How we estimate

What a home would most likely sell for today on the open market, estimated by Princeps’s model and shown as a range.

Where Princeps cannot estimate a home on its own, the page shows what a typical home in its area would sell for, and says that the figure is the area’s.

It is not an appraisal, not the value a municipality taxes the home on, not an advertised price, and not a statement that the home is for sale.

Against prices the model had not seen. A test sets the most recent months aside, estimates the homes that changed hands in them and compares each estimate with the price. Where sale prices are not public, the test is against advertised prices.

The range on a page is the span that held 8 in 10 of the test’s estimates for homes like that one. The page sums up the usual miss as “Typically within about X%”.

When Princeps cannot estimate the home, or its area, within a range narrow enough to be useful; when the model has not passed its test for homes of that kind; or when the area has no price index to follow.

On vacant land the sale value reads “Not available” until a test for land exists.

About what a year of property tax on a home comes to, from Princeps’s own estimate of its value and the local effective rate: the share of a home’s value that a year of tax comes to in its municipality.

Every year on the page’s chart is modelled. None is a past bill.

It is not the tax bill. It does not include exemptions, caps or phase-ins, or deferrals. Where a municipality bills a flat charge with the tax, the estimate spreads it over the value.

Against tax amounts the model had not seen, as the sale value is against prices. The range is the span that held 8 in 10 of those estimates, and the page sums up the usual miss in the same words.

When the sale value is “Not available”, since the tax is computed from it; in Nova Scotia and the territories, where Princeps does not estimate the tax; and on vacant land until a test for land exists.

Shown on vacant land only: what it would cost today to build a new house of the average size for the area, at local construction prices. The page shows the size it assumed.

It is not the value of the land and not a builder’s quote.

It does not say that a house may be built: zoning, servicing, setbacks and flood rules are not checked. Land, servicing, municipal approvals and development charges are not included.

No test against an outcome exists for this figure. Its range shows how far its assumptions can move it.

A house that has not been built has no price to compare the estimate with.

When Princeps cannot work out construction prices for the region.

A score from 0 to 100 that compares an area with every other in Canada on safety, daily amenities, services, nature, natural hazards, market strength and growth. 50 is the national median, and higher is better.

Each of the seven parts is scored from 0 to 100 as well, and the page shows them.

It rates an area, not a home, and not the people who live there. It is not a school ranking, not a price forecast and not a statement about safety at an address.

No test against an outcome exists for this figure. Its range shows how far its assumptions can move it.

A score is a comparison, not a prediction: nothing happens later that could prove it right or wrong.

When too few of its parts can be measured for the area.

The monthly payment on a mortgage for the estimated sale value, plus a twelfth of the estimated annual tax. The page shows the total and its two parts.

It is an illustration under one set of assumptions, the same for the whole country and stated on the page:

• a down payment of 20%
• a mortgage repaid over 25 years
• the rate banks lend at on new five-year fixed mortgages, rounded to the nearest quarter point
• interest compounded semi-annually (twice a year), as on fixed-rate mortgages in Canada

It is not a mortgage offer. Home insurance, utilities, condo fees and upkeep are not included. It has no range.

It has no test of its own. It is arithmetic on two figures the page already shows and one assumed rate, so it is as accurate as the sale value and the tax it is computed from.

When the sale value is “Not available”. When only the tax is missing, the page shows the mortgage payment alone and says so.

Vacant land has no monthly cost: a loan on land is not this mortgage.

How Princeps’s price index for a home’s market area and type of home moved over the past 12 months and over five years, stated in words, with the province’s beside it.

Over 12 months, prices “rose” or “fell” when the index moved by more than 2%, and “held steady” otherwise. Over five years the line is 5%. A rise or a fall is given as a whole percent, with “about”.

It is the market area’s trend, not a home’s and not a street’s: the index goes no finer. It describes the past and is not a forecast.

It has no test of its own. The words restate the index that the page’s chart of the sale value follows.

When the area has no price index. A period the index does not reach back to is left out of the sentence.

Where a home’s estimated sale value stands among homes of its type on its street, in its area and in its municipality: in the lower quarter, in the middle half or in the upper quarter.

It compares Princeps’s estimates with one another, not prices paid, and it compares the sale value only.

It has no test of its own. It ranks figures that are on the pages already, each tested as a sale value is.

It never reads “Not available”: a comparison that cannot be made is left out. That is so when a group holds too few homes, when its homes stand at fewer than three addresses, as in one building, and when the page’s figure is the area’s and not the home’s own.

What a home like the one on the page would rent for each month on a new lease today, shown as a range.

Where the model cannot tell homes apart, which is so for most apartments, the figure is the area’s for a home of that type, and the page says so.

The rental yield is twelve months of that rent divided by the estimated sale value, given as “about N% a year, before costs”.

It is not the rent a tenant pays now, not a rent set by law, and not an offer.

The model has no count of a home’s rooms, so the figure is for a home like it and never “this home’s rent”. The yield is before every cost of owning: tax, insurance, upkeep, a mortgage and months without a tenant.

Against advertised rents the model had not seen, as the sale value is against prices. The range is the span that held 8 in 10 of those estimates.

When the model has not passed its test for homes of that kind, when the home’s type is not covered (a building of several dwellings, for one) or when the area has no rent index.

The yield reads “Not available” unless the rent and the sale value are both shown and are both the home’s own or both the same area’s. Vacant land has no rent.

Two bands for the area a home is in, each low, moderate or high.

Flood susceptibility places the area among all areas in Canada: “high” is the tenth of homes in the most susceptible areas, and “moderate” the next fifth. It is not a height of water. Where a mapped flood zone covers part of an area, it can raise the band and never lowers it.

Wildfire history nearby says whether a wildfire is on record near the area in the past 30 years: “high” within 2 km of the centre of the area, “moderate” within 10 km and “low” beyond.

It is a screening of an area from public maps. It is not a flood map or a fire forecast for an address, and not an insurance or safety finding.

The wildfire band is history: it says where fires burned, not where they will. For local maps, ask the municipality.

Neither band is a prediction, so there is no outcome to test it against. Each sums up public maps in one of three words.

When the area has too few homes to band.