Once a week, the 90-day outlook of every card that has one is written down as shown. Ninety days later each is judged against the price the card actually reached. Nothing is edited afterwards and nothing is left out.
Nothing has been judged yet. The first cohort comes due on 2027-01-11.
| Logged on | Rule | Cards | Judged on | Judged |
|---|---|---|---|---|
| 2026-11-03 | outlook-90d.v1 | 5,412 | 2027-02-01 | 0 |
| 2026-10-27 | outlook-90d.v1 | 5,335 | 2027-01-25 | 0 |
| 2026-10-20 | outlook-90d.v1 | 5,270 | 2027-01-18 | 0 |
| 2026-10-13 | outlook-90d.v1 | 5,152 | 2027-01-11 | 0 |
Each card is placed in one of five groups by its score. The ranges come from a test on one year of prices (October 2025 to October 2026) and assume the market as a whole stays flat.
| Group | Says | Likely range (half) | Wide range (8 in 10) |
|---|---|---|---|
| Lowest fifth | Likely to lag the market | −14% to +3% | −22% to +17% |
| Second | Leaning behind the market | −10% to +10% | −17% to +27% |
| Middle | In line with the market | −8% to +14% | −15% to +33% |
| Fourth | Leaning ahead of the market | −6% to +19% | −14% to +38% |
| Highest fifth | Likely to lead the market | −6% to +24% | −15% to +46% |
The price is read on the day the 90 days end, for the same printing the outlook was made on. A card with no price by then is counted as such and kept in the record. “The market” is the median growth of every card logged the same day, so the cards are their own control: every eligible card is logged and there is no selection to flatter.
When enough outlooks have been judged, the ranges cards really landed in replace the test’s ranges, as a new numbered rule. The old rule’s record stays here as it was. How the outlook works.
Count this visit? Google Analytics tells us which pages are worth keeping. Saying yes stores a cookie that links your visits together. Saying no still counts the page, just nothing that identifies you — and the site is identical either way. What we store.