My laptop died. The portfolio didn't notice.

My laptop died early in the week. The screen went dark mid-sentence, the fans cut out, and the machine refused to come back. I dusted off an old tower from the closet and rebuilt my desk on it.

What I lost wasn't the site — that lives on GitHub, and Vercel was happily serving it the whole time I was hunched over a fifteen-year-old box. What I lost was the wiki. Every page of synthesized notes from every investing book I'd read, every paragraph of strategy reasoning I'd written down, every cross-reference I'd built between authors. None of it was backed up.

The data I could replace — Wealthsimple holds the trade ledger, Stock Rover holds the screeners' raw inputs. The thinking, I had to rebuild from memory.

What rebuilding taught me

Rebuilding the Council from scratch is the kind of accident I wouldn't have signed up for and now wouldn't trade.

The Council is my name for the synthesized strategy spec — the document that turns everything I've read about investing into one verdict per holding. v1 took me weeks. v2 took me three days. And v2 came out tighter — the threshold I'd argued for in v1 was the threshold I argued against by paragraph four of v2, because I could finally see which rules had been load-bearing and which were just decoration.

The bigger surprise: rebuilding the Council forced a rebuild of the Blue Portfolio holdings tables. The old verdict logic was grading every position against the old strategy, and I couldn't ship the v2 Council without also re-grading every line on the dashboard against it. So I did both. Every Blue holding now carries a Council-aligned verdict (Verified Buy / Buy-Ready / Conviction Hold / Warming Up / Wait / Reject), plus three little chips showing which of my screeners — Bedrock (large-cap stability), Prospector (strict-quality conviction), Quantum (strict-momentum smaller-cap) — currently surfaces the name.

The point of the chips is overlap. A name passing two screeners is a stronger signal than one. A name passing all three is the strongest signal the screener layer can produce. A name I hold that passes zero is a question worth asking myself before next Saturday.

For the first time I can look at my own book and see, at a glance, how the positions I'm carrying line up with the rules I've written down. Some of them don't. That's the diagnostic the dashboard is supposed to give me, and until this week it didn't.

How the portfolio did

Down −0.39% on the week. A quiet, slightly red five days.

The two best names were both ones I've held for a while and nothing happened to either of them this week — BDT (Bird Construction, a Canadian builder) up +9.07% and CNQ (Canadian Natural Resources, an oil sands producer) up +8.52%. The worst was K.TO (Kinross Gold, a Canadian gold miner) down −9.51% on no news I can find. It's a Council Conviction Hold for now.

The two gold miners I bought last week — NEM and AEM — gave back about −6% each. That was the one cluster I was watching closely after last week's note, and the answer is less of a moment than it looked on Friday. With K.TO also off, the whole gold corner of the book had a soft week. I'm not adding or trimming. Positions still inside their first two weeks deserve more rope than that.

What I'm watching

Whether the Council's verdicts and my gut keep agreeing. The first week I had this on the dashboard, every position I'd have called by feel matched what the chips said. That's reassuring and also slightly suspicious — the test is the first name where they disagree. I want to be wrong at least once before I trust this fully.

A back-office week

It's the first weekly note I've ever written where the most important thing that happened isn't a trade. That's a feature, not a bug. The dashboard is the proof. The writing is the interpretation. The week the laptop died was the week the dashboard finally started doing the interpretation for me.

— Mark

— Mark