How we reach our answers

Every property is grouped into a cluster — same ZIP code and same room count (e.g. "8700 · 3.5-room") — together with the rental comparables and other tracked properties that match it. Clusters are the unit both rent estimates and cross-property benchmarking are built on; the comparison's "Cluster benchmark" card shows a property against its cluster peers on size, construction year, price/m², and gross yield, plus a detail line per property covering condition, floor, outdoor space, view, orientation, and parking.

Those detail-line fields are extracted from each listing's own research notes/description — not independently verified or newly researched — so a field showing "not stated" or "not researched" means the source material didn't mention it, not that the property lacks that feature.

• Properties: the internet is screened nightly for currently or recently listed residential properties, for sale and for rent, across Canton Zürich (ZIP codes 8000–8999). The scan re-runs and replaces itself every night, so the exact count and research date shift day to day — see the figure shown on each comparison page itself. New-build listings that don't disclose an exact street address pre-sale are marked "address on request."
• Mortgage rates: indicative fixed rates for each tracked lender, screened nightly from public rate-comparison sources and cross-checked against the lender’s own published rates where available — shown on the same indicative, not-a-binding-offer basis those sources use themselves.
• Rental comparables: asking rents for similar units in the same neighbourhood, used to estimate a realistic achievable rent per property — using the same cluster-based method described in §3 below.

This data is a point-in-time snapshot, not a live feed. Rows or providers marked with an asterisk (*) in the financing matrix are carried forward from an earlier research round and have not been reconfirmed in the most recent refresh — treat those numbers as indicative only and confirm directly with the lender.

This canton-wide nightly scan is the single master dataset behind three of this site's four pillars — Investment, Living and Benchmarking. (The Market pillar draws on this same listings data too, aggregated by ZIP code and week instead of per-listing clusters — see §9.) Beyond what a search-results page shows, the scan also captures parking and outdoor-space (balcony) availability and total floors in the building — it does not yet capture written condition assessments, which floor a specific unit is on, or compass orientation. Rental listings from this scan are independently browsable, not just used as comparables. Buy-side rent estimates and gross yield use the same cluster-based method as §3 below. The full financing-cascade analysis (§4) runs directly on any for-sale listing in this scan; a listing whose price, size and room count aren't enough on their own (e.g. one you type in yourself under "Enter my own") routes through a "custom property" entry instead.

Like with like: every night we read the text of each listing and set aside exchange offers, search ads, reserved listings, limited-term, furnished and shared rentals, and copies of the same home on several portals, so medians, rent estimates and benchmarks compare like with like. Listings offered on different terms — for example a minimum price in a bidding process, a building right (Baurecht) or a furnished flat — stay searchable and carry a clear label. Each property appears once, in its most complete live version, together with how long it has been on the market and any price change since it was first listed.

Rent on the right basis: on Living we compare the full monthly rent including service charges (Nebenkosten), so renting and owning are set against all running costs. On Investment we work with the net rent, the amount the owner actually receives, and calculate the gross yield on it, as is usual in Switzerland. Where a listing does not state its service charges, we use the typical charges per m² of comparable homes in the same area. Very large homes from 300 m² stay searchable with a label; our benchmarks focus on ordinary homes, so their medians stay comparable.

Within each cluster, rent is estimated one of two ways, depending on how much comp data is available:

• Per-m² method (preferred): if a cluster has at least 2 rental comps with a known floor area, their CHF/m²/month is computed and the median taken; this rate is then multiplied by each property's own size. This corrects for the common case where a matching-room-count comp is considerably smaller or larger than the subject property — a flat median of raw rents would otherwise mis-price it.
• Flat median (fallback): if fewer than 2 sized comps exist in the cluster, rent falls back to the plain median of that cluster's comp rents, regardless of size — the same method used throughout earlier versions of this tool.

Which method applied to a given property, and how many comps backed it, is recorded and shown in the comparison's cluster benchmark card. Some clusters (ZIP × room-count combinations) still only have 1 sized rent comp or fewer — the flat-median fallback exists for exactly that case, and its numbers should be treated as rougher estimates than a per-m²-backed one.

Estimated yield — shown on every property card and in the cluster benchmark table — is exactly this estimated annual rent ÷ the property’s purchase price. It is a pre-financing figure: it does not include mortgage interest, tax, maintenance or vacancy, so it is not the same as what you’d actually keep after financing the purchase (see §4 below for the numbers that do account for that). Its only real source of uncertainty is the rent estimate behind it — the confidence label and comp count on the cluster benchmark card tell you how many comparable rentals actually back that figure for this specific property, not a signed lease for the unit itself. The full analysis’s own “Gross Rental Yield” figure starts from this same estimate too, but recalculates live from whatever monthly rent is currently set in the rent field below the property — so it stops being an estimate, in either direction, the moment you drag the slider or type a rent of your own.

Unless you change them, the comparison starts from that night’s cheapest 5-year fixed rate and the selected lender’s own minimum required equity share (20% for every tracked lender except Glarner KB/Hypomat, at 34%) — both freely adjustable with the controls above, and both feeding directly into the Return on Equity and net cash flow figures below. For a given property, lender, fixed-rate term, equity share, and monthly rent, the model then runs through:

1. Equity / mortgage split. Mortgage amount = purchase price − equity. Each lender has its own minimum equity share and maximum loan-to-value (LTV); the comparison won't let you select an equity share below a lender's published minimum.
2. First vs. second charge. Most Swiss lenders split the mortgage into a first-rank portion (up to 65% LTV, interest-only, no scheduled repayment) and a second-rank portion (the slice between 65% LTV and the lender's max LTV), which by regulation must be amortised down to 65% LTV — this model assumes straight-line, over 15 years unless the lender's own published amortisation period is shorter.
3. Interest. The selected rate (fixed-term or SARON/variable) applied to the outstanding mortgage balance.
4. Maintenance. A flat 1% of purchase price per year — the standard Swiss banking rule-of-thumb for ongoing upkeep and reserves, not a property-specific quote.
5. Tax effect — modeled differently by pillar, not as one blanket rule. On Investment (buy-to-rent-out): a flat 28% marginal rate applied to net taxable rental income (rent minus interest and maintenance), floored at zero so the model never turns a loss into a subsidy — unrelated to the Eigenmietwert reform below and unaffected by it, so this stays exactly as it is. On Living (buy-to-live-in): both the imputed rental value (Eigenmietwert) tax charge and mortgage-interest deductibility are abolished starting in 2029 per the confirmed 2025-09-28 referendum, and this model already applies that — there is no tax effect from ownership here at all, not a saving from one offsetting a cost from the other, but a genuine zero. See §6 below for the full Eigenmietwert explanation. Your actual Investment-side rate depends on canton, commune, and personal tax bracket, and can differ substantially from 28%.
6. Net cash flow = rent − interest − amortisation − maintenance − tax effect. Shown both per year and per month.
7. Break-even equity. The comparison separately sweeps the equity share from the lender's minimum up to 50% to find the smallest equity stake at which net cash flow crosses zero, holding rate and rent constant.

SARON and market-best rates

SARON (variable rate) is shown for 8 of the 9 tracked lenders and refreshed every night, together with the fixed rates. The rate shown is what a borrower pays today: the lender's margin plus the compounded SARON rate, which is reset every 3 months (while SARON is at or below zero, the rate equals the margin). The market best row shows, for each term, the lowest published rate across more than 30 Swiss lenders, also refreshed every night, so you can see at a glance how a tracked lender compares with the whole market. The comparison opens on the cheapest 5-year fixed lender of the day — whichever that is, not a fixed name.

Split / mixed-tranche financing

You can split the mortgage into two rate tranches from the same lender — e.g. 50% SARON + 50% 5-year fixed — with a slider controlling the split. The effective interest rate used in the calculation cascade is simply the weighted average of the two tranches' rates; amortisation and the rank1/rank2 LTV split are unaffected, since those depend on the loan-to-value, not on which rate product is attached to which slice. Mixed mode is disabled for interest-only/fixed-equity products (see below), since those don't use the standard rank1/rank2 structure at all.

Special case: interest-only / fixed-equity products

A small number of lender products (e.g. Glarner Kantonalbank's "Hypomat" line) don't follow the first/second charge split at all — instead they require a fixed equity share (currently modeled at 34%) with no scheduled amortisation. The comparison detects these products and switches to that simpler branch automatically; the equity slider locks to the required percentage when one of these is selected.

Separate from the buy-to-rent-out model above, the comparison also offers a standard Swiss owner-occupier affordability check: enter your own annual gross income and it computes your affordability ratio — the ratio of imputed housing costs to income, independent of any rental scenario.

• Imputed housing cost = mortgage × a standardized 5% calculatory interest rate, plus 1%/yr maintenance, plus any required amortisation (same rank2/15-year rule as the main model).
• The 5% rate is a deliberate bank stress-test convention, not today's actual mortgage rate — it's what Swiss lenders typically underwrite against, so a mortgage stays serviceable even if rates rise.
• Loan-to-Value (LTV) = mortgage ÷ purchase price, shown alongside for reference.
• Guideline: affordability should generally stay at or below ~33%; some lenders allow up to ~35–36%.

Separate again from both the buy-to-rent-out model and the affordability check, this compares the cost of buying a property to actually live in against renting a comparable one — the classic rent-or-buy question, using the general approach independent rent-or-buy comparison tools take.

• Down-payment sweep: the comparison is run across five representative down payments from 20% to 50%, each showing its down payment, annual interest, maintenance (1%/yr of price), the combined annual outlay, this cluster's rent, the net yearly saving vs. renting, amortisation, and the net cash flow after amortisation (F − G — amortisation is cash leaving your account, not a saving). Amortisation drops to zero once loan-to-value falls to 65% or below (35% down payment and above here) — that's the same rank1/65% threshold used throughout this tool, not a new rule.
• Imputed rental value (Eigenmietwert) and mortgage-interest deductibility. Swiss owner-occupiers have had to declare a notional rental income on their own home, taxed net of deductible mortgage interest and maintenance. Both are abolished starting in 2029. This comparison already models the post-2029 rules rather than today's still-current pre-2029 ones: no Eigenmietwert tax charge, and no interest deduction either.
• Net yearly saving vs. renting compares, cash-for-cash: (annual interest + maintenance — no tax effect, see above) against this property's own cluster market-rent estimate — the same "similar home" rent already used throughout the tool. Amortisation is deliberately excluded from this figure, since it's your own money moving into home equity, not money that's gone the way rent is.
• Opportunity cost of equity. Following the same general approach independent rent-or-buy comparison tools take — money spent on a down payment could otherwise have been invested — an adjustable slider (default 3%/yr) lets you set your own assumption for what that capital might otherwise earn. "Total benefit of buying vs. renting" combines the net yearly saving vs. renting, the equity built through amortisation, and the assumed home-price appreciation below, then subtracts this opportunity cost, giving a single figure for the currently selected down payment.
• Assumed annual home price appreciation. Applied to the full property price, not just your equity — the leverage effect of a mortgage means you own 100% of the gain on a home you only part-paid for. A second adjustable slider (default 1%/yr) lets you set your own assumption; 0% assumes prices stay flat. There is no reliable ZIP-level historical appreciation data behind this site today (see "What the model deliberately ignores" below), so this is a user-adjustable starting point, not a fitted or researched figure — treat it the same way as the opportunity-cost assumption above.
• Not applicable to fixed-equity products (e.g. Glarner KB's Hypomat) — those require a fixed 34% down payment rather than a swept range, so the comparison shows a note instead of a table when one is selected.

• Transaction costs — notary fees, property transfer tax, brokerage commission, mortgage set-up fees.
• Renovation reserves beyond the flat 1%/yr maintenance assumption.
• Vacancy risk — the model assumes the unit is rented every month at the selected rent.
• Value appreciation or depreciation of the property itself, outside the Living buy-vs-rent comparison's adjustable appreciation assumption (see above) — the Investment pillar's returns and cap-rate figures still assume a static price, and there is no reliable ZIP-level historical appreciation data behind either.
• Canton- and commune-specific tax rules beyond the flat 28% Investment-side approximation (actual marginal rates, wealth tax, church tax, etc. all vary).
• The current, pre-2029 Eigenmietwert and mortgage-interest-deduction rules (Living pillar only) — this model already assumes the confirmed 2029 abolition applies today, and doesn't offer a toggle to see the pre-2029 numbers instead.

Both datasets behind this site now refresh nightly, fully automated — including the ZIP-level Market aggregates (§9), which are recomputed from that same nightly scan. The canton-wide property scan (§2) behind Investment, Living and Benchmarking re-runs and replaces itself every night. Swiss lender mortgage rates are screened the same way: each tracked lender’s headline fixed rate is re-checked nightly against public rate-comparison sources and the lender’s own site where available, and shown as an indicative rate — the same basis those sources use themselves, not a binding offer. A handful of rows, marked with an asterisk in the financing matrix, are carried forward from an earlier research round rather than the nightly screen — see §2. Separately again, a lightweight automated nightly check re-publishes this website whenever either underlying dataset changes — it does not perform any new research itself. See the About page for more on how these fit together.

The Market tab draws on the same canton-wide listings data used elsewhere on this site, but aggregated differently: by ZIP code and week rather than per-listing clusters. Current Status shows canton-wide, listing-count-weighted price, rent and gross-yield medians by ZIP code and room range, pooled over a rolling 7-day window and deduplicated across portals. Short-Term Trends and Historical Evolution chart those same weekly ZIP-level medians over time.

Gross yield is estimated, not measured: for each ZIP code, the median asking price per m² is divided by an estimated annualised rent, since none of these units are actually let out. The monthly rent figure itself comes from multiplying a ZIP's median CHF/m²/month rent by its own median rental apartment size (falling back to the canton-wide median of 85 m² where a ZIP has fewer than 5 rental comparables). ZIP codes below a medium-quality threshold (at least 5 comparable buy and rent listings in the trailing week) are shown as no-data rather than with a shaky estimate.

The Forward-Looking tab uses a different, Gemeinde-level dataset: the Casaris Blended Outlook is an equal-weighted average of a UBS-derived sub-score and each Gemeinde's own ZKB price-level percentile, both rescaled to 0–1. This is reported at the Gemeinde level (162 municipalities) because that is the finest granularity either public source publishes — UBS confirms only one canton-wide data point in its report text, with ZKB's Preisniveau rating supplying the genuine Gemeinde-to-Gemeinde detail.

Net rents and like-for-like data on the Market page: all rents on the Market page are net rents, excluding service charges (Nebenkosten) — the amount an owner actually receives — and the gross yield is calculated on them, as is usual in Switzerland. Where a listing does not state its service charges, we deduct the typical charges per m² of comparable homes in the same area. The weekly medians use comparable listings only: furnished, temporary and shared rentals, swap offers, auctions and very large homes stay searchable elsewhere on the site but are set aside here, so each week compares like with like since the series began on 14 September 2026.