Deck · Aging & accuracy
How conclusions hold up over time
Aged 12mo, still correct
61%
firm-wide
Stale queue
4
awaiting review
Best-aging source
Drill
85% correct at 24mo
Fastest decaying
EZVC
61% correct at 24mo
Accuracy at 6 / 12 / 24 months
Share of entries per source still judged correct
| Source | Entries | Correct at 6mo | Correct at 12mo | Correct at 24mo |
|---|---|---|---|---|
| Conclave | 126 | 78% | 71% | 64% |
| EZVC | 214 | 72% | 68% | 61% |
| Pulse | 102 | 88% | 74% | — |
| Northstar | 44 | 91% | 84% | 79% |
| Drill | 68 | 94% | 89% | 85% |
Firm-wide aging curve
12-month rolling accuracy
Confidence calibration
Stated confidence vs. realised accuracy
Marked high confidence74% correct
Marked medium confidence58% correct
Marked low confidence41% correct
Decay curves by source
Toggle sources to compare
Stale entry review queue
Over 8 months old and never revisited
L-355 — Fintech infrastructure thesis
8mo old · No falsifier or entry criteria ever attached.
L-341 — Manager concentration cap
10mo old · Two GP relationships have since crossed 7% concentration.
L-298 — Real estate sleeve — proptech adjacency
14mo old · Proptech deal flow has picked up materially since logging.
L-276 — Hedge sleeve convexity
13mo old · Macro regime classifier has since flagged elevated tail risk twice.