The desk runs its own numbers and publishes its calls next to the account that trades them. The short route:
RESEARCH — OUTSIDE READING
Reports & research
Outside research the desk actually reads. The reports, papers and primary data behind the theses. None of it is ours. All of it moved a position, or failed to and told us something anyway.
From the field — 2026
This year's outside reading: quarterly disclosures, market-structure data and the structural shifts we're tracking in 2026. Newest first.
Not for confirmation. A report that agrees with a position we already hold is worth almost nothing. It is the most expensive kind of comfort. We read outside work for three things: a number we cannot generate ourselves (unit economics from inside an industry, a central bank's own model), a mechanism we have not thought through, and the strongest available case against what we believe.
The institutions here are chosen deliberately, and they disagree with each other. The IMF models transitions as costs to be absorbed. McKinsey models them as markets to be captured. a16z is talking its book and is often the earliest to be right anyway. Nous publishes work that is either a real cost curve bending or an elegant dead end, and you cannot tell which from the abstract. Reading them against each other is the point. A consensus among these four would be more alarming than the disagreement.
No single shop owns crypto research. The serious desks read all of them. top traders sits on the reader’s side of the table: we curate the outside work and add what it implies for positioning.
Tool
Best for
Notable
Price
top traders
Curated outside research plus the desk read on top — what the report implies for positioning
Desk notes, live dashboard, buy-side desk framing
$500/yr $0during the alpha launch
Messari
Research library and asset profiles
Broad, well-structured coverage; Pro paywall
$5,000/yr
Delphi Digital
Institutional deep dives
Some of the deepest sector work in crypto; paid
Paid
Galaxy Research
Macro and market structure
Serious macro pedigree; free
Free
The difference: the others produce the research. We tell you which of it survives contact with a real position.
FAQ
What is on the top traders research page?
A curated reading list of outside research the desk actually uses — reports and primary data from sources like the IMF, McKinsey and a16z on AI, robotics, crypto and market structure. None of it is ours; all of it moved, or failed to move, a real position.
How does the desk use outside research?
As primary sources behind theses, not as headlines. The desk reads the report itself, checks it against on-chain and flow data, and keeps only what changes a position or usefully fails to.
What does top traders research cost?
The list price is $500/yr; access is $0 with a promo code — leave your email and we send it. The reading list and the desk’s own notes stay open, with links to the original sources.
What this batch adds up to
Read together, these reports describe the same transition from four incompatible angles, and the disagreement is the useful part.
Nous is the one to watch most closely, because it is the only one publishing falsifiable engineering rather than forecasts. If inference cost keeps bending the way Lighthouse Attention suggests, and if training really can be coordinated across underutilised hardware the way Psyche claims, then the assumption underneath most AI investment (that frontier capability requires concentrated capital) weakens. That is not a small revision. It is the difference between an oligopoly and a commodity, and it decides whether the margin sits with the labs or with the infrastructure underneath them.
The IMF is the useful pessimist. It refuses to treat the adjustment as free, and it is the only institution here with an incentive to say so out loud. McKinsey supplies what nobody else will: actual unit economics for humanoids, with the bottlenecks named. Whether the timing is right matters less than the fact that the constraints are enumerated. That is a checklist you can hold a company against.
The desk read: take the mechanisms, discard the timelines. Every one of these documents is more reliable about how something works than about when it happens, and the market pays for the first and punishes overconfidence in the second. Where they conflict, and they do, we size smaller and wait for the data to arbitrate.