Why this article exists (before we look at any charts)
In the System Baseline edition of The Reliability Illusion, we established a simple but critical shift:
Vessel schedule reliability doesn’t break at arrival.
It breaks inside the cargo receiving window.
That window - defined by the Earliest Return Date (ERD) and the CY Cutoff - is where export execution actually happens.
This edition applies the same framework to Yang Ming (YMLU) to understand how cargo receiving windows behave in practice - where they stay predictable, and where they don’t.
All terms used here are defined in the Reliability Series - Methodology Appendix:
https://www.tradelanes.co/blog/reliability-series-methodology-appendix
Data scope (Yang Ming sample)
This analysis is based on an observational system sample of executable export port-calls and is not a statistically randomized sample.
- Port-calls: 118
- Vessels: 45
- Ports: 7
- Carrier: YMLU (Yang Ming)
Filters applied:
- ERD and CY Cutoff both required
- Drift >40 days treated as data error and excluded
Section 1 - How often do Yang Ming receiving windows actually move?
A receiving window is considered moved if either ERD or CY Cutoff shifts by one calendar day or more from its originally published value.
Figure 1 - Receiving Window Stability (Yang Ming)
- Stable receiving windows: 35.59%
- Moved receiving windows: 64.41%
Plain English meaning:
For Yang Ming in this sample, receiving window movement is the majority state. That doesn’t automatically mean something is “wrong” - but it does mean exporters should expect plans to require re-validation more often than not.
Section 2 - Drift isn’t chaos; it has a shape
Drift measures how far ERDs or CY Cutoffs move between original and final values, expressed in calendar days.
Figure 2 - ERD Drift Distribution (Yang Ming)
- 0-1 day: 51.69%
- 1-2 days: 8.47%
- 2-3 days: 7.63%
- 3+ days: 32.20%
Plain English meaning:
Yang Ming shows a meaningful tail: nearly one in three port-calls experienced 3+ days of ERD drift. That tail is where execution pain concentrates.
Static buffers are built for the middle of the curve.
Operational pain lives in the tail.
Section 3 - ERD vs CY: where Yang Ming risk concentrates
Across this sample, ERD drift is higher on average than CY drift, even though CY still exceeds ERD at the 1-day threshold.
Figure 3 - ERD vs CY Drift (Yang Ming)
Average drift:
- Mean ERD drift: 2.11 days
- Mean CY drift: 1.76 days
Threshold comparison:
- ≥1 day drift: ERD 48.31% vs CY 49.15%
- ≥2 days drift: ERD 39.83% vs CY 28.81%
- ≥3 days drift: ERD 32.20% vs CY 22.03%
Plain English meaning:
This is a useful reminder that “the constraint” can differ by carrier. In this sample, ERD movement becomes the dominant driver as drift grows (2+ and 3+ days), while CY remains the more common issue at the 1-day threshold.
Operationally, that often feels like:
- early acceptance dates shifting enough to force replans, and
- cutoffs still moving often enough to create late friction.
Section 4 - Timing matters more than averages
A late-stage change is defined as a change to ERD or CY Cutoff that occurs within the final 72 hours before the receiving window opens.
Figure 4 - Late-Stage Receiving Window Changes (Yang Ming)
- ERD changed in last 72 hours: 33.05%
- CY Cutoff changed in last 72 hours: 22.03%
Plain English meaning:
Roughly one in three ERDs and about one in five CY Cutoffs changed inside the final 72 hours. That is the execution lock-in problem: the plan can look workable until close-in changes reduce options.
So far, we’ve looked at how windows move.
Next, we look at where.
Section 5 - Volatility is not evenly distributed across ports (Yang Ming)
The Port Volatility Index (PVI) reflects how quickly static planning assumptions break at a port.
Figure 5 - Port-Level Drift (Yang Ming)
Below are Yang Ming’s port environments in this sample, ordered by PVI. Ports with very small sample sizes should be interpreted cautiously.
USSAV (PVI 10.0, n=14)
- Mean ERD drift: 4.71 days
- Mean CY drift: 4.57 days
- Stable window rate: 7.14%
- ERD late-stage change: 35.71%
- CY late-stage change: 28.57%
What this feels like:
This is where volatility concentrates for Yang Ming in this dataset. Both ERD and CY move far enough that static buffers stop being reliable. Execution becomes a re-validation workflow.
USNYC (PVI 5.3, n=9)
- Mean ERD drift: 1.56 days
- Mean CY drift: 3.60 days
- Stable window rate: 33.33%
- ERD late-stage change: 33.33%
- CY late-stage change: 22.22%
What this feels like:
A “CY-dominant” port environment. ERDs move, but CY moves more and can break plans late.
USTIW (PVI 3.2, n=62)
- Mean ERD drift: 2.38 days
- Mean CY drift: 1.32 days
- Stable window rate: 30.65%
- ERD late-stage change: 41.94%
- CY late-stage change: 14.52%
What this feels like:
ERD volatility dominates here, and late-stage ERD change is high. Exporters can be forced into replans earlier than expected.
USLAX (PVI 0.4, n=25)
- Mean ERD drift: 0.79 days
- Mean CY drift: 0.82 days
- Stable window rate: 56.00%
- ERD late-stage change: 12.00%
- CY late-stage change: 28.00%
What this feels like:
More predictable overall, with CY timing still worth watching closer to execution.
USOAK (PVI 0.0, n=6)
- Mean ERD drift: 0.23 days
- Mean CY drift: 0.33 days
- Stable window rate: 66.67%
- ERD late-stage change: 33.33%
- CY late-stage change: 33.33%
What this feels like:
Small sample, but a common pattern: low average drift does not guarantee low late-stage risk.
USORF (PVI 9.4, n=1) and USCHS (PVI 0.3, n=1)
These ports have one port-call each in this sample. They should not be over-interpreted.
Section 6 - Severity still exists, even when averages look manageable
Figure 6 - Top 10 Highest-Severity Yang Ming Events
Top examples:
- YM TRILLION (USSAV): ERD 18d, CY 18d, late-stage 1, drift score 37
- YM TRAVEL (USSAV): ERD 9d, CY 8d, late-stage 2, drift score 19
- YM WARRANTY (USNYC): ERD 6d, CY 10d, late-stage 2, drift score 18
Plain English meaning:
These are stress tests, not typical shipments. They show how quickly drift can stack when multiple changes coincide, especially in the highest-volatility port environments.
Static buffers fail in these scenarios by design.
Section 7 - The KPI that matters for Yang Ming
Figure 7 - Receiving Window Movement Rate (Yang Ming)
- Moved receiving windows: 64.41%
- Stable receiving windows: 35.59%
- Scope: 118 port-calls • 45 vessels • 7 ports
Plain English meaning:
For Yang Ming in this sample, movement is the norm. The operational risk is driven by both magnitude (the 3+ day tail) and timing (late-stage changes).
Section 8 - Why static buffers fail (and why this repeats)
Figure 8 - Static Buffer vs Dynamic Time Buffer (DTB)
Plain English meaning:
When drift has a long tail and late-stage changes are common, fixed buffers are routinely exceeded. Planning must adapt to observed behavior, not assumptions.
Before we move to the next carrier
A vessel can be “on time” and still break export execution if the receiving window shifts underneath it.
This Yang Ming edition shows:
- receiving window movement is the majority state in this sample,
- ERD drift becomes the dominant driver at higher thresholds (2+ and 3+ days), and
- volatility is highly port-dependent, with USSAV carrying the strongest signal.
Methodology and definitions:
Reliability Series - Methodology Appendix
https://www.tradelanes.co/blog/reliability-series-methodology-appendix
Next in the Carrier Reliability Series
Maersk - publishing soon.
Leave a Comment