The Open Book

Every published call, replayed against the weekly-close record. The calls that could not be replayed are listed too — the point is the whole book, not the good half.

£10,000 following every call since 7 August 2026: £9,900.

The same money in the Nasdaq-100: £9,944.

The book

-1.0%

Since the start date

the Nasdaq-100

-0.6%

Same money, same weeks

Deepest fall

-4.8%

Peak to trough, weekly

Called right

4/5

Graded calls only

The curve

Contributions

What each call contributed to the book
Call	Ticker	Weeks	Weight	Move
NVIDIA: paying a market multiple for a franchise financing its own demand	NVDA	5	1	+10.6%
Micron: priced like a cyclical at the top, contracted like a utility	MU	5	1	+8.6%
Lumentum: The Optics Play That Bridges Power and Memory	LITE	5	1	+1.2%
TSMC: own the bottleneck, not the brand	TSM	5	1	+0.2%
Broadcom: the right custom-silicon thesis at the wrong price	AVGO	5	Not counted	-8.7%
SK Hynix: the largest HBM rent, valued like a Korean cyclical	000660.KS	5	1	-25.6%

Not replayable

Calls the replay could not include, and why
Call	Why
KV Cache Compression Will Cut Inference Costs 50% Before HBM4 Ships at Scale	No ticker
Transformer Successors: State-Space Models and Linear Attention Are Worth Watching	No ticker
Nuclear Is the Only Power Source That Scales for AI Data Centers	No ticker
Energy Ownership Is the New AI Moat	No ticker
BTC Miners Pivoting to AI Infrastructure: The Power Arbitrage	No ticker
Inference Costs Will Fall 90% by 2028 — The Memory Stack Is the Mechanism	No ticker
Parameter Scaling Is Dead. Multi-Dimensional Scaling Is the New Paradigm.	No ticker

How this is computed

One close per ticker per Friday, from the stored record — weekly closes — intra-week drawdowns are not visible. Each call is measured in percent from its own entry price, so a non-dollar ticker contributes like any other and no exchange rate is applied anywhere; the notional multiplies the final percentage only. A week where a close was missing holds the previous one and is dotted on the curve. This is a paper replay of published views — it is not advice, and it is not what any account actually did.