / r["totalAreaM2"] for r in rows),
      count_at_or_below(r["price"] / r["coveredAreaM2"] for r in rows))

ml_rows = mercado["items"]
assert len(ml_rows) == len({r["listing_key"] for r in ml_rows})
units = [r for r in ml_rows if r["listing_kind"] == "unit"
         and r["unit_correspondence"] == "matched" and r["currency"] == "USD"
         and (r["price"] or 0) > 0 and (r["surface_covered_m2"] or 0) > 0
         and not r["surface_covered_m2_is_range"]]
rates = [r["price"] / r["surface_covered_m2"] for r in units]
print("Mercado Libre", len(ml_rows), Counter(r["listing_kind"] for r in ml_rows))
print("Eligible", len(units), f"{median(rates):,.0f}", count_at_or_below(rates),
      "with total area", sum(bool(r["surface_total_m2"]) for r in units))

Apify run IDs

Search Collected (UTC) Run ID Dataset ID
ZonaProp Palermo and Belgrano sale apartments 2026-11-17 cTsP6HdTvmJASLcRA mtf5Oe5vLl7UXHUNU
Mercado Libre Palermo sale apartments 2026-11-14 NLcLW2J7P95AQJUt1 D4pkAfoesnJV6Lxg3

Affiliate disclosure: The Apify Store links carry our referral code. If you sign up for a paid plan through one, ssh.autos may earn a commission at no cost to you.

Sources in this recipe

Machine-readable: openapi.json · llms.txt

Some links to Apify are affiliate links. We may earn a commission if you sign up and become a paying customer.