From Parcel to Buildable: Setbacks, Coverage, FAR, and What Public Records Reveal

A buildable envelope is a subtraction. Start with the gross area of the parcel, take away the land the code reserves as yards, cap what is left by whatever lot coverage rule applies, multiply the surviving footprint by the stories the height limit allows, and check the result against the floor area ratio the district tabulates. Five steps, four parameters, one geometry. Any analyst can do it on paper in ten minutes.

The arithmetic is not the hard part. The hard part is that every input to it lives in a different record, kept by a different office, on a different schedule, and two of the five inputs are usually missing. San Francisco publishes one of the better assessor rolls in the country, and on the roll year 2025 file the lot area field is populated on 148,638 of 211,547 records, the building area field on 198,803, the lot depth field on 39,980 and the lot frontage field on 607 (San Francisco Office of the Assessor-Recorder, Assessor Historical Secured Property Tax Rolls, current as of June 26, 2026, queried August 24, 2026). Frontage and depth are the two numbers the setback rules operate on. They are present on 18.9% and 0.29% of the records respectively.

The second problem is subtler and more expensive. Analysts screen on floor area ratio because it is the one parameter that is sometimes published as a number, and because it produces a single answer from a single multiplication. It is also the parameter least likely to be the binding constraint. In New Hampshire, the one state this research found publishing the full set of dimensional standards as data, a floor area ratio exists in 106 of 3,784 mapped zoning districts, which is 2.8%, while front, side and rear setbacks are coded on 2,878, which is 76.1% (New Hampshire Zoning Atlas, Saint Anselm College Initiative for Housing Policy and Practice with New Hampshire Housing, zoning changes adopted through June 1, 2025, feature service queried August 24, 2026). In San Francisco, Table 124 of the Planning Code tabulates a basic floor area ratio for most districts and Section 124(b) then exempts residential uses in the R, RC, NC and Mixed Use Districts from it entirely, which removes the tabulated ratio from 154,973 of the 211,547 parcel records on the roll, or 73.3% (San Francisco Planning Code, 2026; roll year 2025, queried August 24, 2026; share computed by MMCG).

That is the counterintuitive finding this piece is built on, and it has two halves. Which constraint binds is a function of lot size, so the same zoning district produces setback-bound small lots and coverage-bound large lots and the crossover point can be derived exactly from the code's own numbers. And a screen that checks only floor area ratio passes precisely the parcels where floor area ratio is slack or switched off, because the codes that bother to tabulate a ratio tend to disapply it for the uses people actually build.

What follows is a reproducible method: the envelope arithmetic stated precisely, what public records supply against each input, two worked codes with real cited parameter values, the point at which each constraint takes over, the failure modes ranked by how often they bite, and the honest boundary between a screen and an entitlement opinion. It sits inside the wider argument about the public-data stack for commercial real estate analysis, and it assumes you have already resolved the district, which is the subject of the companion piece on zoning data by parcel.

The envelope is a subtraction, and the order of operations matters

Write the envelope as a sequence rather than a formula, because the sequence is where screens go wrong. For a rectangular lot of frontage f and depth d:

1. Gross parcel area. A equals f times d. This is the recorded lot area, not the surveyed area, and the two differ more often than analysts expect.
2. Yards and setbacks. Subtract the front setback, the rear yard and any side yards. Setbacks are measured from lot lines and are the only envelope parameter that touches the boundary of the parcel, which is why they are the parameter most sensitive to whether the geometry is right.
3. Lot coverage. Cap the surviving footprint at the coverage limit, expressed as a share of the lot or as a formula keyed to lot size. Coverage does not subtract from a particular place on the lot; it caps the total.
4. Height. Convert the footprint into floor area by the number of stories the height limit permits. Height rules specify a measurement datum, and the datum matters on a slope.
5. Floor area ratio. Cap the total floor area at the tabulated ratio times the lot area, if the ratio applies to the use being proposed.

Two properties of that sequence do the analytic work. First, steps 2 and 3 are alternative constraints on the same quantity, the footprint, and only one of them binds. Second, steps 4 and 5 are alternative constraints on floor area, and again only one binds. So a buildable envelope is really the minimum of two pairs, and the interesting question is never what the numbers are but which of the four is doing the work.

The order also determines what a missing input costs you. If the coverage cap binds, you do not need the depth of the lot at all, because coverage is a share of an area you already have. If the setbacks bind, you need frontage and depth to the foot, because the subtraction runs on linear dimensions. That asymmetry is why a screen built on a parcel file with reliable area and unreliable dimensions is defensible for some districts and worthless for others, and why the first question to ask of any envelope estimate is which constraint it thinks is binding.

A note on vocabulary, because codes are not consistent. Setback and yard are not always synonyms: many codes use yard for the required open area and setback for the distance from a lot line to a building wall, and a few use the terms interchangeably. Lot coverage sometimes means the footprint of buildings only and sometimes buildings plus impervious surface, and the two produce very different answers on a site with parking. Floor area ratio is defined by its numerator, and every code defines gross floor area differently. San Francisco's Section 102 measures gross floor area from the exterior faces of exterior walls in most districts, but along the glass line at a height of four feet above the finished floor in the C-3, Central SoMa and Van Ness Special Use Districts, and it excludes basements and cellars used for building services, rooftop mechanical areas and qualifying underground accessory parking (San Francisco Planning Code Section 102, 2026). None of that is exotic. All of it changes the answer.

The five inputs, and what public records supply against each

Take the five inputs in turn and ask what a public record actually gives you.

Parcel geometry. The polygon comes from a county assessor or a statewide compilation, and its lineage is the subject of a separate discussion of openness and quality of parcel records by state. What matters here is that the recorded area and the recorded dimensions are frequently not independent measurements. New York City's Department of Finance calculates lot area by multiplying lot frontage by lot depth for every tax lot not flagged as irregular, and computes it from the digital tax map only for irregular lots (New York City Department of City Planning, PLUTO Data Dictionary, August 2022 edition). The consequence is measurable: across Kings County, 93.8% of lots with positive frontage, depth and area reproduce the recorded area from frontage times depth to within 5%, but among the lots the assessor flags as irregularly shaped only 49.0% do (MMCG tabulation from PLUTO 26v1, queried August 24, 2026, reported in full in the companion analysis of land metrics computed from parcels). The rectangle holds exactly where the agency built the number as a rectangle.

The zoning district. Published almost everywhere as a polygon, and almost nowhere with the parameters attached. San Francisco is the exception worth studying, because its Planning Department layer carries a link back to the ordinance on every polygon: 10,617 polygons, 77 distinct districts, a code section on 9,709 of them and a deep link to the hosted code text on all 10,617 (City and County of San Francisco Planning Department, Zoning Map Zoning Districts, updated August 23, 2026, queried August 24, 2026). Then look at where the links point. The field holds 88 distinct values, of which 86 are section numbers and two are free text: 33 polygons carry a placeholder string instead of a section, and three carry the name of a redevelopment plan. Of the 86 section numbers, 85 sit in Articles 2, 2.4, 7, 8 and 9, which establish the districts and set out what may be done in them, and the single exception is a special height exception in Article 2.5. Not one points at Section 124, which tabulates floor area ratio, or Section 132, which sets front setbacks, or Section 134, which sets rear yards and lot coverage, or Section 260, which fixes how height is measured. The best-linked zoning layer examined for this piece links you to the use table, not to the envelope.

The dimensional parameters. There is no national source. The nearest thing is the National Zoning Atlas, a nonprofit compilation whose published method codes the types of residential uses allowed, permitting procedures, lot development requirements such as minimum lot sizes and parking mandates, and building parameters such as height caps (National Zoning Atlas, How the National Zoning Atlas is Made, 2024). Setbacks, coverage and floor area ratio are not in that summary. The one state-level exception found for this piece is New Hampshire, whose atlas service carries, per district and per housing type, minimum lot size in acres, front, side and rear setbacks in feet, minimum road frontage, maximum lot coverage for buildings and for buildings plus impervious surface, minimum parking, maximum height in stories and in feet, floor area ratio, minimum unit size and maximum density. That is the full envelope parameter set, published as data, for 3,784 districts across 263 jurisdictions in one state.

Terrain. USGS 3DEP is the national elevation source, and its coverage is the least of the problems: 3DEP-quality data were available or in production for 48% of the nation as of May 2018 and acquisition reached 89% national coverage by the end of 2022 (U.S. Geological Survey, 3D Elevation Program documentation, 2018 to 2026). Resolution is the problem. A 1/3 arc-second surface at roughly 10-meter spacing and a QL2 one-meter surface are not interchangeable for a development pad, which is the argument developed in the companion piece on slope and cut-fill screening.

Constraint overlays. Flood, wetlands and the rest remove land before the code does, and they are covered in the section below. Easements do not appear on any of them.

Set those five against each other and the pattern is consistent. Geometry is published widely and measured inconsistently. Districts are published widely and linked to their rules almost never. Parameters are published in one state. Terrain is published nationally at uneven resolution. Overlays are published nationally with disclaimers that forbid the use most analysts make of them. The envelope arithmetic is trivial; the provenance of its inputs is the entire job, which is the general point argued in the piece on why every number needs a source and a date.

Setbacks: the only parameter measured from the lot line

Coverage and floor area ratio are ratios. Height is a plane. Setbacks are the only envelope parameter expressed as a distance from a boundary, and that gives them three properties the other three do not have.

They are non-linear in lot size. A fixed setback consumes a share of the lot inversely proportional to the lot's dimensions, so the same rule that costs a five-acre site a tenth of its area can cost a small infill lot everything. Take the New Hampshire Zoning Atlas as the arithmetic source, because it is the one public dataset that publishes the numbers. Across the 2,878 districts that carry one-family setbacks, the mean front setback is 35.8 feet, the mean side setback 21.2 feet and the mean rear setback 25.0 feet, with a front setback range from 0 to 575 feet and a side setback range from 0 to 150 feet (New Hampshire Zoning Atlas, feature service queried August 24, 2026; means computed by MMCG). Apply those mean values to a square lot and the buildable share runs from nothing at 3,000 square feet to 5.6% at 5,000 square feet, 24.8% at a quarter acre, 56.5% at one acre and 79.1% at five acres. The zero point is exact: when the side of a square lot falls below the sum of the mean front and rear setbacks, 60.8 feet, the front and rear reservations meet in the middle and the buildable area is nil. That is a lot of 3,701 square feet.

They interact with the shape of the lot, not just its size. The subtraction above assumes a rectangle. Codes know this and say so: Portland's own zoning title states that its development standards are generally written for houses on flat, regularly shaped lots and that other situations are addressed through special regulations or exceptions (City of Portland, Title 33 Planning and Zoning, Chapter 33.110, Single-Dwelling Zones, as amended by Ordinance 191961 effective March 1, 2025). A flag lot, a wedge, a lot with a curved street frontage or a corner lot with two front setbacks all break the rectangle, and the direction of the error is not predictable. New York City flags 136,272 of its 858,244 tax lots as irregularly shaped, which is 15.9% (New York City Department of City Planning, PLUTO 26v1, queried August 24, 2026; share computed by MMCG).

They depend on facts about the neighbors. San Francisco's front setback rule is the clean example. Section 132 sets the required front setback in RH, RTO and RM Districts at no less than the setback of the adjacent building with the shortest front setback, or one half of that depth where only one adjacent building exists, measured as the horizontal distance from the street property line to the nearest building wall, excluding projections, decks and garages. Section 132(d) then caps the requirement at 10 feet, or at 15 feet where more than 75% of the properties on the block face and both adjacent parcels are set back 15 feet or more (San Francisco Planning Code Section 132, as last amended by Ordinance 10-26, effective March 16, 2026). The required front setback on a given parcel is therefore a function of two neighboring buildings and a block-face census. No parcel file carries it, no zoning layer carries it, and it cannot be computed from the subject parcel alone.

Corner and through lots are the standard variants and they are worth naming because they change the count of yards rather than their depth. A corner lot fronts two intersecting streets and typically carries two front setbacks, which on a small lot can leave a buildable strip rather than a rectangle. A through lot connects two streets with frontage on both, and codes usually treat both frontages as front lot lines, which eliminates the rear yard and replaces it with a second front setback. New York City's lot type field is the only place a screen can pick these up at scale: on PLUTO 26v1, 100,043 lots are coded corner, 3,602 through, 734,472 inside, 2,402 interior with no street frontage, 2,490 alley lots too narrow to accommodate a building and 1,865 waterfront (PLUTO 26v1 and the PLUTO Data Dictionary lot type code list, queried August 24, 2026).

The practical consequence for a screen is that setbacks should be flagged rather than computed wherever the code makes them conditional. A screen that applies a fixed front setback in an RH District in San Francisco is not approximating the rule; it is applying a different rule that happens to share a name.

Which constraint binds is a function of lot size

Now the arithmetic that gives this piece its title. Write the buildable footprint two ways for a rectangular lot of depth d, with a front setback F, a rear yard expressed as a share r of the depth subject to a floor of 15 feet, and a lot coverage cap c expressed as a share of the lot:

• From the setbacks, the footprint share is 1 minus F divided by d, minus the greater of r and 15 divided by d.
• From the coverage rule, the footprint share is simply c.

Both scale with frontage, so frontage cancels and the comparison depends on depth alone. Setbacks bind when the first expression is smaller than the second, which happens below a crossover depth of F divided by the quantity one minus r minus c.

Put San Francisco's numbers in. Section 134(c) sets the basic rear yard at 30% of total lot depth, never less than 15 feet, in the RH, RM-1, RM-2, RTO-1 and RTO-M Districts, and at 25% subject to the same 15-foot floor in the other districts it lists. Section 130(b) requires front setbacks and rear yards to extend the full width of the lot, with depth measured at right angles to the lot line. Section 134(l) limits development lots of two acres or more to 60% coverage in Residential Districts other than RTO-C, and to 65% above the second story elsewhere, with the C-3 Districts excepted (San Francisco Planning Code Sections 130, 132 and 134, 2026). With r equal to 0.30, c equal to 0.60 and the front setback at its 10-foot cap, the crossover depth is 10 divided by 0.10, which is exactly 100 feet.

Below 100 feet of depth the setbacks bind and the coverage cap is slack. Above 100 feet the coverage cap binds and the setbacks are slack. At exactly 100 feet the two rules produce an identical footprint, 60% of the lot, and a screen could apply either and get the same answer. Move the front setback to the 15-foot figure that Section 132(d) allows on a block face where more than 75% of properties are set back 15 feet or more, and the crossover moves to 150 feet. Remove the front setback altogether, which is what happens where no adjacent building establishes one, and the coverage cap binds at every depth above 50 feet.

There is a second, sharper version of the same effect, and it runs the other way. Portland caps building coverage in its single-dwelling zones with a formula rather than a percentage. Table 110-5 allows 50% of lot area below 3,000 square feet; 1,500 square feet plus 37.5% of the area over 3,000 for lots from 3,000 to just under 5,000; 2,250 square feet plus 15% of the area over 5,000 for lots from 5,000 to just under 20,000; and 4,500 square feet plus 7.5% of the area over 20,000 above that (City of Portland, Title 33, Chapter 33.110, Table 110-5, as amended by Ordinance 191961 effective March 1, 2025). Expressed as a share of the lot, that formula falls from 50.0% at 2,500 square feet to 45.0% at 5,000, 30.0% at 10,000, 22.5% at 20,000, 14.4% at one acre and 10.9% at two acres. The coverage cap is steeply regressive in lot size, exactly the opposite of the setback rule, which is steeply progressive in lot size.

Combine the two and the third constraint falls out. In Portland's R5 zone the tabulated floor area ratio is 0.5 to 1 for a lot with one dwelling unit and 0.8 to 1 for four or more, the maximum height is 30 feet, and the coverage cap follows Table 110-5 (City of Portland, Title 33, Chapter 33.110, Table 110-4). Divide the floor area the ratio allows by the footprint the coverage cap allows and you get the number of stories the code's own numbers require in order to use the full ratio. At 0.8 to 1 that quotient is 1.78 stories on a 5,000 square foot lot, 2.67 on a 10,000 square foot lot, 3.00 at 12,857 square feet and 3.56 at 20,000 square feet. Three stories is about what fits inside a 30-foot height limit at a ten-foot floor-to-floor, so somewhere around 12,900 square feet the tabulated floor area ratio stops being reachable and the coverage cap plus the height limit takes over. Below that lot size the ratio binds; above it the ratio is a number that cannot be spent.

That is the general result, and it is not specific to these two cities. Fixed-dimension setbacks are regressive on small lots and negligible on large ones. Percentage coverage caps are constant unless the code makes them a formula, and where it does they are usually regressive on large lots. Floor area ratio is flat in lot size and therefore binds in the middle of the distribution, where neither of the other two has taken over. Any screen that reports a single binding constraint for a whole district is reporting the constraint that binds at the district's median lot, and saying nothing about the tails.

A worked example: San Francisco, stage by stage

Run the sequence on real parameters. The lot is the modal San Francisco lot: 2,500 square feet, which is the single most common recorded lot area on the roll year 2025 file, appearing on 11,541 records, ahead of 2,495 square feet on 11,186 records, 3,000 on 10,905 and 2,996 on 10,402 (San Francisco Office of the Assessor-Recorder, roll year 2025, queried August 24, 2026). A 2,500 square foot lot is a 25-foot frontage on a 100-foot depth. The district is RH-2, Residential House Two Family, which carries 35,993 records on the same file, second only to RH-1.

Note what the roll cannot confirm. Across all nineteen roll years on that dataset, 2007 to 2025, and all 3,934,467 records, the lot depth field reaches 100 feet or more on 3 records and the lot frontage field reaches 100 feet or more on 38, and in eighteen of the nineteen roll years the maximum recorded depth is 99.96 feet and the maximum recorded frontage 99.39 feet. The modal recorded lot area is the product of a 25-foot frontage and a 100-foot depth, and the file's own depth field never reaches 100. The dimensions have to come from the parcel geometry, not from the assessment record, which is the practical form of the frontage and depth dependency documented in the sibling analysis of land metrics.

Now the stages.

1. Gross parcel area: 2,500 square feet.
2. Rear yard. Section 134(c) sets the basic rear yard in RH Districts at 30% of total lot depth, never less than 15 feet. Thirty percent of 100 feet is 30 feet, which clears the floor. Section 130(b) requires it to run the full width of the lot, so it removes 25 times 30, or 750 square feet. Remaining: 1,750 square feet.
3. Front setback. Section 132 matches the requirement to the adjacent building with the shortest front setback and Section 132(d) caps it at 10 feet. At the cap it removes 25 times 10, or 250 square feet. Remaining: 1,500 square feet, which is 60.0% of the lot. Where no adjacent building establishes a setback, this stage removes nothing and the footprint stands at 1,750 square feet, or 70.0%.
4. Lot coverage. Section 134(l) applies its 60% cap to development lots of two acres or more. This lot is 0.057 acres, so no percentage coverage cap applies and the footprint stays at 1,500 square feet.
5. Height. Sections 250(b) and 260(a) govern. The height and bulk layer's two most common values are a 40-foot limit: 2,305 polygons at one 40-foot designation and 1,886 at another, against 67 at 45 feet and 39 at 65 feet (San Francisco Planning Department, Zoning Map Height and Bulk Districts, updated August 23, 2026, queried August 24, 2026). At a 40-foot limit and a ten-foot floor-to-floor, four stories. Four times 1,500 is 6,000 square feet.
6. Floor area ratio. Table 124 tabulates 1.8 to 1 for RH-2, which on a 2,500 square foot lot is 4,500 square feet. Section 124(b) exempts residential uses in the R, RC, NC and Mixed Use Districts from the floor area ratio limits, so for a residential project on this lot the tabulated ratio does not apply at all (San Francisco Planning Code Table 124 as last amended by Ordinance 245-25 effective January 12, 2026, and Section 124(b)).

Read the result carefully, because this is where a floor area ratio screen fails in the most misleading way available. Such a screen returns 4,500 square feet. The envelope at three stories is 4,500 square feet. The two agree, and the agreement is a coincidence: one number is a legal limit that has been switched off for this use, and the other is the product of a rear yard, a front setback and a height limit. At four stories the envelope is 6,000 square feet, a third more than the ratio screen reported, and the ratio screen has no mechanism to notice.

Run the same code on a two-acre development lot and the binding constraint changes. Take 87,120 square feet as a 300-foot frontage on a 290.4-foot depth in an RM-1 District, which Section 134(c) also puts in the 30% rear yard group. The rear yard is 87.1 feet running the full width, removing 26,136 square feet. A 10-foot front setback removes 3,000 more. The setback footprint is 57,984 square feet, or 66.6% of the lot. Section 134(l) now applies and caps coverage at 60%, or 52,272 square feet, so the coverage rule binds and the setback geometry is slack by 5,712 square feet. Reshape the same two acres as a 1,089-foot frontage on an 80-foot depth and the arithmetic flips: the rear yard falls to 24 feet, the setback footprint falls to 50,094 square feet or 57.5%, and the setbacks bind while the coverage cap goes slack. Same district, same code, same acreage, opposite binding constraint, decided entirely by depth.

Two caveats travel with these numbers and both belong in any report that uses them. First, the product of footprint and stories is not gross floor area as Section 102 defines it, because that definition excludes basements and cellars used for building services, rooftop mechanical areas and qualifying underground accessory parking, and measures from the exterior faces of exterior walls in most districts and along the glass line four feet above the finished floor in the C-3, Central SoMa and Van Ness Special Use Districts. Second, the hosted code carries a currency disclaimer, so every parameter above is cited with the ordinance number and effective date under which it was read.

What codes do that a subtraction cannot model

The five-stage sequence assumes that every rule either removes land or caps a ratio. Real ordinances contain at least four kinds of rule that do neither, and a screen that does not recognize them will produce a number with the wrong sign of error.

Maximum setbacks. Portland's Commercial and Mixed Use zones set a minimum building setback of none at the street lot line, and a maximum building setback of 10 feet, rising to 20 feet on selected civic corridors, in all six of the CR, CM1, CM2, CM3, CE and CX zones (City of Portland, Title 33, Chapter 33.130, Table 130-2, as amended by Ordinance 191961 effective March 1, 2025). The rule pushes the building toward the street rather than away from it. A screen that subtracts setbacks from the lot will subtract nothing here and be right by accident; a screen that treats the 10-foot figure as a minimum will subtract land the code requires you to build on.

Geography-dependent parameters. The same Portland table sets maximum building coverage at 85% in the CR, CM1 and CE zones and 100% in CM2, CM3 and CX inside the Inner pattern area, and at 75% and 85% respectively in the Eastern, Western and River pattern areas, with the pattern areas identified on a separate map. Coverage in a given zone therefore has two values and the parcel's location decides which applies. New Hampshire's atlas records a related conditionality directly: 941 of its 3,784 districts, or 24.9%, are flagged as having zoning that varies with the availability of water or sewer utilities, and 462 of the 2,856 districts with both figures carry a larger minimum lot size under worst conditions than under ideal conditions, with the statewide mean rising from 1.64 acres to 1.85 acres (New Hampshire Zoning Atlas, queried August 24, 2026; counts and means computed by MMCG).

Bonuses and transfers. Portland's Table 130-3 raises the maximum floor area ratio in the CM2 zone from a base of 2.5 to 1 to 4 to 1 with bonuses, in CM3 from 3 to 1 to 5 to 1, and in CX from 4 to 1 to 6 to 1, with defined increments for inclusionary housing, affordable commercial space and planned development, and separate provisions for floor area transferred from historic resources. San Francisco runs the same mechanism in the other direction: Section 123(c)(1) caps gross floor area on a C-3-O lot at 18 to 1 even with transferred development rights, against a tabulated 9.0 to 1 in Table 124. A tabulated ratio is a base, not a ceiling, and the gap between the two can be the whole project.

Definitional exclusions. Portland's Section 33.130.205 states that outside the CR zone floor area excludes structured parking meeting an electric-vehicle charging condition up to a maximum of 0.5 to 1, long-term bicycle parking not located in a dwelling unit up to a maximum of 0.5 to 1, and indoor common area used to meet the required residential outdoor area standard, and it states that adjustments to the maximum floor area ratios are prohibited. Up to a full point of ratio can therefore sit outside the numerator while the ratio itself is unadjustable. Multiplying a published ratio by a lot area does not reproduce that.

The layers that remove land before the code does

Zoning decides what may be built on the land that is available. Three federal layers decide how much land is available in the first place, and they operate before the envelope arithmetic starts.

Flood. The binding rule is not the flood zone but the floodway. Under 44 CFR 60.3(d)(3) a participating community must prohibit encroachments, including fill, new construction, substantial improvements and other development, within the adopted regulatory floodway unless hydrologic and hydraulic analyses performed in accordance with standard engineering practice demonstrate that the encroachment would not increase flood levels during the base flood discharge. Before a floodway is designated, 44 CFR 60.3(c)(10) limits development in Zones A1-30 and AE to what will not raise the base flood elevation more than one foot anywhere in the community. Floodway land is therefore not buildable land in any practical sense, and it should be removed from the parcel before the setbacks are applied rather than noted as a risk factor afterward. The National Flood Hazard Layer covers more than 90% of the United States population (Federal Emergency Management Agency, National Flood Hazard Layer page, last updated April 3, 2025), while about 15% of the maps are more than fifteen years old (U.S. Government Accountability Office, December 1, 2021), and the statutory duty in 42 U.S.C. 4101 is only to assess the need to revise flood hazard areas at least once every five years. The mechanics of reading the layer are the subject of the companion piece on FEMA NFHL flood zones.

Wetlands. The National Wetlands Inventory holds more than 37 million wetland and deepwater features and is released twice a year, and its own metadata sets the terms on which it may be used: the data are intended for use with base maps and imagery at 1:12,000 or smaller, the primary intended use is regional and watershed display and analysis rather than specific project analysis, and there is no attempt in either the design or the products of the inventory to define the limits of proprietary jurisdiction of any federal, state or local government (U.S. Fish and Wildlife Service, National Wetlands Inventory documentation, 2024 to 2026). A wetland polygon over a parcel is a reason to order a delineation, not a subtraction from the envelope, and the difference is set out in the companion piece on wetlands screening before a site visit.

Terrain. Slope removes buildable land economically rather than legally, by making grading and retaining structures necessary. It also interacts with the code directly: San Francisco's Section 260(a) fixes the height measurement point at curb level for a lot level with or sloping down from the street, applied over a lot depth of up to 100 feet or to the midblock line, whichever is greater, and measures upward-sloping lots from curb level within 10 feet of the street property line and from average ground elevation at each building cross-section beyond that. On a steep lot the height limit is not one number but a series of them. The elevation product under the parcel decides whether that can be modeled at all, which is the argument in the national screening piece on terrain and slope at national scale.

The instructive comparison is between the reach of these layers and the reach of the zoning parameters. Elevation, flood and wetlands are national programs with published coverage figures in the high double digits. The dimensional standards that decide the envelope are published as data in one state. A screen assembled from public sources will therefore know more about the ground under a parcel than about the rules over it, which is the reverse of what most credit files assume. MMCG Analytics is built on federal, state and public data sources with source and vintage provenance carried on displayed values, and its analytical layers include parcels, flood, wetlands, wind risk, terrain, traffic and demographics; zoning is not among them, precisely because the parameters do not exist as a national layer to build one from.

The failure modes, ranked

Ranked by how often they bite rather than by how badly, the ways a parcel-to-buildable screen goes wrong are these.

1. The parameters are not in the data. This is the base case, not the exception. California's statewide zoning compilation publishes 568,741 polygons drawn from 535 of the state's 539 jurisdictions and carries no lot size, setback, height, floor area ratio or parking value at all; Oregon's statewide layer holds 114,823 polygons and carries the local district code on 653 of them, which is 0.6% (both reported in the companion piece on zoning data by parcel). Every envelope estimate from those layers is an estimate of a district name.

2. The geometry is missing or derived. Lot depth is absent from 81.1% of San Francisco's roll year 2025 records and lot frontage from 99.7%. Where dimensions are present they may be back-calculated from area, or the area from them, and the direction of the dependency differs by jurisdiction.

3. Overlays and disapplication rules. The rule that removes a parameter is usually in the same code as the parameter. Section 124(b) is the type specimen: Table 124 tabulates a floor area ratio for most San Francisco districts, and the very next subsection exempts residential uses in R, RC, NC and Mixed Use Districts from it, which covers 154,973 of 211,547 roll records. In New York City, 185,938 of 857,951 tax lots, or 21.7%, carry a second base district, a commercial overlay, a special purpose district or a limited height district on top of the primary district, a count reported in full in the companion piece on zoning data by parcel.

4. Irregular lots. 15.9% of New York City tax lots are flagged irregular, and it is on exactly that population that the recorded area and the recorded dimensions stop reconciling.

5. Nonconforming and vested conditions. Existing buildings are not evidence of what is now permitted. Across the 815,367 New York City lots that carry both a built floor area ratio and a maximum permitted one, 54,943, or 6.7%, are already built at or above the maximum, and 55.6% are built below half of it (MMCG tabulation from PLUTO 26v1, queried August 24, 2026). In San Francisco the same pattern appears where the tabulated ratio has been disapplied: 1,697 of the 9,851 RH-3 records with both a building area and a lot area, or 17.2%, show a built ratio at or above the 1.8 to 1 that Table 124 tabulates for the district, as do 20.4% of RM-1 records. A screen that applied the tabulated ratio would report one in six existing RH-3 buildings as impossible.

6. Split-lot zoning. 19,997 New York City tax lots, or 2.3%, are split between multiple zoning features, and 22 of the 106 zoning code values on the San Francisco roll name two districts in a single field. A split lot has no single set of parameters and no defensible single envelope.

7. Easements and private encumbrances. Discussed below, and unquantifiable from public layers by construction.

The ranking matters because the remedies differ. The first two are fixed by going to the primary source: the ordinance for the parameters, the digital tax map or a survey for the geometry. The third and sixth are fixed by reading the whole code rather than one table. The fourth and fifth are fixed by flagging the parcel out of the automated population rather than by computing harder. The seventh is not fixed at all.

Easements, covenants and the limit of every public layer

A utility easement, an access easement, a conservation easement, a party wall agreement, a recorded covenant restricting building height for a neighbor's view: each removes buildable area, none appears in a parcel file, a zoning layer or a hazard layer, and several cannot be drawn even when the document is in hand.

The profession that does locate them says so in its own standard. The Minimum Standard Detail Requirements for ALTA/NSPS Land Title Surveys exist, in their stated purpose, because title insurers are asked to insure title without exception as to the many matters which might be discoverable from survey and inspection, and which are not evidenced by the public records. Section 4 requires that the surveyor be provided with the most recent title commitment or other title evidence satisfactory to the title insurer, together with the recorded easements, servitudes and covenants burdening the property, because the surveyor cannot be expected to find them otherwise. Section 6.C.ii then requires the plat to carry a note for each right of way or easement whose location cannot be determined from the record document, and a separate note where the interest is a blanket easement over the whole parcel (American Land Title Association and National Society of Professional Surveyors, 2026 Minimum Standard Detail Requirements for ALTA/NSPS Land Title Surveys, adopted October 2025, effective February 23, 2026, superseding the 2021 edition).

Read those two provisions together and the boundary of a public-records screen is clear. The class of encumbrances that reduce buildable area includes members whose location is not determinable from the record, and members that burden the entire parcel. No amount of public data closes that gap, because the gap is not a data availability problem. It is what the recording system does and does not require to be located.

The practical rule for a screen is therefore to state an envelope as a figure before encumbrances, and to say so in the same sentence rather than in a footnote. That convention keeps the screen honest and puts the title commitment where it belongs, in the sequence of a data-first due diligence checklist rather than at the end of it.

What a lender does with an envelope

The envelope earns its place in a credit file by answering one question quickly: does the proposed project fit inside what the code permits on this parcel, with margin. That is a screening question, and it has three legitimate answers.

The project fits with room to spare. The envelope exceeds the proposed gross building area by enough that no plausible correction to the inputs closes the gap. This is the common case for infill on a large lot, and it means the envelope is not the constraint on the deal. Move on.

The project fits, but only just. The proposed area sits within a few percent of the computed envelope. Given that the front setback may depend on two neighboring buildings, that the coverage cap may have a second value in another pattern area, and that the recorded lot area may itself be a product of two other fields, a few percent is inside the error of the method. This answer is a request for the survey and the planning department, not a conclusion.

The project does not fit. The proposed area exceeds the envelope by a margin that no input correction explains. This is the answer worth having early, because it means the sponsor is relying on something the base code does not give: a bonus, a transfer, a variance, a conditional use, a planned development, or a vested right in an existing nonconforming structure. Each of those is a process with a timeline and a probability, and each belongs in the credit narrative rather than in the site analysis. It is exactly the case the 30-minute pre-term-sheet site screen is designed to surface before a term sheet is drafted.

What the envelope is not is an entitlement opinion. A screen computed from a published code and a published parcel file is a statement about two documents. It is not a statement about whether the jurisdiction will issue a permit, because that depends on discretionary review, on pending amendments, on the status of any nonconforming use, and on facts about the site that no record carries. The planning department is the only party whose reading of its own code binds anyone, and the title company is the only party whose reading of the record binds anyone on encumbrances. MMCG Analytics provides the data and the analytics with the source and vintage carried on every value; the credit decision and the entitlement judgment rest with the lender and its counsel.

Two reporting conventions follow. State the binding constraint, not just the envelope, because a reader who knows that coverage binds knows immediately that lot dimensions do not matter and that a survey correction will not change the answer. And state the envelope as a range across the conditional parameters rather than a point, because on a San Francisco RH lot the front setback is legitimately 0, 10 or 15 feet depending on the block face, and the honest output is the three answers rather than the middle one.

A reproducible sequence, and where it stops

The method in order, with the provenance each step must carry.

1. Resolve the jurisdiction and the district. Intersect the parcel polygon, not a centroid, with the local zoning layer and with every overlay, special district and height district layer the jurisdiction publishes. Record the layer name, the publisher and the update date. Centroid errors are a real source of misassignment, as the work on the Census geocoder against commercial services sets out.
2. Pull the geometry and test it. Record lot area, frontage and depth, then check whether frontage times depth reproduces the area. If it does not, or if the irregular-lot flag is set, stop treating the lot as a rectangle.
3. Read the code, not the layer. Take the district code to the ordinance and record five parameters with their section numbers, ordinance numbers and effective dates: setbacks and yards, lot coverage, height and its measurement datum, floor area ratio, and any subsection that disapplies one of them for the proposed use.
4. Subtract the constraint layers. Remove floodway, and flag rather than remove wetlands and steep slope. Record the layer vintage in each case; an effective flood map is a statement about a map's age as much as about a parcel's risk.
5. Compute both pairs. Footprint from setbacks and footprint from coverage; floor area from height and floor area from ratio. Report the minimum of each pair and name which one bound.
6. State the range. Where a parameter is conditional on neighbors, utilities, pattern area or use, compute the envelope at each value and report the set.
7. Label it as pre-encumbrance. Say in the same sentence that easements, covenants and vested rights are outside the scope of any public-records screen.

Where the sequence stops is as important as where it runs. It stops at the boundary of the base code: bonuses, transfers, variances, conditional uses and planned developments are outside it by construction, and a screen that silently includes them is not a screen. It stops at the boundary of the record: encumbrances whose location is not determinable from the record document cannot be modeled. It stops at the boundary of the data: a district with no published parameters produces no envelope, and reporting one anyway is the single most common failure in this work. And it stops at the boundary of discretion: nothing in a public file predicts how a planning commission will exercise judgment.

Inside those boundaries the method is worth the time it takes. It is reproducible, every number in it carries a section and a date, and it answers the one question a lender needs answered before anything else is commissioned: whether the project the sponsor described can physically and legally sit on the parcel the sponsor owns. The companion pieces on parcel data options, on what assessor rolls record about buildings and on public-records hazard screening cover the inputs; this one covers what to do with them.

Frequently asked questions

How do you calculate buildable area from a parcel?

Take the gross parcel area, subtract the yards and setbacks the zoning district requires, cap the remaining footprint at the lot coverage limit, multiply the surviving footprint by the number of stories the height limit allows, and cap the resulting floor area at the district's floor area ratio if that ratio applies to the proposed use. Setbacks and coverage are alternative constraints on the footprint and only one binds; height and floor area ratio are alternative constraints on floor area and only one binds. The useful output is not the envelope alone but the name of the constraint that produced it.

Which constraint usually binds, setbacks, coverage or floor area ratio?

It depends on lot size, and predictably so. Fixed-dimension setbacks consume a share of the lot inversely proportional to its dimensions, so they bind on small and narrow lots. Percentage coverage caps are flat unless the code writes them as a formula, and where it does they are usually regressive, so they bind on large lots. Floor area ratio is flat in lot size and therefore binds in the middle. In Portland's R5 zone the tabulated ratio binds on a 5,000 square foot lot and becomes unreachable somewhere around 12,900 square feet, where the coverage cap and the 30-foot height limit take over. In San Francisco's RH and RM Districts, with a 30% rear yard and a 10-foot front setback, the crossover against the 60% coverage cap that applies to development lots of two acres or more sits at exactly 100 feet of lot depth.

Do public zoning datasets include setbacks and floor area ratio?

Almost never. Statewide compilations publish the district polygon and a standardized class, and the dimensional standards stay in the ordinance text. California's statewide zoning files carry no lot size, setback, height, floor area ratio or parking value across 568,741 polygons. The exceptions found for this analysis are New York City, whose PLUTO file carries a maximum allowable residential, commercial and community facility floor area ratio per tax lot, and New Hampshire, whose statewide zoning atlas codes minimum lot size, front, side and rear setbacks, frontage, coverage, parking, height and floor area ratio for 3,784 districts across 263 jurisdictions. Both are exceptions, and neither is federal.

Is the floor area ratio in a zoning table the maximum you can build?

Not usually. It is a base, and codes move it in both directions. San Francisco tabulates 9.0 to 1 for the C-3-O District and separately caps gross floor area on a C-3-O lot at 18 to 1 even with transferred development rights. Portland raises the CM2 base of 2.5 to 1 to 4 to 1 with bonuses, and excludes qualifying structured parking, long-term bicycle parking and certain indoor common area from the floor area that counts. And a tabulated ratio can be switched off: Section 124(b) of the San Francisco Planning Code exempts residential uses in the R, RC, NC and Mixed Use Districts from the floor area ratio limits altogether, which removes the tabulated ratio from 73.3% of the parcel records on the city's roll year 2025 file.

Why does the assessor's lot area not match the survey?

Because in many jurisdictions the recorded lot area is not a measurement. New York City's Department of Finance calculates lot area by multiplying lot frontage by lot depth for every tax lot not flagged as irregular, and goes to the digital tax map only for irregular lots. Across Kings County, 93.8% of lots reproduce the recorded area from frontage times depth to within 5%, but only 49.0% of the lots flagged irregular do. Assessment records are built for valuation, not for envelope arithmetic, and the two uses have different tolerances.

What removes buildable land before zoning does?

Regulatory floodway is the strongest case: 44 CFR 60.3(d)(3) requires participating communities to prohibit encroachments in the adopted floodway, including fill and new construction, unless engineering analysis shows no increase in base flood levels. Wetlands are a flag rather than a subtraction, because the National Wetlands Inventory states in its own metadata that it is intended for regional and watershed analysis rather than specific project analysis and does not define any agency's regulatory jurisdiction. Slope removes land economically rather than legally, and it also changes how height is measured: San Francisco measures height from curb level on a level or downward-sloping lot and from average ground elevation at each building cross-section beyond ten feet on an upward-sloping one.

Can a public-records screen account for easements?

No, and the professional standard for locating them explains why. The 2026 ALTA/NSPS Minimum Standard Detail Requirements exist because title insurers must deal with matters discoverable from survey and inspection that are not evidenced by the public records, they require the surveyor to be furnished with the title commitment and the recorded easements burdening the property, and they require a plat note for any easement whose location cannot be determined from the record document and for any blanket easement. An envelope computed from public records is a figure before encumbrances, and should be labeled as one.

What is the difference between a buildable envelope and an entitlement opinion?

An envelope is arithmetic on two published documents, a code and a parcel record, and it can be reproduced by anyone with the same inputs. An entitlement opinion is a judgment about what a jurisdiction will approve, and it depends on discretionary review, pending amendments, the status of nonconforming uses and site facts that no public record carries. The planning department is the only party whose reading of its own code binds anyone. A screen is worth running first precisely because it is cheap and reproducible, and it is worth labeling clearly for the same reason.

Sources

1. San Francisco Office of the Assessor-Recorder, Assessor Historical Secured Property Tax Rolls, DataSF dataset wv5m-vpq2, 3,934,467 rows across roll years 2007 to 2025, roll year 2025 current as of June 26, 2026: record counts, field population for lot area, lot depth, lot frontage and property area, lot area and lot depth distributions, zoning code distribution and built ratio counts, queried August 24, 2026. https://data.sfgov.org/resource/wv5m-vpq2.json
2. San Francisco Planning Code, Article 1, Section 102, Definitions: floor area ratio, gross floor area and its measurement conventions and exclusions, lot, and the height cross-reference to Section 260, 2026. https://codelibrary.amlegal.com/codes/san_francisco/latest/sf_planning/0-0-0-49051
3. San Francisco Planning Code, Article 1.2, Sections 123 and 124 with Table 124: basic floor area ratio limits by district, the Section 123(c)(1) ceiling of 18 to 1 on a C-3-O lot with transferred development rights, and the Section 124(b) exemption of residential uses in R, RC, NC and Mixed Use Districts; Table 124 last amended by Ordinance 245-25, effective January 12, 2026. https://codelibrary.amlegal.com/codes/san_francisco/latest/sf_planning/0-0-0-18102
4. San Francisco Planning Code, Article 1.2, Sections 130 to 132: front setbacks and rear yards extending the full width of the lot with depth measured at right angles, the front setback matched to the adjacent building with the shortest setback, and the Section 132(d) caps of 10 feet and 15 feet; Section 132 last amended by Ordinance 10-26, effective March 16, 2026. https://codelibrary.amlegal.com/codes/san_francisco/latest/sf_planning/0-0-0-62918
5. San Francisco Planning Code, Article 1.2 Section 134 (basic rear yard of 30% or 25% of lot depth subject to a 15-foot floor, lot coverage in C Districts, and the 60% and 65% limits for development lots of two acres or more) and Article 2.5 Sections 250 and 260 (height and bulk limits and the curb-level measurement rule), 2026. https://codelibrary.amlegal.com/codes/san_francisco/latest/sf_planning
6. City and County of San Francisco Planning Department, Zoning Map Zoning Districts, DataSF dataset 3i4a-hu95, updated August 23, 2026: polygon count, distinct district count, and the population and content of the code section and ordinance link fields, queried August 24, 2026. https://data.sfgov.org/resource/3i4a-hu95.json
7. City and County of San Francisco Planning Department, Zoning Map Height and Bulk Districts, DataSF dataset h9wh-cg3m, updated August 23, 2026: height district value distribution, queried August 24, 2026. https://data.sfgov.org/resource/h9wh-cg3m.json
8. New York City Department of City Planning, Primary Land Use Tax Lot Output (PLUTO), version 26v1, released May 28, 2026: tax lot counts, lot area, frontage and depth population, irregular lot and split zone flags, lot type distribution, and built against maximum allowable residential, commercial and community facility floor area ratio, queried August 24, 2026. https://data.cityofnewyork.us/resource/64uk-42ks.json
9. New York City Department of City Planning, PLUTO Data Dictionary, August 2022 edition (22v2): the lot area calculation rule, the irregular lot code, the lot type code list, and the definitions of built and maximum allowable floor area ratio, read August 24, 2026. https://www.nyc.gov/assets/planning/download/pdf/data-maps/open-data/PLUTODD.pdf
10. City of Portland, Title 33 Planning and Zoning, Chapter 33.110, Single-Dwelling Zones: Table 110-4 development standards (floor area ratio by unit count, maximum height, front, side and rear setbacks) and Table 110-5 maximum building coverage by lot size, as amended by Ordinance 191961, effective March 1, 2025. https://www.portland.gov/code/33/100s/110
11. City of Portland, Title 33 Planning and Zoning, Chapter 33.130, Commercial/Mixed Use Zones: Table 130-2 development standards (maximum floor area ratio, base height, minimum and maximum building setbacks, maximum building coverage by pattern area, minimum landscaped area), Table 130-3 bonus floor area ratio and height, and Sections 33.130.205 and 33.130.220, as amended by Ordinance 191961, effective March 1, 2025. https://www.portland.gov/code/33/100s/130
12. New Hampshire Zoning Atlas, Saint Anselm College Initiative for Housing Policy and Practice with New Hampshire Housing: the dimensional standards coded for each district, including minimum lot size, front, side and rear setbacks, minimum frontage, maximum lot coverage, minimum parking, maximum height, floor area ratio, minimum unit size and maximum density; includes zoning changes adopted through June 1, 2025, read August 24, 2026. https://www.anselm.edu/about/offices-centers-institutes/finance-administration/office-partnerships/initiative-housing-policy-and-practice/new-hampshire-zoning-atlas
13. New Hampshire Zoning Atlas, NH Atlas Zoning Districts Buildable feature service: full field list, 3,784 district features across 263 jurisdictions, setback and minimum lot statistics, and the population of the floor area ratio, lot coverage, height and utility-conditionality flags, queried August 24, 2026. https://services1.arcgis.com/aguSsLS841Hp3EC4/arcgis/rest/services/NH_Atlas_Zoning_Districts_Buildable/FeatureServer/0
14. National Zoning Atlas (Land Use Atlas, Inc.), How the National Zoning Atlas is Made, including the published method How To Make a Zoning Atlas 2.0 (2024) and the list of regulatory characteristics coded per district, read August 2026. https://www.zoningatlas.org/how
15. California Governor's Office of Land Use and Climate Innovation, California Statewide Zoning North and South feature services: 568,741 polygons from 535 of the state's 539 jurisdictions, field lists carrying no dimensional standard, published 2026, queried August 24, 2026 and reported in full in the companion article on zoning data by parcel. https://services8.arcgis.com/BcP3FB8Ua7DeCr93/arcgis/rest/services/California_Statewide_Zoning_North__19162237743447578111/FeatureServer/0
16. Oregon Department of Land Conservation and Development, statewide Zoning feature service, ArcGIS Online item created July 19, 2023, data from 229 local jurisdictions as of June 30, 2023: 114,823 polygons with the local district code populated on 653, queried August 24, 2026. https://services8.arcgis.com/8PAo5HGmvRMlF2eU/arcgis/rest/services/Zoning/FeatureServer/0
17. 44 CFR 60.3, Flood plain management criteria for flood-prone areas, paragraphs (c)(10) and (d)(3): the prohibition on encroachments in the adopted regulatory floodway and the one-foot rise limit before a floodway is designated; 41 FR 46975, October 26, 1976, as amended, read August 24, 2026. https://www.law.cornell.edu/cfr/text/44/60.3
18. 42 U.S.C. 4101, Flood elevation determinations and flood hazard mapping: the duty to assess the need to revise and update floodplain areas and flood risk zones at least once every 5 years, 2023 edition. https://www.law.cornell.edu/uscode/text/42/4101
19. Federal Emergency Management Agency, National Flood Hazard Layer: digital flood hazard data covering more than 90 percent of the United States population, page last updated April 3, 2025. https://www.fema.gov/flood-maps/national-flood-hazard-layer
20. U.S. Government Accountability Office, WatchBlog, How Can FEMA and Flood Insurance Keep Up with Rising Flood Risks?, December 1, 2021: about 15 percent of FEMA's flood maps are more than 15 years old. https://www.gao.gov/blog/how-can-fema-and-flood-insurance-keep-rising-flood-risks
21. U.S. Fish and Wildlife Service, National Wetlands Inventory, Wetlands Data Layer and dataset metadata, 2024 to 2026: more than 37 million wetland and deepwater features, twice-yearly release, the 1:12,000 scale limit, the regional and watershed intended use, and the statement that the inventory does not define the limits of proprietary jurisdiction of any federal, state or local government. https://www.fws.gov/program/national-wetlands-inventory/wetlands-data
22. U.S. Geological Survey, 3D Elevation Program, About 3DEP Products and Services and Topographic Data Quality Levels, 2018 to 2026: 3DEP-quality data available or in production for 48 percent of the nation as of May 2018, acquisition reaching 89 percent national coverage by the end of 2022, and the QL0, QL1 and QL2 specifications. https://www.usgs.gov/3d-elevation-program/about-3dep-products-services
23. American Land Title Association and National Society of Professional Surveyors, 2026 Minimum Standard Detail Requirements for ALTA/NSPS Land Title Surveys, adopted October 2025 and effective February 23, 2026, superseding the 2021 edition: the statement of purpose on matters not evidenced by the public records, the Section 4 records research requirements, and the Section 6.C plat note requirements for easements whose location cannot be determined from the record document and for blanket easements. https://www.alta.org/topics/topic-land-survey-standards
24. National States Geographic Information Council, 2025 Geospatial Maturity Assessment, published August 14, 2026 from results collected in the second half of 2025: state-level digital parcel coverage and public access grades, as tabulated in the companion article on the state of United States parcel records. https://nsgic.org/wp-content/uploads/2026/08/2025-GMA-Full-Report-20260814.pdf
25. MMCG Research, Buildable Envelope Series: the San Francisco roll year 2025 field population, lot area and depth distributions and built ratio counts; the New York City built against permitted floor area ratio distribution, irregular lot, split zone and lot type tallies; the New Hampshire setback and minimum lot statistics and their application to square lots; the San Francisco setback against coverage crossover arithmetic; and the Portland coverage share and story-requirement arithmetic. All computed August 24, 2026 from sources 1 to 24; MMCG database, 2026. https://mmcganalytics.com/methodology/