Independent quantitative research
Research archive.
Eight validated city-level machine-learning studies, organized into four matchups. Each publication separates Zillow’s underlying ZHVI history from ChrionML® dataset construction, QA, modeling, evaluation and interpretation.
8 cities published642 monthly observations352 rolling backtests64-city program
Research library
Four matchups · Eight researched cities
Featured research · Week 04Published · October 14, 2026
Brooklyn vs. Philadelphia
Brooklyn · 11201
$1,308,132
+0.16% forecast
Philadelphia · 19107
$257,048
-0.29% forecast
Brooklyn leads forecast index value; Philadelphia has lower scale-adjusted backtest error (0.170% vs. 0.285%).
October 2026 · Holt damped trend / Holt damped trend
Explore Week 4 box score →Featured research · Week 03Published
Back Bay vs. Hanover
Back Bay · 02116
$1,300,581
+0.12% forecast
Hanover · 03755
$521,201
+0.00% forecast
Back Bay leads forecast index value; Hanover leads historical appreciation (+57.5% vs. +10.77%).
September 2026 · Drift / Last Value Baseline
Explore Week 3 box score →Featured research · Week 02Published
Cambridge vs. Ithaca
Cambridge · 02139
$888,570
+0.19% forecast
Ithaca · 14850
$289,770
+0.00% forecast
Cambridge leads forecast index value; Ithaca leads historical appreciation (+29.5% vs. +17.8%).
September 2026 · Drift / Last Value Baseline
Explore Week 2 box score →Featured research · Week 01Published
Boston vs. Providence
Boston · 02114
$779,306
+0.06% forecast
Providence · 02903
$438,308
+0.44% forecast
Boston leads forecast index value; Providence leads historical appreciation (+53.2% vs. +4.9%).
September 2026 · Drift / Drift
Explore Week 1 box score →Latest ML projects
Each city is evaluated independently before matchup-level comparison.
Week 4 · Final · October forecast
Brooklyn vs Philadelphia
Brooklyn leads forecast index value; Philadelphia has lower scale-adjusted validation error.
Read the publication →Project 05 · Back Bay 02116
Highest September forecast in the archive
Drift narrowly improves on Last Value across 44 chronological backtests.
$1.301MForecast
+0.12%Momentum
DriftWinner
Project 06 · Hanover 03755
Strong trajectory with a baseline win
Last Value has the lowest error; observed growth reaches 57.50% since January 2020.
$521KForecast
+57.50%Since 2020
Last ValueWinner
Projects 03–04 · Week 2
Cambridge vs Ithaca
Cambridge leads value and month-ahead momentum; Ithaca leads observed cumulative appreciation.
Read Week 2 →Projects 01–02 · Week 1
Boston vs Providence
Boston leads forecast value; Providence leads projected momentum and observed appreciation.
Read Week 1 →Published matchups
Publication-style summaries for technical and nontechnical readers.
Research standard
Comparable evidence without pretending every market behaves the same.
The protocol stays consistent while model selection remains local.
80–81Months per city
44Backtests per city
5Candidate models
1Shared protocol
