Washington DC · annual report

Washington DC Crime Rate — 2025 in Review

A year of crime trends, summarized.

An annual companion to the monthly briefings: the anomalies that mattered, the structural shifts that emerged, and where the model got it right (or wrong). 12 briefings, condensed.

01

The big picture

Washington DC closed 2025 with 23,934 bucketed incidents down 17.4% against 28,987 the year before. 756 tracked signals were raised across 12 briefings — 61 spikes, 665 drops + sustained shifts, and 6 rare-event / streak-break signals.

The monthly volume chart at right shows where the year was busy and where it was quiet, against the prior-year monthly average (dashed line). The categories that moved most are broken out below.

FIG 1.1 · MONTHLY INCIDENT VOLUME · 2025VS 2024
640127919192559JanFebMarAprMayJunJulAugSepOctNovDecprior year monthly avg
−17.4%total volume vs 2024
23,934total incidents
756signals tracked
8baselines reset
69%forecast accuracy
41 / 41neighborhoods covered
03

Crime by category

All ten categories, ranked by 2025 total volume.

#CategoryYear totalYoYTrendNote
01Other Larceny11,525−10%Down 10% vs 2024. baseline shift flagged (5 months).
02Theft from Vehicle5,307−20%Down 20% vs 2024. baseline shift flagged (8 months).
03Motor Vehicle Theft3,917−23%Down 23% vs 2024. baseline shift flagged (4 months).
04Robbery1,326−37%Down 37% vs 2024. baseline shift flagged (6 months).
05Aggravated Assault930−9%Down 9% vs 2024. 88 below-trend months.
06Burglary719−28%Down 28% vs 2024. 2 below-trend months.
07Homicide111−35%Down 35% vs 2024. no flags raised.
08Sexual Assault99−26%Down 26% vs 2024. 5 rare-event flags.
05

Crime forecast scorecard

Of 96 monthly point-estimate forecasts issued for 2025, 66 (69%) landed inside their 95% prediction intervals. Below: month by month.

Jan
7/8
Feb
7/8
Mar
7/8
Apr
8/8
May
7/8
Jun
3/8
Jul
5/8
Aug
5/8
Sep
4/8
Oct
4/8
Nov
4/8
Dec
5/8
Inside 95% CI Outside 95% CI (model miss)

Per-bucket coverage, MAPE, and bias details live on the methodology page.

06

Methodology updates

Logged inline with the code that runs the model.

  • 2025

    10-bucket NIBRS-aligned categories

    Replaced an earlier 6-bucket scheme (which collapsed homicide, robbery, aggravated assault, and sexual assault into one “violent” bucket). Each bucket now maps to FBI UCR Part 1 / NIBRS Group A — the cross-city common denominator for adding new cities.

    VIEW DETAIL →
  • 2025

    Sustained-shift Poisson rate-ratio test

    Added a Poisson Z-test (|Z|>2.576, ratio differs by ≥25%) for sustained shifts between recent vs prior 12-mo windows — distinct from the spike/drop signals which compare against the multi-year baseline.

    VIEW DETAIL →
  • 2025

    Prophet forecasts with low-count gating

    Per-(neighborhood, bucket) forecasts now skip cells averaging <2 incidents/month over the trailing 24 months. Violent-bucket forecasts skip at the neighborhood level and surface via rare-event / streak-break signals instead.

    VIEW DETAIL →
07

What we'll watch in 2026

3 distinct patterns from 2025we expect to keep moving — drawn from the year's recurring sustained signals, not the single-month spikes already covered above.

  1. 01

    Walter Reed · other larceny

    The past 12 months saw 27 incidents — about 437% above the 5 average from prior years. Surfaced in 8 of 2025's 12 briefings — the persistence is what puts it on the watch list.

  2. 02

    Cleveland Park · other larceny

    The past year logged 299 incidents — down 37% from 476 the year before.

  3. 03

    Twining · robbery

    The past year logged 66 incidents — down 57% from 155 the year before.

END OF REPORT · WASHINGTON DC · 2025

Cite as: Public Analyst.ai, “Washington DC2025in review,” auto-generated annual report. Permanent URL: /dc/2025/year-in-review.

All briefings →Methodology →