COE Prices by Month: What Historical Seasonality Can Tell You
Compare monthly Category A premiums within each year, understand the limits of the data, and avoid treating a seasonal average as a buying signal.
A month-by-month comparison can describe the archive. It cannot tell you what your next COE will cost. Price cycles, quota changes and holidays can move together, making an apparently cheap month look more predictable than it is.
Compare months within the same year
Pooling nominal premiums across decades gives expensive years more influence. The table below first compares each month with its own year’s average, then averages those percentage differences across eligible years. It uses Category A, so its results should not be presented as findings for every category.
| Month | Complete years included | Mean deviation from that year |
|---|---|---|
| January | 10 | -5.48% |
| February | 10 | -6.88% |
| March | 10 | -2.46% |
| April | 10 | 3.36% |
| May | 10 | 2.74% |
| June | 10 | 0.85% |
| July | 10 | -2.38% |
| August | 10 | 2.31% |
| September | 10 | 1.45% |
| October | 10 | 3.86% |
| November | 10 | 2.45% |
| December | 10 | 0.18% |
Only years with all twelve months represented are included. Each monthly mean is divided by that same year’s exercise-weighted mean; deviations are then averaged equally across years. This descriptive statistic does not control for within-year price trends, quota changes or moving holidays and does not establish statistical significance or a future buying advantage.
A negative percentage means that month was below the corresponding annual benchmark on average in the included years. It does not mean every observation fell below the benchmark, or that a buyer could have predicted the annual average at the time.
Read the sample count
The calculation includes only years with all twelve months represented. This matters for 2020: bidding was suspended from April through June, so the year is excluded by that rule. A represented month is not proof that every expected exercise has been captured; check the underlying archive when reproducing the study.
What the comparison does not establish
A sustained rise within a year can make early months appear cheap and late months expensive without a repeatable calendar effect. Chinese New Year also moves between January and February. Quota announcements use allocation windows that do not match calendar quarters.
This table contains no confidence intervals, controls for those influences or test on future unseen data. Earlier claims that particular months were statistically proven bargains have been withdrawn. No tested trading or buying advantage is established here.
Use it alongside an actual purchase decision
Record the package you can buy now, the cost of waiting and your deadline. Check whether a future premium reduction would change that package under its rebate terms. A lower auction result is not necessarily an equal reduction in your invoice.
For research, repeat the comparison over different time windows and inspect individual years. A pattern that disappears when the period changes deserves less confidence. For a purchase, use the exercise results and a complete written quote rather than a calendar slogan.
Sources and review date
Reviewed on 13 September 2026. Historical results retain their exercise dates; worked budgets are assumptions, not quotations.