Track record does the model beat the anchor?
Every week the model's fleet TCE and dividend for each company's open quarters are frozen, and beside them the anchor: a simple estimate from the company's own booked rate and payout ratio (how it is built). When the quarter reports, both are scored against the realized TCE and the declared dividend. A model that cannot beat that anchor has not earned its complexity, and this page says so either way.
Where it stands
The model's dividend has been scored for 5 company-quarters so far. It was 0.0% off for CCEC and NORDEN, 10.1% for KCC.OL, 40.0% for STNG and 56.1% for ASC: 2 within 10%. None can yet be compared with the anchor. These early records do not identify their dividend method.
366 weekly forecasts frozen so far, 42 of them with an anchor beside them. 5 company-quarters can be scored today: 5 with the model, 0 with both the model and the anchor.
Model against anchor: not yet scored. Forecasts frozen before the anchor was added to the record carry none and are never given one after the fact. Each weekly forecast frozen since freezes its anchor beside it, so the first comparison comes when a quarter reports after a week with an anchor on file.
By dividend path
The path is the method that produced the model's dividend in the week scored. Errors are a percentage of the outcome; a hit is a forecast within 10% of it.
| Dividend path | Model DPS | Anchor DPS | DPS: model beat anchor | Model TCE | Anchor TCE | TCE: model beat anchor | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Mean abs. error | Hit rate | n | Mean abs. error | Hit rate | n | Mean abs. error | Hit rate | n | Mean abs. error | Hit rate | |||
| Method was not recorded for these early forecasts | 5 | 21.2% | 40% | 0 | not yet scored | not yet scored | 3 | 7.2% | 67% | 0 | not yet scored | not yet scored | ||
9 other paths with no scored quarter yet
- Stated payout policy applied to our earnings estimate
- Share it has actually paid, applied to its named payout base
- Share it has actually paid, applied to cash above break-even
- Held flat at the last declared dividend
- The company's own announcement, or a stated decision to pay none
- Payout ratio on modelled earnings (earlier method)
- Index-linked monthly forecast
- Fit of past dividends against the fleet rate (earlier method)
- Earlier cash-walk method
By company
| Company | Model DPS | Anchor DPS | DPS: model beat anchor | Model TCE | Anchor TCE | TCE: model beat anchor | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Mean abs. error | Hit rate | n | Mean abs. error | Hit rate | n | Mean abs. error | Hit rate | n | Mean abs. error | Hit rate | |||
| ASC | 1 | 56.1% | 0% | 0 | not yet scored | not yet scored | 1 | 10.3% | 0% | 0 | not yet scored | not yet scored | ||
| CCEC | 1 | 0.0% | 100% | 0 | not yet scored | not yet scored | 0 | not yet scored | 0 | not yet scored | not yet scored | |||
| KCC.OL | 1 | 10.1% | 0% | 0 | not yet scored | not yet scored | 0 | not yet scored | 0 | not yet scored | not yet scored | |||
| NORDEN | 1 | 0.0% | 100% | 0 | not yet scored | not yet scored | 1 | 8.0% | 100% | 0 | not yet scored | not yet scored | ||
| STNG | 1 | 40.0% | 0% | 0 | not yet scored | not yet scored | 1 | 3.2% | 100% | 0 | not yet scored | not yet scored | ||
Every scored forecast
| Company | Quarter | Frozen | Outcome published | Declared DPS | Model DPS error | Anchor DPS error | Realized TCE | Model TCE error | Anchor TCE error |
|---|---|---|---|---|---|---|---|---|---|
| STNG | 2026Q3 | 2026-10-21 | 2026-10-26 | $0.75 | -40.0% | — | $37,508/day | +3.2% | — |
| NORDEN | 2026Q3 | 2026-10-21 | 2026-10-28 | DKK 2.00 | +0.0% | — | $14,683/day | +8.0% | — |
| KCC.OL | 2026Q3 | 2026-10-21 | 2026-10-27 | $0.36 | -10.1% | — | $36,116/day | — | — |
| CCEC | 2026Q3 | 2026-10-21 | 2026-10-22 | $0.15 | +0.0% | — | — | — | — |
| ASC | 2026Q3 | 2026-10-30 | 2026-11-03 | $0.27 | +56.1% | — | $25,358/day | +10.3% | — |
How it is scored
- The forecast scored is the last one frozen strictly before the outcome was published. Each track is dated from the company's earliest print of that quarter's figure: a pre-announcement, the results release on the calendar, or the dividend declaration, whichever came first. A forecast frozen on that day or later is never scored on that track, and a quarter with no date on file is not scored at all. A correct forecast of no dividend counts as a hit.
- The anchor uses the company's own stated booked rate for the quarter and its own earnings margin to estimate earnings. It then applies the company's stated payout ratio, or the median share of earnings it has declared where it states none. It is frozen with the forecast from what was on file that week. Where an input is missing the anchor is left empty and the missing input is recorded; no peer figure stands in.
- The anchor is a dividend in US dollars per share, so it is not scored against a dividend declared in another currency. A model dividend is scored only in the unit the company declares in.
- Model beat anchor counts the quarters where both were scored and the model's error was smaller. A tie is not a win.
- Frozen forecasts are never revised. A score is fixed when its outcome is first seen; a later correction by the company is shown as a note. Nothing here is a re-run of today's model over past quarters.
How the weekly record is kept is on the methodology page.