This dashboard answers whether the office has been thinking and whether it has been right. It is a record of twenty decisions rather than a picture of today. The current position is at the foot, because it is one quarter of twenty.
The realised drawdown constraint bound at 15 of 20 meetings. That is the most important fact in this record: from December 2022 the binding constraint was the fund's own drawdown against the board limit, not any signal. The signal was rarely what set position size.
Click any row for the full reasoning tabled at that meeting. Click a column heading to sort.
| Date | Decision | What moved | Signal | Regime | TE | Binding constraint | Compliance | Earned |
|---|---|---|---|---|---|---|---|---|
| z | bps | bps |
8 decisions hurt, and they are not eight unrelated bad calls. All 8 of them carried a Treasury duration position, and the sign flipped halfway through the record. 4 were duration underweight with a large cash overweight, from 2022-09 to 2023-09, while the tightening cycle peaked and the long end recovered. 4 were duration overweight funded from equity, from 2024-09 to 2026-03, after the same signals flipped positive. That is one error made twice with the sign reversed: a trend signal at a turning point, on the single line whose out-of-sample R² against the expanding mean is most negative. Five unrelated bad calls and five instances of the same bad call are different findings, and only the second is fixable. The recommendation removes the duration position entirely. The deeper cause is that the programme was sized to a tracking-error budget rather than to demonstrated skill, and a budget is permission to take risk rather than a reason to.