The Analyst, Analyst Oracle's AI investment committee personified
The Analyst Where greed is good, and supercharged by AI
A private investment committee · Powered by AI

Your own committee of legendary investors, plus a community of real ones.

Jim Simons turned investing into a science. This is that idea rebuilt for fundamental research, in the age of AI. A swarm of agents that distill legendary investor minds into an algorithm, then reason independently, argue it out, and ground every claim in real expertise.

Not financial, investment, or tax advice. No affiliation with, or endorsement by, any person or company named. The investor personas are AI interpretations of public material, not the real individuals. For research and entertainment purposes only. Investing involves risk, including the possible loss of principal. Verify every source and consult a licensed professional before investing.

Trained on decades of data, with a live feed of the latest signals
SEC EDGAR10-K / 10-QEarnings callsSell-side reportsInterviewsPodcastsX / TwitterPublic commentaryMarket driversAlt dataLive market data
The investment committee
AI Value investor persona Value
AI Quant investor persona Quant
AI Macro investor persona Macro
AI Growth investor persona Growth
AI Contrarian investor persona Contrarian
AI Rational investor persona Rational
9:41
Analyst Oracle Research Note · Sun 9:00 ET
NVDA
NVIDIA Corp · Semis
$172.40
▲ 1.8%
WATCH · accumulate on weakness
1Y · +94%
75%Gross margin
31×Fwd P/E
+94%Rev YoY
Thesis

Data-center demand still outruns supply and CUDA lock-in is intact. The real debate is the durability of hyperscaler capex through 2027, not business quality.

Entry point
Accumulate $115 to $140spot $172
Macro lens

Rate cuts into ’26 support long-duration growth multiples; watch 10Y & hyperscaler capex guides.

Committee 4 Accumulate1 Watch1 Pass
Sourced to 10-K, Q3 call & EDGAR ↳
9:41
Analyst Oracle 247 online
# trading-floor # semis # macro-desk +6
SOX +1.9% NVDA +2.4% TSM +1.1% AVGO +0.8% AMD −0.6% SMH +1.7%
M
Marcus9:36
10Y backing off into the close, risk-on. Semis leading the tape, $SOX +1.9% on the session.
P
Priya9:37
TSMC 3nm booked through ’26, supply is the bottleneck, not demand. Hyperscaler capex guides still climbing.
NVDA▲ 2.4%172.40
quant signal flipped long SMH at the open
B
Ben9:39
trimmed AAPL at 31×. rotating into AVGO, custom-ASIC tailwind looks underpriced here.
🔥 14💯 6
D
Diego9:40
the real NVDA debate is ASIC share + capex durability into ’27, not the GPU moat. watch gross margins.
Analyst Oracle
Semis Deep Dive: NVDA posted
Thesis · entry · kill-criteria ↳
Gavin is typing…
Message #trading-floor
Tap to switch the screen ↑

Let's be honest

5 things every
retail investor is quietly losing at.

1

You're up against funds with hundreds of analysts and a billion-dollar data budget. Your edge? A YouTube thumbnail, a tweet, or some random Discord chat.

2

You read one bullish thread, feel the FOMO, and buy a thesis you only ever heard one side of.

3

You'd do the homework. But who has a weekend to read a 200-page 10-K, build a complete pro forma, and listen to every CEO interview, all for one stock?

4

The "research" you do get from the internet is typically a confident take with zero sources. But dig deeper, and you find that source puts out thousands. Just one randomly went viral.

5

Everyone screams about what to buy. Nobody tells you when to sell, or when the thesis breaks.

"None of that is your fault. You've just never had the right tools. "

The Hive

Why we built this. A metaphor for investing.

One bee is nearly blind. However, the hive always finds flowers.

No single bee is in charge. No single bee can even see the whole field. And yet every day, a hive of thousands solves a problem none of them could crack alone. Which flowers, how far, and what direction? And it gets the answer astonishingly right.

A scout flies out, finds a patch of flowers, and comes home to dance. The angle of her waggle points to the sun. The length tells the hive how far and how rich. Dozens of scouts dance at once. From all those independent reports, with no vote whipped and no queen deciding, the hive commits to the best flowers in the field. It is almost always right.

Markets reward the same trick. When a prediction market like Kalshi or Polymarket embarrasses the pundits, it isn't magic. It's that independent guesses, pooled together, cancel out each other's blind spots. The crowd is wrong in a thousand directions at once. The consensus that survives is right. The instant everyone copies one loud voice, the edge is gone.

No bee runs the hive. That's the entire reason it works.

This isn't a metaphor we reached for. It's already how the best money managers on earth are built. Firms like Millennium, Citadel, and Point72 run dozens of independent teams under one roof. Each one chases its own edge. None of them are allowed to quietly bend the others. The structure is the alpha. We rebuilt that machine out of distilled minds instead of trading floors, so one person can have what a multi-manager fund spends a fortune to maintain.

That one idea is the whole company. We didn't build a single genius AI to tell you what to buy. We built a hive. A swarm of distilled investor minds that forage the filings alone, each one coming back with an honest, sourced finding, then pull together into one signal: the watchlist. Roles stay fluid. Whichever lens the problem needs takes point. No dictator. No groupthink. Just the dance.

A hive beats a genius. You're not buying a tip. You're claiming a cell in the comb while it's still being built.

Fluid roles

Whichever lens the problem needs takes point: value, macro, quant. Status shifts with the question, never the org chart.

No dictator

No central voice forces the call. The signal emerges from many independent reports, the way a swarm picks its next home.

The waggle dance

Each mind returns with an honest, sourced finding. The watchlist is all of them integrated. Pluralism, not one opinion wearing a suit.

The mechanism

Introducing The Committee Method
Independent minds, one room, your call.

Most chat rooms, watchlists, and models all fall short for one reason. They are fixated on one strategy that works well at one point in time. When that point in time is over, it no longer works. Real fund managers have teams of analysts focused on sectors, constantly bringing in ideas to trade. That drives alpha. Not some silly prompt or a random group chat your buddy runs.

01

Scout

Like worker bees, the analysts fan out every day, scouting the market for the next idea.

02

Investigate

The most promising ideas get filtered through, then the analysts go deeper on each one.

03

Discuss

The committee talks it out and argues, working its way to the ground truth.

04

Publish

The highest-conviction names rise to the top, published as your Weekly Watchlist.

What lands in your watchlist

It doesn't give you a tip.
It shows its work.

Every name on the watchlist is read by the full committee, each lens working alone from the same filings. Here is one stock through all six minds, disagreement and all. (Illustrative of the research, not a recommendation.)

AI The Value Investor persona

Pass

Value lens · NVDA

"A genuinely exceptional business. But at this price there is no margin of safety, and it sits at the outer edge of my circle of competence."

AI The Quant persona

Accumulate

Quant lens · NVDA

"The numbers hold up. Gross margin above 70%, elite ROIC, clean cash conversion. Quality and momentum both screen green."

AI The Rationalist persona

Watch

Rational lens · NVDA

"Invert it. What kills this thesis? A CUDA crack, or custom-ASIC share at scale. I want those ruled out before I act."

AI The Macro Strategist persona

Watch

Macro lens · NVDA

"The thesis lives or dies on hyperscaler capex through 2027. I need that guide intact before I change my view."

AI The Contrarian persona

Pass

Contrarian lens · NVDA

"When everyone owns it and everyone loves it, the edge is already gone. I am not paying up for consensus."

AI The Growth Investor persona

Accumulate

Growth lens · NVDA

"A durable compounder. The reinvestment runway across AI infrastructure is still wide open, and the unit economics back the story."

The part nobody else sells

Most trading circles only give you buy lists. This committee is built to give you ground truth.

A typical Discord just feeds you content so you stay engaged. The committee is built to do the opposite. It acts only when signals actually emerge.

It passes, and says why

When a name sits outside a member's circle of competence, that member says so and stands down instead of inventing a confident take. "I don't know" is a valid, logged answer here.

Every yes ships with its exit

No call is just "buy." Each one carries the exact conditions that would break the thesis, and the price at which the committee would change its mind, written down before you ever act on it.

It tells you when it cracks

The committee keeps watching the indicators behind each thesis. The moment one trips, your Thesis-Change Alert fires. You're the first to know a view has flipped, not the last.

Illustrative of the committee's method, not a recommendation to buy or sell any security. Figures trace to the latest filing. Every kill-criterion is inspectable.

Meet the committee

Six minds. One room.
Each with its own lens.

These aren't costumes on one model. Each member is distilled from public letters, interviews, and books into a distinct framework. Each one asks the questions that mind would actually ask before it signs off on a name.

01

01

The Value Investor

A Buffett-style lens: moat, management quality, owner earnings, and a margin of safety before anything else.

02

02

The Rationalist

A Munger-style mental-models lens that inverts the thesis and hunts for the reasons you're fooling yourself.

03

03

The Macro Strategist

A Dalio-style read on cycle position, rates, inflation, and the cross-asset backdrop the company sits inside.

04

04

The Quant

A systematic lens that screens the numbers: growth durability, margins, ROIC, and cash conversion.

05

05

The Growth Investor

A lens for durable compounders: TAM, reinvestment runway, and the unit economics behind the story.

06

06

The Contrarian

An Icahn-style activist lens that fades the crowd, pressures management, and hunts for the value a consensus trade has left on the table.

The swarm

Independent findings,
one honest signal.

Each mind forages alone and returns with a sourced finding. The hive integrates them, and what survives the dance becomes the week's watchlist. Every forage is remembered too, wired into a comb of past coverage that compounds.

Node Analyst Research team Finding Company Theme Consensus
AAPL MSFT NVDA Value Macro Quant Growth Rationalist Contrarian Consensus

Spot the empty cell

When a name has only been foraged by one bee, the comb makes the gap obvious, and the committee sends out a second scout.

Pull a thread

Start from a theme like "AI infra spending" and every finding, company, and mind that ever touched it surfaces at once.

Trace any claim

Follow any cell straight down to the filing, table, or transcript sitting underneath the conclusion.

Tens, then hundreds of forages. All remembered, all connected. The difference between getting a tip and keeping a hive.

6

Independent minds

100%

Calls cite their evidence

Weekly

Fresh watchlist

24/7

The room never closes

Why not just…

The honest side-by-side.

Analyst Oracle is a research newsletter, a private trading community, and the AI engine that powers both. Here is how the whole stack compares to buying just one piece of it.

Analyst OracleStock Trading DiscordA Standalone NewsletterDIY
Two+ master investors, each ruling independently
A clear, written thesis on every pick ~ ~
Bull and bear, and where the analysts clash ~
Every number traced back to its source
Unsourced claims get cut, not published ~ ~
A live community to trade alongside
A fresh watchlist every Sunday
Never forgets. Ties this week to every past call
Tells you when to pass and when to sell ~ ~
Time you spend per week ~15 min Hours ~30 min A weekend

Research to inform your own decision, never a signal to copy. ✓ = yes · ~ = partial · ✗ = no.

Is this for you?

Built for people who
don't care for hype.

📊

The Self-Directed Investor

You run your own account and want a real research team in your corner. Not a hot take. Not a signal to blindly copy.

🔍

The Diligence-Heavy Researcher

You already read filings. You want a room of distinct lenses to pressure-test your thesis and surface the angle you missed.

🧠

The Curious Operator

You're fascinated by where AI and markets collide, and you want a front-row seat as a hive-mind research process gets built in public.

🤖

The Technologist

You want access to the best investing models available. You care as much about the engine as the picks it produces.

Be honest. Skip this if:

  • You want guaranteed buy/sell signals to copy without thinking
  • You want a bot that promises to beat the market for you
  • You don't care whether a call is backed by a real source

Under the hood

The desk that never closes.

Analyst Oracle is always running: tracking signals, sweeping sectors, pulling fresh filings the moment they land, and answering your questions when you have them. A real research desk, distilled minds grounded in primary sources, graded in the open. Each layer does one job.

L1

The Watch Floor

The desk never closes. Four streams flow in around the clock: live market signals it watches for regime changes, sector sweeps across every vertical it covers, fresh filings and prints the moment they publish, and your questions when you have them. The engine, not a human, decides what's worth a research pass.

Always onSignal monitoringSector sweepsYour questions too
L2

The Director

A signal trips, a sector moves, or a question lands, and the Director triages it, decides whether it earns a full research pass, and turns it into a typed plan: a dependency graph of the exact steps, in the right order. It runs on a clock, not a command. Nothing fires on vibes.

Autonomous triageTyped task graphRuns on a schedule
L3

The Research Analyst

A tireless first-year does the grind: pulls the filings, the prints, the transcripts, and verifies every load-bearing fact against a live source before it moves forward. No claim reaches the committee unsourced. Never gets tired, never skips the footnotes.

Live fact-checkingEvery claim sourcedBuilds the brief
L4

The Distillation Engine

Proprietary

Everything else in this stack is engineering. This is the part that's ours.

We reduce a lifetime of public reasoning, every letter, interview, filing, and recorded decision, into a structured representation of the investor's decision function: the priors they weight, the signals they trust, the constraints they will not violate. That representation is fused with a recency-weighted semantic retrieval layer over a speaker-attributed corpus of their own words, so every inference is anchored to a verifiable citation, never a model's hallucinated memory. A fidelity harness scores each output against the source before it ever leaves the engine.

Decision-function modelingRetrieval-grounded inferenceRecency-weighted corpusFidelity-scored
L5

The Committee

The distilled minds research the same company, each in isolation, so no single confident voice can quietly bend the room. Independent takes first, the clash surfaced second, a human decides last.

Persona agentsIndependent reasoningNo groupthink
L6

Evidence & the Second Brain

Every artifact is stored with its lineage. A knowledge graph stitches tickers, memos, analysts, and themes into a memory you can actually query: research that compounds instead of resetting every session.

Evidence chainFull provenanceKnowledge graph
L7

Primary Sources

The same raw material an institutional desk reads (filings, prints, calls, wires), pulled fresh and read directly, never paraphrased from a model's memory.

SEC EDGARMarket dataEarnings callsNews wire

Inside a single pick

Every name on the watchlist runs a typed research pipeline.

Every name takes the same path through the desk you just saw: flagged, planned, sourced, argued, remembered. Each step declares what it needs and what it produces, and an automated gate rejects any claim that isn't backed by a source before it ever reaches the watchlist. The path runs the same way every time. Open-ended judgment is reserved for the one step that truly needs it.

Runner Deterministic Hybrid Agent
Phase 1 · L1

The Watch Floor

Flags the name for a research pass.

watch_signal
sector_sweep
filing_listener
coverage_brief
Phase 2 · L2

The Director

Plans it into a typed task graph.

triage
scope_check
plan_graph
dependency_sort
Phase 3 · L3

The Research Analyst

Sources and verifies every fact.

fetch_filings
fetch_market
parse_transcripts
extract_facts
gate_check
Phase 4 · L5

The Committee

Argues it out, then the clash surfaces.

classified
surface_clash
score_thesis
assemble_memo
Phase 5 · L6

The Second Brain

Remembers it with full lineage.

link_entities
index_memo
graph_update

How it works for you

Joining takes about ninety seconds.

1

Take your seat

Claim a founding spot and land in the private Discord the moment you join.

2

Get the watchlist

Every Sunday, high-conviction names drop. Each with a well thought out thesis.

3

Discuss it live

Pressure-test the picks with the committee and members in the Discord.

The full stack

Everything you get the moment you take your seat.

A Bloomberg terminal runs about $32,000 a year. A junior analyst, about $80,000. A six-mind committee, every Sunday, costs you less than a dinner out.

The Weekly Watchlist

The core deliverable

The committee's highest-conviction names, every Sunday. Each with the full thesis and the filings that back it.

A Seat in the Private Discord

The live community

Live committee discussion, ask-the-analyst threads, and a members-only community of serious, no-hype investors.

The Coverage Archive

The compounding edge

Every past pick, thesis, and the evidence behind it. Searchable, connected by theme, and compounding week over week.

Thesis-Change Alerts

The early warning

The moment the committee flips a view on a name you care about, you're the first to know. Not the last.

The real cost

What it would cost to build this yourself.

A serious research desk is not a subscription. It is a stack of expensive inputs, an engineer to wire them together, and an analyst to run it, every week. Here is the bill, before anyone reads a single filing.

The data (recurring, every year)

  • Bloomberg Terminal

    One seat. The terminal every desk runs on.

    ~$31,000/yr
  • Premium subscriptions

    WSJ, NYT, FT, Bloomberg, The Economist.

    ~$2,000/yr
  • Fundamental + market-data APIs

    Filings, fundamentals, prices, estimates.

    ~$6,000/yr
  • Transcripts + expert calls

    AlphaSense and Tegus-grade access.

    ~$18,000/yr
  • Screening + charting

    A real screener and charting stack.

    ~$1,500/yr
The data alone, before you write one line of code ~$58,500/yr

The people + tech (to actually make it run)

  • A research analyst to run it

    Or ~500 hours of your own: 10 hours, every week, all year.

    ~$90,000/yr
  • Engineering

    Build the decision engine and keep it live.

    ~$40,000 (one-time)
  • AI inference + cloud hosting

    Running the committee on every name.

    ~$5,000/yr
  • Corpus retrieval / vector store

    The memory that makes it compound.

    ~$1,000/yr
Year one, all in ~$190,000+

~$150,000/yr after that. The data and the analyst never stop.

That's ~$3,650 a week.

You're skipping the $90k analyst, the $58k data bill, and your own 500 hours, for less than a lunch.

A real research desk costs six figures and a full-time hire. We rebuilt it as one subscription. Your seat starts at $25 a week.

Take your seat →

Take your seat

Join the Investment Committee.

Your seat includes

  • The Weekly Watchlist: the committee's highest-conviction picks
  • The full thesis + sourced evidence behind every name
  • A built-in exit on every call: the kill-criteria that would flip the thesis
  • A seat in the private member Discord
  • The searchable Coverage Archive (your second brain)
  • Thesis-change alerts the moment the room shifts a view
  • No-hype community of serious, self-directed investors
  • Your rate locked for as long as you're a member

Building this desk yourself runs

~$150,000/yr ~$3,650/week
See how we get there ↑

Choose your billing

Weekly

$25 /week

Billed weekly · cancel anytime

Start weekly

Monthly

$60 /month

Billed monthly · cancel anytime

Start monthly
Best value · 5 months free

Annual

$420 /year

Just $35/mo, billed annually

Get annual

Secure Stripe checkout · Cancel anytime

Not ready to commit?

See last week's watchlist,
on the house.

Get one full Weekly Watchlist, free. Every name, every thesis, every source link. See the quality before you pay a cent.

  • Real names with the committee's reasoning
  • Every call traced to its filing or transcript
  • No spam. Unsubscribe in one click.

Free sample · No card required · One-click unsubscribe

Illustrative research for educational purposes only. Not financial advice or a recommendation to buy or sell any security.

Before you join

Common questions .

What exactly do I get for my membership?

Two things. A seat in the private Discord, the room where the committee's work and the community live. And the Weekly Watchlist: the committee's highest-conviction picks, each with the full thesis and the filings behind it. Plus the searchable archive of everything it's ever covered.

Is this financial advice or a signal service?

No. Analyst Oracle is research and education: sourced, inspectable theses that argue both sides, to inform a decision you make yourself. We don't place trades, and we don't hand you signals to copy blindly. The whole point is that you understand the why, not just the what.

How often do I get the watchlist?

A fresh watchlist every Sunday. The committee's highest-conviction names, not a firehose of tickers. Each comes with the full thesis and the filings behind it, so you always know why a name made the cut.

Can I cancel anytime?

Yes. Cancel in two clicks from your account, any time. The founding rate you lock in stays with you for as long as you're a member.

The committee meets every week.
The only question is whether you're in the room.

A real research desk for less than a lunch. Your seat starts at $25 a week, and you can cancel anytime.

Take your seat →

Instant access · Cancel anytime