You already sense it.

The market is confusing participation with power.

Infinity Thesis identifies the few companies becoming foundational to the next economy — then prices their five-year value distribution across one million simulations.

LiveSee the report on any company on the record →
Thesis95
5-yr value$6.5T
SignalSTRONG BUY
Free live preview · See the market gauge →
Today's reading

NVIDIA, in full.

Seven thesis factors. Bear-to-tail price algorithm. Insider activity. Reasoning trace. The reading you subscribe to — in one card.

NV
NVIDIA
NVDA · $4.72T · Semiconductors
Rank
#1
of 100 on the record
Thesis
95
geometric mean of seven factors
Signal
Strong Buy
+13% / yr · priced below its own downside
5-Year Outlook
Tap a band

+6% / yr · Even a poor outcome tops today's $4.7T — the whole modeled range sits above where it trades now.

Why it moved

No material movement in the last 14 days.

Strongest factor

Vision (97). Jensen Huang's contrarian thesis — that general-purpose CPU computing has permanently run its course — articulated since 2006 and still being prosecuted today.

Weakest factor

No factor weakness flagged — every scored factor sits above the 70 concern threshold.

Thesis Quality — Seven Factors
Tap for reasoning

Jensen Huang's contrarian thesis — articulated continuously since 2006 and still actively prosecuted — that general-purpose CPU computing has permanently run its course.

Price Quality — The Price Algo

A real one-million-path Monte Carlo · click a band

today $4.7TBearBaseBullTail
Reality Check

Even a poor outcome ($6.3T) tops today's $4.7T — the market has it priced below its own downside. Upside to the tail ($24.1T) is 5.1×.

Insider Activity
Routine · Normal

Net buying or de minimis selling — within the typical insider-activity range.

Cluster sellingCEO is a top seller
Top sellers
HUANG JEN HSUNPresident and CEO$-1.0B377 txns
STEVENS MARK A$-601.9M11 txns
Puri Ajay KEVP$-208.7M14 txns
SEAWELL A BROOKE$-122.3M54 txns
Candidate thesis-break triggers
TRIM

NVDA EBITDA margin falls below 15% of current for two consecutive quarters.

TRIM

Forward revenue growth falls below the Constraint-Current band's lower bound.

HARD SELL

The Moat verdict drops from its current band to a lower one.

The market read

Two lenses on the same market.

Before we ever price a company, we read the market two ways — structurally and tactically. Both are free.

Structural · since 1900

Where we stand, against a century.

Deep value020Attractive2040Fair4060Elevated6080Irrational8090Extreme9010019001920194019601980200020201929DepressionDot-comGFC low2021
Exuberance IndexS&P 500 · real, log
The higher the reading, the weaker the decade that followed.
A high reading doesn't call a crash — it means the odds and risk-reward for new money have gotten worse. The y-axis above is this same scale; today sits in Elevated.
020Deep value+11%/yrGenerational opportunity — the market is deeply cheap.
2040Attractive+7.3%/yrBelow-average valuations; patient capital is rewarded.
4060Fair+6.5%/yrAround the long-run normal; stock selection matters most.
6080Elevatedtoday+2.4%/yrExpensive and demanding — where 1929, 2021, and today sit. Be selective with new money.
8090Irrational exuberanceno recordThe record extreme — only the dot-com peak (82 in 2000) ever reached here. Maximum discipline.
90100Extreme bubbleno recordBeyond the entire 125-year record — no market has ever priced here. Capital preservation.
Tactical · since 2021

Hot at the tops, cold at the bottoms.

Market Heat· 0–100S&P 500
Cold · oversold ≤38n = 5
+7.9%next 3 mo

Average next-3-month S&P return when the market is washed out. Fear snaps back.

Hot · overheated ≥62n = 7
-0.5%next 3 mo

When it runs hot, the near-term edge fades — well below the usual quarter.

Baseline — the S&P's average quarter since 2021 is about +2.5%.

Both gauges are free. The reports are the product.

The few companies becoming foundational to the next economy — priced across a million simulations.

See the reports →
The spectrum

Four verdicts. Never a vague “buy.”

No hand-wave. Every reading resolves to one of four verdicts, and each threshold is a real financial line— anchored to the market’s own cost of equity (the risk-free rate plus the equity risk premium). So you always know exactly what return you’re being paid to take the risk. And it moves with rates automatically.

Signal 01
Strong Buy
Pays above the market's own return — plus a margin of safety

The rare one. Over five years the math pays you well above what the market itself expects to return, with room to spare — priced below its own downside. Load-bearing conviction.

Signal 02
Accumulate
Pays the full equity risk premium

Priced to beat the market's own expected return. You're paid the full premium for taking the risk — deploy on strength, add to what you own.

Signal 03
Hold
Beats bonds, but below the market's return

The thesis is intact, but the price has caught up to it. Own it; wait for a better entry before you add new capital.

Signal 04
Wait
Below the risk-free rate

The upside is already in the price. At today's level you'd earn less than Treasuries for taking equity risk — so you wait.

Backtests beat S&P
8 / 8
Jul 2026Source →
Outperformance vs S&P
2 – 10×
Jul 2026Source →
Companies on the record
100
Jul 2026Source →
Vetoes directionally correct
6 / 7
Jul 2026Source →
On the record since
Dec 2016
Jul 2026Source →
The permanent record

Timestamped at the moment of the call.

“In December 2022 the consensus said tech was dead. The report flagged NVDA a Strong Buy at $14.61 — with every citation on the record. It returned +1,300%.The same discipline flagged PLTR as priced-in at the top and caught this June's exuberance two days before the selloff.”

∞ 14,287
The permanent recordEvery call timestamped — wins and misses
Dec 28, 2022
NVDA — Strong Buy at $14.61
→ +1,300% by mid-2026
Dec 12, 2024
PLTR — Wait at $75, thesis elite
→ sidestepped the −36% drawdown
Jun 5, 2026
Regime flagged Extreme Exuberance
→ two days before the AI-semi selloff

Backtested and live calls, Dec 2016 – Jul 2026. Past performance does not guarantee future results.

NVDA
Conviction case · December 2022
“Scored 86 and flagged a Strong Buy while consensus said tech was dead — at $14.61, split-adjusted. Every citation from that run is still on the record.”
Snapshot #4,112 · entered Dec 2022 · +1,300% by mid-2026
PLTR

Wait at $75 — an elite thesis, but the price had run past it.

Price discipline
MEI · 99

Flagged Extreme Exuberance two days before the June selloff.

Capital preserved
Primary sources

Evidence-first.

Every score is traced to filings, transcripts, and primary data. Never sentiment feeds. Changes beyond five points require new evidence.

SEC EDGAR10-K · 10-Q · 8-K
POLYGONprices · fundamentals
FREDrates · macro
TRANSCRIPTSearnings calls
FINRAmargin · leverage
SHILLER / YALECAPE, 1871→
DAMODARANrisk premium
RITTER / UFIPO quality
GRID / FERCpower constraint
Join our

Community

𝕏X
YouTube
Discord
GitHub
Substack
inLinkedIn
Don't miss our

Updates

Join our weekly synthesis. Regime read, top movers, and the month's most-cited calls.

Become a

Subscriber

Access the full record — every reading, every reasoning trace, every trigger. Read the market through the instrument.

FAQ

Ask.

What exactly is Infinity Thesis?
A research publication for the sovereign investor. Seven thesis factors score every company on the record; a deterministic price algorithm computes the Bear/Base/Bull/Tail distribution; every reading is timestamped, cited, and permanent.
Who is this for?
For investors who invest in decades, not quarters. Who think for themselves. Who suspect the institutions pricing the future are still running Industrial-Age software.
How is the price algo different from AI?
It's deterministic. A million-path Monte Carlo produces the Bear/Base/Bull/Tail 5-year market cap distribution. No AI calls in the price side. No hallucination. No bias. Pure math, rigorously calibrated.
What are the 7 thesis factors?
Vision · Execution · Problem Hardness · Moat · Reach · Constraint (Current) · Constraint (Forward). Each with its own rubric, each auditable to primary sources.
Do I ever run the instrument myself?
No. You read the reading. The instrument runs continuously — evaluating triggers, re-scoring on material events, updating the record.

Read the record.

Every company, timestamped. Every score, cited. Every position, falsifiable — before you buy.