Orlando Deep Dive · 2025 / 26
A collision spike, a vanishing fleet, and the one segment that tells the truth — across 263,877 Orlando rides, 2025 vs 2026.
·
A collision spike that isn't what it looks like.
This study compares Nexar fleet activity in Orlando, Florida between January–May 2025 and January–May 2026 — covering 263,877 rides, 2.97 million miles, and 143,450 hours of dashcam footage.
Orlando's collision rate rose +62.3% (464.6 → 754.2 /M; Z = 4.91, p < 0.001). The fleet simultaneously became 90.7% Taxi/Rideshare (up from 69.8%). But within that segment alone — 93,000+ rides each year — collisions doubled (+117.5%). The deterioration is real, not just compositional.
1. Collision rate +62.3% (DI 162). 2. Fleet flipped to 90.7% Taxi/Rideshare. 3. Within that segment, collisions doubled (+117.5%, DI 218) — the genuine signal. 4. Q2 2026 (939.7/M) nearly tripled Q2 2025 (DI 292). 5. Timing held: 5 PM peak unchanged.
What this study measures.
Year-over-year changes in traffic volume, safety incident rates, trip characteristics, driver-type composition, and temporal patterns in the Orlando, Florida metro area.
| Scope | Detail |
|---|---|
| City | Orlando, FL (city core) · ride origin |
| Excluded | Kissimmee, Maitland, Winter Park |
| Baseline | Jan 1 – May 31, 2025 |
| Study | Jan 1 – May 31, 2026 |
| Grain | Monthly · 5 full months each |
Consumer and Truck vehicles largely exited the Nexar-monitored Orlando fleet between 2025 and 2026. Any rate-based metric is affected. Every finding must be read through this lens.
Data source and metrics.
All data comes from the Nexar Dashcam Fleet — proprietary telemetry from dashcam-equipped vehicles, aggregated at monthly grain and filtered to Orlando starting-point city.
Rate = events ÷ rides × 1,000,000. Danger Index = (2026 ÷ 2025) × 100. Z = (p − p₀) ÷ √(p₀(1−p₀)/n).
Volume fell −22.5% — but the fleet, not the city.
Total ride volume declined −22.5% (133,456 → 103,421). Miles fell −26.3%, hours −27.3% — driven by the exit of Consumer and Truck vehicles, not a real-world reduction in Orlando traffic.
| Metric | Jan–May 2025 | Jan–May 2026 | Change |
|---|---|---|---|
| Total rides | 133,456 | 103,421 | −22.5% |
| Total miles | 1,707,570 | 1,258,810 | −26.3% |
| Total hours | 83,040 | 60,410 | −27.3% |
| Miles per ride | 12.80 | 12.17 | −4.9% |
| Hours per ride | 0.622 | 0.584 | −6.1% |
Every metric worsened.
All four incident metrics deteriorated year-over-year per million rides. Normalization by rides is essential — raw counts would mislead given the −22.5% volume decline.
| Incident Type | 2025 /M | 2026 /M | % Change | DI |
|---|---|---|---|---|
| Collision | 464.6 | 754.2 | +62.3% | 162 |
| Harsh Acceleration | 1,042 | 1,886 | +81.1% | 181 |
| Sharp Cornering | 5,522 | 8,045 | +45.7% | 146 |
| Hard Brake | 21,018 | 25,275 | +20.3% | 120 |
The deterioration is accelerating.
Splitting the window into quarters shows it is accelerating — the Q2 2026 collision rate nearly tripled its 2025 baseline.
| Incident | Q1 2025 | Q1 2026 | Q1 Δ | Q2 2025 | Q2 2026 | Q2 Δ |
|---|---|---|---|---|---|---|
| Collision /M | 567.2 | 624.5 | +10.1% | 322.1 | 939.7 | +191.8% |
| Hard Brake /M | 19,361 | 20,394 | +5.3% | 23,317 | 22,855 | −2.0% |
Q2 2026 Collision Danger Index
April–May 2026 hit 939.7 per million rides — nearly triple the Q2 2025 baseline of 322.1 (DI = 292, +191.8%). The primary driver of the overall YoY increase.
Robust and reliable.
Both headline safety metrics clear the very-highly-significant threshold by a wide margin. Neither is an artifact of the smaller 2026 ride pool.
| Metric | 2025 Rate | 2026 Rate | Z-Score | P-Value | 95% CI (2026) |
|---|---|---|---|---|---|
| Collision /M | 464.6 | 754.2 | 4.91 | <0.001 | [638.6 – 869.8] |
| Hard Brake /M | 21,018 | 25,275 | 10.84 | <0.001 | — |
Same shape, lower volume.
Both years build from the morning commute, sustain through midday, and peak at the evening commute (5 PM local). The hourly profile is consistent — the fleet change altered how safely Orlando drives, not when.
| Hour (Local ET) | 2025 Rides | 2026 Rides | YoY Change |
|---|---|---|---|
| 7 AM | 6,044 | 4,530 | −25.1% |
| 12 PM | 8,848 | 6,733 | −23.9% |
| 3 PM | 9,488 | 6,915 | −27.1% |
| 5 PM · Peak | 10,291 | 8,265 | −19.7% |
| 6 PM | 8,887 | 6,931 | −22.0% |
| 9 PM | 4,621 | 3,594 | −22.2% |
The weekday squeeze.
Weekdays remain dominant, but 2026 weekday rides declined more steeply than weekends — narrowing the gap. The pattern points back to the fleet shift.
| Day | 2025 Rides | 2026 Rides | Ride Change |
|---|---|---|---|
| Monday | 14,722 | 12,284 | −16.6% |
| Tuesday | 18,681 | 14,729 | −21.2% |
| Wednesday | 19,601 | 14,781 | −24.6% |
| Thursday | 20,958 | 15,135 | −27.8% |
| Friday | 20,930 | 15,581 | −25.5% |
| Saturday | 21,030 | 16,349 | −22.3% |
| Sunday | 17,534 | 14,562 | −16.9% |
Where the truth lives — the Taxi/Rideshare rate doubled.
Taxi/Rideshare is the only segment that held volume (+0.8%) while all others collapsed — both the dominant segment and the most deteriorated.
| Driver Type | 2025 | Share | 2026 | Share | Δ |
|---|---|---|---|---|---|
| Taxi/Rideshare | 93,106 | 69.8% | 93,853 | 90.7% | +0.8% |
| Consumer | 28,372 | 21.3% | 5,977 | 5.8% | −78.9% |
| Truck | 8,280 | 6.2% | 1,893 | 1.8% | −77.1% |
| Bus | 3,698 | 2.8% | 1,697 | 1.6% | −54.1% |
From 333.0 to 724.5 per million rides (+117.5%, DI 218). With 93,000+ rides in both years, this is reliable — Orlando's core 2026 safety concern, and the one finding that survives the composition caveat.
The fleet became 90.7% Taxi/Rideshare.
The most significant structural change is the near-complete transition to Taxi/Rideshare dominance — essential for interpreting every other metric.
The headline +62.3% blends a composition effect with a real deterioration effect. Even holding segment fixed, collisions doubled — and only the second is fixable through coaching.
Shorter and faster — on average.
Trip characteristics shifted toward shorter, faster trips in 2026 — but the segment-level breakdown reveals divergent behavior beneath the fleet average.
| Driver Type | 2025 Mi/Ride | 2026 Mi/Ride | Δ |
|---|---|---|---|
| Taxi/Rideshare | 12.02 | 11.56 | −3.8% |
| Consumer | 9.74 | 10.71 | +9.9% |
| Truck | 34.27 | 51.07 | +49.0% |
| Bus | 7.65 | 7.75 | +1.3% |
| All Types | 12.80 | 12.17 | −4.9% |
A novel finding with no benchmark.
Where published benchmarks exist, the 2025 baselines validate cleanly and the 2026 changes are explainable by the fleet shift. One finding has no comparable benchmark at all.
The 90.7% Taxi/Rideshare dominance has no published comparable benchmark. No other organization has this granular fleet-composition data at city level — a uniquely valuable external data asset.
The spike, the segment that owns it, and what held steady.
Six actions.
Monitor Taxi/Rideshare safety in Orlando closely
The segment's collision rate doubled YoY. Determine whether it reflects driver behavior, route, or data-capture changes — and alert fleet managers or rideshare partners.
Investigate the Q2 2026 collision spike
The DI 292 spike is 3×. Before external use, validate whether it's a real increase or a detection/classification change.
Disclose fleet composition in external presentations
State that the 2026 fleet is 90.7% Taxi/Rideshare. Raw rate increases without this context risk misinterpretation as a city-wide claim.
Compute fleet-composition-adjusted rates
Standardize to a fixed fleet mix and recompute both years to isolate real safety change from composition change.
Expand to full Jan–Jul once July 2026 data lands
This study covers Jan–May. Revisit for a firmer read on the Q2 trajectory.
Showcase the fleet-composition depth
The driver-type breakdown is genuinely novel — a high-value external data asset.
What this study can and can't claim.
Partial 2026 period
Covers Jan–May 2026, not the requested Jan–Jul. June–July data was not yet complete.
Fleet composition confound
The dominant confound. Raw rates alone cannot claim Orlando is "more dangerous" without the composition qualifier.
Small 2026 samples
Consumer (n=5,977), Truck (n=1,893), Bus (n=1,697) carry high uncertainty. Treat individual rates as directional.
City-level geography
Orlando = city starting-point. No street-level enrichment.
Detection model & benchmarks
Detection may differ between 2025/2026 model versions. External benchmarks are curated estimates.