This dashboard reproportions 2026 Point-in-Time (PIT) count data of the adult unsheltered population from the Los Angeles Homeless Services Authority (LAHSA) down to the area within a buffer around LA 2028 Olympic venues, so that the totals shown are an estimate of the unsheltered population near venues.

LAHSA’s PIT Count is an annual single-night count of people experiencing homelessness in the Los Angeles Continuum of Care (CoC). The CoC covers all cities and unincorporated areas in Los Angeles County except Pasadena, Long Beach, and Glendale, which conduct and report their own counts independently. This report uses the 2026 adult unsheltered count, released by LAHSA in July 2026.

Los Angeles’ PIT Count is available at the census tract level and records individuals sleeping on the street, as well as units of vehicles (i.e., cars, vans, and RVs) and tents or makeshift shelters where people are believed to be sleeping. To get from units to people for the vehicle and tent/makeshift shelter counts, we apply occupancy multipliers to estimate the number of people represented by those units (more details below).

Because the PIT Count is reported only at the census tract level, we do not know the exact locations where unsheltered people are sleeping within each tract. At the same time, an Olympic venue’s security buffer may cover only part of a tract or nearly all of it. To estimate how many people fall within the buffer, we use the street network as a proxy for where unsheltered people are likely to be found. More detail about our street-length reproportioning methodology is detailed below.

Since no PIT observation is actually tied to a specific location within a tract, these are estimates; the method rests on an assumption about how people are distributed within a tract: uniform along walkable street length.

We note that our estimates focus on unsheltered adults ages 25 and older, plus family members counted on the street. This follows how LAHSA categorizes this population in the PIT Count, but it also means our estimates do not capture everyone who could be affected by venue-related security policies. In particular, the unsheltered youth population are excluded from these totals. Our figures presented should therefore be understood as a conservative estimate of the number of unsheltered people who could be affected.

Lastly, it is not yet public how officials will define the actual security perimeter around each venue—from the full site boundary, as we do here, or from a single reference point. We use site polygons because a real perimeter is more likely to follow a venue’s physical footprint than a single coordinate, but this choice raises the estimate substantially relative to a point-based buffer of the same radius; treat the figures in this report as reflecting that assumption rather than a confirmed perimeter methodology.

Report tab: narrative summary of the estimates, broken out by geography, dwelling type, and demographics.

Map tab: shows unit-level PIT counts by census tract (Total, Rough Sleepers, Vehicles, Tents/MSS), reproportioned within a 1-mile buffer of official Olympic venues. 0.5-mile and 2-mile buffer rings are shown for visual reference.

Totals tab: aggregates estimates up to the CoC, City of LA, Service Planning Area (SPA), City of LA Council District, and Supervisorial District level, broken out by individual count type (rough sleepers, cars, vans, RVs, tents, makeshift shelters), searchable across all three buffer distances (0.5, 1, and 2-miles).

Data on the location of the Olympic Game Venues was retrieved from the LA County Planning GIS Team (last downloaded August 14, 2026). To approximate venue footprints as polygons rather than points, boundaries were constructed using two sources from LA County eGIS: parcel/building footprints from the County Assessor office, and the Countywide Parks and Open Space layer (both last downloaded August 14, 2026). Each venue point was matched to the corresponding building or park polygon based on location and name, and that polygon was used to represent the venue’s spatial extent in place of a single coordinate. This approach captures the actual footprint of each venue (accounting for venues that span large areas, such as parks or multi-building complexes) for use in buffer and overlap calculations.

Reference buffers of 0.5, 1, and 2 miles were constructed independently around each venue polygon using a fixed-distance buffer applied directly to each venue’s footprint boundary. Because buffers are drawn from the polygon edge rather than a central point, the effective catchment area reflects each venue’s actual size and shape. Where venues were close enough for their buffers to overlap, the individual buffer polygons were dissolved into a single unioned buffer to avoid double-counting overlapping area in tract-level estimates.

Street network data are from Road Segments data from the Countywide Address Management System (CAMS) Program. This file contains a classification code on type of road (primary, secondary, etc.). Before any buffer or reproportioning math happens, every street segment is categorized into “potential unsheltered locations” based on its road-type:

• Include: primary and secondary roads
• Exclude: highway, bike path, motorway/truck trail, alley, driveway, pedestrian walkway, ramp, trail, unpaved road, minor, freeway, private road, railroad, rapid transit, planned road

Only streets classified as “include” are used as the denominator/numerator in the street-length reproportioning method below. Excluded road types are not counted as potential unsheltered locations.

Street-length-weighted reproportioning: For each tract, the fraction of its qualifying street length that falls inside the venue buffer is calculated, and that same fraction of its PIT count is attributed to the buffer. Only streets classified as “include” count toward that fraction (see the street classification note). This assumes unsheltered individuals are distributed evenly along “walkable” street length within the tract. Tracts with little qualifying street length get a low estimate even if they have substantial area overlap with the buffer, and tracts with no qualifying street length inside a given buffer contribute zero to that buffer’s estimate regardless of how much of the tract’s area the buffer covers.

For example, a tract with a PIT Count of 40 people and 50% of its qualifying streets inside the buffer contributes an estimated 20 people to that buffer; a tract with the same count but 100% of its qualifying streets inside the buffer contributes the full 40.

The map shows unit-level PIT counts for Vehicles and Tents/MSS (i.e., the number of cars/vans/RVs and the number of tents/makeshift shelters observed), while Rough Sleepers reflect person counts (i.e., individuals aged 25+ counted on the street and in safe parking sites; individuals aged 18-24 in safe parking sites; and family members counted on the street and in safe parking sites). Total PIT Count is the total number of rough sleepers (persons) + Vehicle units + Tent/MSS units.

Vehicles (cars, vans, RVs) and Tents/Makeshift Shelters are counted during the PIT street count as physical units. To estimate people from these unit counts, each unit type (cars, vans, RVs, tents, makeshift shelters) is multiplied by an occupancy multiplier unique to that type and the region. Region-specific multipliers were used for the CoC, City of LA, and each SPA; Council District estimates borrow the City of LA multipliers, and Supervisorial District estimates borrow the CoC-wide multipliers. Rough sleepers, on the other hand, do not need a multiplier; unsheltered individuals observed directly on the street are counted as people to begin with.

Because multipliers are not provided at the tract-level, the Map shows unit counts for vehicles and tents/makeshift shelters. Each count type (Total, Rough Sleepers, Vehicles, Tents/MSS) is mapped as its own toggleable choropleth layer using Jenks natural breaks, at the 1-mile/street-length reproportioned estimate.

The multiplier-based people estimates for Vehicles and Tents/MSS are available in the Totals tab.

The Totals tab reports the street-length buffer estimates aggregated to five levels: the entire LA Continuum of Care (CoC), the City of LA only, by Service Planning Area (SPA 1–8), by City of LA Council District (CD), and by Supervisorial District (SD). The top-line summary are fixed to the 1-mile buffer scenario and shows both the concentration share (% of that geography’s own unsheltered population) and the contribution share (% of the countywide total). The detail table underneath shows the raw unit estimate and, where a multiplier applies, the converted people estimate, and is searchable by Buffer Distance to compare the default 1-mile estimate against alternatives.

We estimate how much each zone of venues (e.g., Downtown LA, Exposition Park, Venice Beach) contributes to the total affected population. These venue zones correspond to those defined by the LA2028 Olympic Committee; however, in this report Dodger Stadium is kept as its own even though it sits within the broader DTLA zone, since its buffer doesn’t overlap the other downtown venues.

We report two complementary numbers for each zone, at each buffer distance:

• Solo: the estimated affected population within that zone’s buffer alone, as if it were the only venue zone in the analysis. Because nearby zones’ buffers can overlap (e.g., Downtown LA and Exposition Park sit close enough that their one- and two-mile buffers overlap), solo estimates for different zones can double-count the same people, so that summing solo estimates across every zone would overstate the true total.
• Unique: the marginal contribution of that zone (i.e., how much the total estimated affected population would fall if that zone’s venues were removed from the analysis entirely, holding every other venue in place). This nets out any catchment area/overlap shared with other zones.

For most venue zones, solo and unique are close, because the zone’s buffer doesn’t meaningfully overlap any other zone’s. They diverge most for zones that sit near each other – in this case, primarily Downtown LA (DTLA) and Exposition Park zones, whose buffers overlap enough that removing either zone on its own only marginally reduces the combined total.

Pasadena and Long Beach venues are excluded from this calculation, since LAHSA’s PIT count data doesn’t cover those cities.

• Reproportioned counts are estimates, a geometric approximation based on the assumptions above.
• The street classification used is itself a judgment call based on road type, not validated against actual PIT observation locations.
• Occupancy multipliers are themselves estimates with their own margin of error; the people-level figures inherit that uncertainty on top of the geometric reproportioning uncertainty.
• The LAHSA 2026 PIT Count data used in this analysis reflects adults aged 25 and older; it does not include youth ages 18-24, who are counted separately by LAHSA’s youth-specific count methodology. As a result, these estimates do not capture the full unsheltered population near Olympic venue sites, and true totals (particularly for venues near areas with higher concentrations of transition-age youth) are likely higher than reported here.
• All figures reflect the confirmed 2028 Olympic game venues as of this report’s last update (August 14, 2026). As venue plans and security perimeters are finalized, these estimates should be revisited.

• Randall Kuhn, UCLA Fielding School of Public Health.
• Jessie Chien, UCLA California Center for Population Research.

• Benjamin Henwood, USC Center for Homelessness, Housing and Health Equity Research.
• Amanda Landrian Gonzalez, USC Center for Homelessness, Housing and Health Equity Research.

Special thanks to Scot Hickey (USC Center for Homelessness, Housing and Health Equity Research) for their research support, Katheryn Leifheit (UCLA Fielding School of Public Health) for their thoughtful feedback, and the USC Homelessness Policy Research Institute (HPRI) for providing research resources.