Expectations about New York City's congestion pricing program, written down before the data arrived and scored automatically once it did.
Scored Nov 12, 2026 at 06:26 UTC. Page built Nov 12, 2026 at 06:27 UTC.
Registry at commit 3a53f61.
5Held
7Pending
1Not scoreable
Expert's record on prospective expectations: none of the 7 has been scored yet. Their data is not published. This line will report how many held against the confidence the expert stated.
How to read this
An expectation is a written statement of what a metric should do over a set of dates, with a range. It is recorded in a registry that can only be added to. When the data for its dates is loaded, the pipeline computes the observed value and gives a verdict.
The expected ranges here come from an AI agent acting as a subject matter expert, interviewed and checked by the engineer. The full interview is public. In real use the experts would be people.
Held
The observed value fell inside the expected range.
Violated: below
The observed value fell below the expected range.
Violated: above
The observed value fell above the expected range.
Pending
The data for this window is not published or loaded yet.
Data issue
A data check on this expectation's inputs failed, so the result is not trusted.
Not scoreable
No data this project holds shares the expectation's baseline. It is shown for context.
Prospective
Recorded before the data for its window was published.
Derived
An expert's translation of an official projection onto a metric this project measures.
Official
Published by an agency before the outcome.
Retrospective
Recorded after the data was seen. It checks the pipeline or marks what to watch. It does not test a forecast.
Where the gap comes from
For expectations about a change between two periods, the gap between observed and expected is split by slices such as service or entry point. Every slice is assumed to have been expected to move by the same percentage. This shows where a gap sits. It does not say why.
Prospective expectations
Recorded before the data for its window was published.
PendingProspective
In the week of November 8 to 14, 2026, about 4% fewer vehicles will enter the Congestion Relief Zone than in the matching week of 2025.
Expected -7.50% to -1.00%
The data for this window is not published or loaded yet. Data is loaded through 2026-10-31.
No data yet for Nov 8, 2026 to Nov 14, 2026. The comparison window has 7 days behind it. Solid line: the window. Dashed line: the comparison window.
Assumes baselines use Spark's exact percentile (linear interpolation), as the spec requires. Day of week was requested as a driver but is not a dimension.
Derived expectations
An expert's translation of an official projection onto a metric this project measures.
HeldDerived
If the official forecast's logic held, taxi and ride-hail trips to or from the zone in tolling's first year would be roughly flat against the year before.
Observed +1.72%, expected -2.50% to +4.00%
The observed value fell inside the expected range.
Daily tlc_cbd_touch_trips from Jan 5, 2025 to Jan 3, 2026: 364 days, between 149,927 and 464,845. The comparison window has 364 days behind it. Solid line: the window. Dashed line: the comparison window.
Gaps are in the metric's unit over the whole window. When the result lands near the middle of its range the total gap is small, so shares are large and can have opposite signs.
This window's data was loaded and largely published when this was recorded, so a hit says nothing about forecasting skill. Score it as a test of the projection's logic. The expert's own estimate, given published data, is +0.5% to +3.0%. Not counted toward calibration.
Official expectations
Published by an agency before the outcome.
Not scoreableOfficial
The $9 Phase 1 toll reduces daily vehicles entering the Manhattan CBD by 13.4% relative to No Action.
No data this project holds shares the expectation's baseline. It is shown for context. The public entries data begins at launch, and no action is a modeled counterfactual for a 2023 analysis year, so no series this project observes shares the projection's baseline. the mta's own headline counts also add excluded roadways to zone entries..
A consistency check with no evidential weight, because the value was known when recorded.
HeldRetrospective
The pipeline reproduces the MTA's figure that yellow and ride-hail trips starting or ending in the zone rose about 0.8% from January to September 2024 to the same months of 2025.
Observed +0.79%, expected +0.30% to +1.40%
The observed value fell inside the expected range.
Daily tlc_cbd_touch_trips from Jan 1, 2025 to Sep 30, 2025: 273 days, between 202,062 and 414,107. The comparison window has 274 days behind it. Solid line: the window. Dashed line: the comparison window.
Gaps are in the metric's unit over the whole window. When the result lands near the middle of its range the total gap is small, so shares are large and can have opposite signs.
Reproduces a published figure to validate the pipeline; not a forecast. If the yellow and ride-hail reproductions pass but this misses, check what the unfiltered metric includes.
HeldRetrospective
The pipeline reproduces the MTA's figure that yellow taxi trips starting or ending in the zone rose about 17% from January to September 2024 to the same months of 2025.
Observed +16.71%, expected +15.90% to +17.90%
The observed value fell inside the expected range.
Daily tlc_cbd_touch_trips from Jan 1, 2025 to Sep 30, 2025: 273 days, between 53,035 and 129,715. The comparison window has 274 days behind it. Solid line: the window. Dashed line: the comparison window.
Gaps are in the metric's unit over the whole window. When the result lands near the middle of its range the total gap is small, so shares are large and can have opposite signs.
Reproduces a published figure to validate the pipeline; not a forecast.
HeldRetrospective
The pipeline reproduces the MTA's figure that ride-hail trips starting or ending in the zone fell about 5% from January to September 2024 to the same months of 2025.
Observed -4.86%, expected -5.60% to -4.20%
The observed value fell inside the expected range.
Daily tlc_cbd_touch_trips from Jan 1, 2025 to Sep 30, 2025: 273 days, between 142,609 and 293,809. The comparison window has 274 days behind it. Solid line: the window. Dashed line: the comparison window.
Gaps are in the metric's unit over the whole window. When the result lands near the middle of its range the total gap is small, so shares are large and can have opposite signs.
Reproduces a published figure to validate the pipeline; not a forecast.
Data health
The latest result of each data check. A check is a claim about the data, such as "this column is never empty". A failing check on an expectation's inputs turns its verdict into a data issue.
Check
Status
Pass
Warn
Fail
Detail
cbd_zone_membership
Warning
0
1
0
2025: 1,823,293 of 1,922,810 trips that start and end in the same in-zone zone carry the fee (2025-01-05 to 2025-12-31)
dropoff_before_pickup
Pass
90
0
0
dropped_share
Pass
90
0
0
entries_days_complete
Pass
22
0
0
fee_before_launch
Pass
1
0
0
fee_presence
Pass
90
0
0
freshness
Pass
3
0
0
regression_totals
Pass
6
0
0
required_nulls
Pass
90
0
0
schema_required_columns
Pass
90
0
0
silver_volume
Pass
84
0
0
Freshness
How far each source is loaded against what it has published.
Source
Status
Detail
crz_entries
Pass
Loaded through 2026-10-31; published through 2026-10-31