β AI & Startups
Anthropic IPO plus commitments (Silicon Republic)
~$64bn + $26bn
SpaceX IPO, June 2026 (NPR)
$75bn
Saudi Aramco IPO incl. green shoe, 2019 to 2020
$29.4bn
Record listings have started strongly before and then struggled: Alibaba in 2014 and Saudi Aramco in 2019 both rose sharply on debut, then fell below their offer prices within a year, while SpaceX, the record holder since June 2026, ended its first week 37% above its IPO price. Anthropicβs offering, which with $26 billion of share commitments would come to about $90 billion, has not yet priced, so these precedents show a range of outcomes rather than a forecast.
⬑ Crypto
200-week moving average, 13 Nov 2026 (CoinDesk)
~$68,000
Bitcoin's fall over the week to 14 Nov (Bitcoin Magazine)
~13%
Record high, 6 Oct 2025 (Fortune)
$126,198
Bitcoin fell below its 200-week moving average, around $68,000, on 13 November and kept falling, to about $63,500 late on 14 November. The line has been tested before: it held on a weekly-close basis in December 2018, was broken only briefly in March 2020, and in 2022 bitcoin stayed below it until October 2023, after a further fall to $15,480.
π PokΓ©mon
PSA Value price from 29 Oct 2026
$39.99
PSA backlog, 28 Oct 2026
7.2 million
2021 Value suspension
Mar 2021 β Jun 2022
PSA reopened its Value grading tier on 29 October 2026 at $39.99 per card, just under five months after pausing it with about 10 million cards in its backlog. It has done this once before: in 2021 PSA suspended its cheaper services for the same reason, and Value only returned in June 2022, about 14 months later, after Express had come back $50 dearer.
β· Watches
WatchCharts index, Q3 2026
+1.2%
Index vs April 2022 peak
About β29%
Swiss watch exports, 2009
β22.3%
The watch industry has climbed out of two slumps since 2008: Swiss watch exports rebounded 22.1% in 2010 after falling 22.3% in 2009, but rose only 2.7% in 2017 after two years of decline. Pre-owned prices have now risen for five straight quarters yet remain about 29% below their 2022 peak, and record inventory is what sets this recovery apart.
β AI & Startups
S&P 500 fall, 5 Nov 2026
β4.4%
Nvidia one-day value loss, 27 Jan 2025
$589bn
Nasdaq peak to trough, 2000 to 2002
β78.4%
Yes, tech has been through scares like this before: Nvidia's record one-day loss in the DeepSeek sell-off of January 2025 was regained within five months, while the telecom build-out bust of 2000 to 2002 took the Nasdaq down 78.4% and fifteen years to recover. The sell-off of 5 November 2026, driven by worries that AI data-centre construction may slow, has so far looked more like a sharp scare than either of those, but the precedents show both endings are possible.
⬑ Crypto
2026 airdrop tokens below first-week median (Bitquery)
35 of 46
Uniswap airdrop per address, Sept 2020
400 UNI
2024 airdrop tokens that declined (Keyrock)
88%
Past airdrops have more often been followed by falling prices than rising ones: Uniswap's token rose in its first week in 2020, but Arbitrum and Wormhole traded well below their early highs within days, and a Keyrock study found 88% of tokens launched alongside airdrops in 2024 declined. Vendor research on this year's airdrops points the same way, with 35 of 46 tokens studied trading below their first-week median price.
π PokΓ©mon
Delta Reign English release
6 Nov 2026
Roaring Skies English release
6 May 2015
Evolving Skies cards incl. secret rares
237
Yes: Delta Reign, released on 6 November 2026, is not the first English PokΓ©mon set to put Rayquaza up front, after Roaring Skies featured Mega Rayquaza in 2015 and a Rayquaza VMAX became one of the most valuable cards in Evolving Skies in 2021. The earlier sets show that a Rayquaza headline is no guarantee the set's top card will be a Rayquaza, and they arrived in very different markets.
π Historical Precedents
Spot gold record, 29 Jan 2026
$5,594.82
Spot gold, 6 Nov 2026
$4,070.59
Gap between 1980 and next record
Jan 1980 β Jan 2008
Gold has been here before: after its peaks in January 1980 and September 2011, it did not set a new record until January 2008 and July 2020. Spot gold is now near $4,070, well below January's record of about $5,595, and the speed and causes of this year's swing differ from both precedents in ways worth understanding.
β AI & Startups
OpenAI valuation reported 6 Nov
$1.25tn
Target reported 29 Sep
$30bn at $1.4tn
WeWork valuation, Jan to Oct 2019
$47bn β ~$8bn
Yes: Facebook in 2012, Uber in 2019 and WeWork in 2019 all priced below what had been expected, and what followed ranged from a slow recovery to bankruptcy. OpenAI's reported $11 billion raise at a $1.25 trillion valuation, below the $1.4 trillion target reported in September, is a private round rather than a listing, which changes how far those precedents carry.
⬑ Crypto
CLARITY Act Senate procedural vote
49β50
Bitcoin first above $100,000
5 Dec 2024
FTX bankruptcy filing
11 Nov 2022
US elections have mattered to crypto mainly through what Congress and regulators do next, and the last two showed how differently that can play out: a policy-driven rally after 2024, and an industry collapse unrelated to the vote in the same week as the 2022 midterms. The 2026 result, with Democrats taking both the House and the Senate, puts the stalled CLARITY Act at the centre of that question.
π PokΓ©mon
Released worldwide
16 Sep 2026
30th Celebration master set
199 cards
Celebrations (2021) set
25 + 25 classics
Yes: the 30th Celebration set is the third time in a decade that PokΓ©mon has marked an anniversary by bringing back its most famous early designs, after Evolutions in 2016 and Celebrations in 2021. The precedents show how differently each set was built and sold, and the 2021 one arrived in a year so heated that a major US retailer temporarily stopped selling trading cards in its stores.
⬑ CryptoHPI
Top HPI match
Bitcoin 2013
Maturity window
180β540 days
Signal Intelligence has detected a strong pattern match between Ethereum's current signal profile and Bitcoin's 2013 accumulation phase β 11.3 years before Bitcoin achieved mainstream recognition.
π PokΓ©monHPI
Inventory compression across three major platforms, combined with PSA population data, suggests a supply contraction consistent with early accumulation phases seen before major price movements.
β AI & StartupsHPI
Signal Intelligence has run HPI analysis across 110 tracked AI entities. Twelve are showing signal profiles consistent with OpenAI's pre-breakthrough accumulation phase from 2017β2019.
β· Watches
Signal category
Allocation + scarcity
Signal type
Allocation shift
Typical lead
Months, not weeks
Across multiple historical watch market surges, one signal consistently appeared first: a quiet shift in waitlist length, months before retail prices moved at all.
β¦ GenesisHPI
Core question
"Seen this before?"
Six validated historical cases. One question they all answer: have we seen this pattern before? That question is the entire premise behind Signal Genesis.
⬑ Crypto
Domains live
Crypto & DeFi
Update frequency
Every 15 min
By the time a token trends on social media or a headline mentions a protocol by name, the signal that actually mattered has usually been building for weeks.
π PokΓ©mon
Detector type
Population Scarcity
Data sources
PSA, CGC, BGS
PSA, CGC, and BGS population reports are public data. The information has never been hidden. What's been missing is someone watching it continuously enough to notice when it shifts.
β AI & Startups
Over a 90-day tracking window, four entities that surfaced on GitHub Trending subsequently appeared in major tech press β with an average lead time of 14 days between the technical signal and the headline.
β· Watches
Signal category
Demand vs allocation
Signal type
Waitlist length
Visibility
Low β by design
Waitlist length is one of the quietest signals in any collector market β and historically, one of the most reliable. It moves before the price list does, and almost nobody is checking it continuously.
β¦ GenesisHPI
Rank in library
Highest score
NVIDIA's pre-recognition signal profile preceded mainstream attention by 7.2 years, with a peak validated opportunity score of 0.91 β the highest of all six precedents in the HPI library.
⬑ Crypto
Signal type
Governance frequency
Governance proposal frequency is one of the least glamorous signals Signal Intelligence tracks β and one of the most consistently informative, precisely because almost nobody else is watching it closely.
π PokΓ©monHPI
Signal class
Coordinated acquisition
Typical pattern
Quiet, then sudden
Buyout Detection exists to catch coordinated acquisition activity β market sweeps and listing removals that happen before any single high-profile sale makes the activity publicly obvious.
β AI & Startups
Collectors
GitHub, HN, Product Hunt
There is no point in any company's history where it was obviously going to become what it eventually became β except in hindsight. The signal profile that precedes recognition looks, at the time, like noise.
β· Watches
Signal category
Sentiment + scarcity
Signal type
Sentiment shift
Community sentiment in collector forums tends to shift in tone well before any official price list reflects the change β a lead-time pattern the Watches domain is being built specifically to capture.
β¦ GenesisHPI
Method
Pattern shape, not outcome
Precedents
6 validated cases
Mark Twain's observation that history rhymes rather than repeats is, in a precise and measurable sense, exactly what Historical Precedent Intelligence is built to detect.
⬑ CryptoHPI
Corroboration level
3 sources
Quality gate
Strongest available
Cross-source corroboration is the single strongest quality gate in the Signal Intelligence scoring model. When three independent collectors confirm the same pattern simultaneously, the probability of noise drops sharply.
π PokΓ©monHPI
Validated cases
2 (1st Ed. Charizard, Lugia)
Pattern type
Print-run scarcity
Detector
Population + Inventory
1st Edition Charizard and Lugia 1st Edition share a documented signal signature in the historical replay data β a pattern that, when it reappears, is now actively tracked across the rest of the Collectibles domain.
β AI & Startups
Signal type
Post-launch acceleration
Reliability factor
Sustained, not single-day
Launch day on Product Hunt is inherently noisy and partly driven by promotional effort. What's harder to manufacture artificially is sustained engagement acceleration in the weeks that follow.
β· Watches
Signal category
Upstream allocation
Signal type
Allocation shift
Allocation decisions β which dealers receive which references, in what quantities β are made internally well before any retail-facing change becomes visible. The decision precedes the visibility by a wide margin.
β¦ GenesisHPI
SpaceX lead time
9.4 years
OpenAI lead time
6.8 years
Three of the six validated historical cases in the HPI library, side by side: Tesla's 9.7 year lead time, SpaceX's 9.4 years, and OpenAI's 6.8 years β each preceded by a signal-only phase that looked, at the time, unremarkable.
⬑ Crypto
Inputs
Frequency, recency, diversity
Behaviour
Compounds over time
Collective Intelligence Asset scoring is the layer that turns repeated observation into a compounding measure β tracking frequency, recency, and source diversity over time rather than scoring any single signal in isolation.
π PokΓ©mon
Signal philosophy
Lead, not lag
Detector count
5 active types
Data type
Non-price-derived
Price is the lagging output of a pattern, not the pattern itself. By the time price moves enough to attract broad attention, the inventory, population, and auction signals that drove it have typically already played out.
β AI & Startups
Signal type
Sustained discussion
Reliability marker
Engagement depth
Front-page placement on Hacker News correlates with a specific, identifiable pattern of sustained technical discussion β distinct from viral but shallow engagement that fades within hours.
β· Watches
Signal category
Allocation + scarcity
Core driver
Structural scarcity
Hype is loud and visible by nature. Genuine scarcity β the structural kind that actually drives sustained secondary-market appreciation β tends to be quiet and only becomes obvious well after the fact.
β¦ GenesisHPI
Function
Historical validation
Cases available
6, fully validated
Output
Genesis Score 0-100
A signal tells you something unusual is happening. It doesn't, on its own, tell you whether to take that signal seriously. Genesis exists to answer the second question with documented historical evidence rather than intuition.
⬑ Crypto
Collector cadence
Continuous
Update frequency
Every 15 min
The volume of announcements across the crypto ecosystem makes it genuinely difficult for any single human to track which ones matter. Continuous structured collection doesn't have that limitation.
π PokΓ©monHPI
Detector firings
4 in past 14 days
Pattern type
Sustained build
Pikachu Illustrator's status as one of the most significant cards in the hobby wasn't established in a single moment. The signal data shows a sustained build, documented through historical replay using only pre-recognition information.
β AI & StartupsHPI
Confirmed matches
12 vs OpenAI 2017
Most of the activity that eventually defines a major platform happens in full public view β commits, launches, discussions. What's missing is usually not visibility. It's continuous, structured scoring of that visible activity.
β· Watches
Signal category
Discontinuation tracking
Signal type
Discontinuation event
Historical pattern
Documented precursor
Discontinuation announcements are a well-documented historical precursor to secondary-market appreciation in collector watch categories β and they're announced publicly, well ahead of the resulting price movement.
⬑ Crypto
Crypto update cycle
Every 15 min
AI/GitHub cycle
Every 30-60 min
The structural advantage of continuous collection isn't speed for its own sake. It's that some patterns are only visible across multiple closely-spaced observations β and a once-daily check structurally cannot capture that.
β¦ GenesisHPI
Press coverage at the time
Effectively none
Bitcoin's pre-recognition signal profile preceded mainstream attention by 11.3 years β the longest lead time of all six validated cases in the HPI library, built during a period with essentially zero mainstream press coverage.
π Historical Precedents
Asset type
Accumulating database
Competitive moat
Time-based
Dashboards, charts, and alerts are the visible surface of Signal Intelligence. The actual commercial asset is underneath all of it β the accumulating database that makes every one of those surfaces possible.
π‘ Insurance Intelligence
Lead indicator
Regulatory filings
Signal class
Regulatory + claims pattern
Founder background
Infrastructure patent holder
Insurance Intelligence is one of three domains currently in development, built around a category most signal platforms ignore entirely: regulatory change as a leading indicator, not a lagging compliance cost.
π Property
Lead indicator
Planning applications
Signal class
Planning + infrastructure
Planning applications, infrastructure announcements, and demographic shifts are all public record β and all tend to precede meaningful property price movement by a window most buyers and investors don't track continuously.
β¦ GenesisHPI
Rank in library
Shortest of six
Shopify's pre-recognition signal profile carries the shortest lead time of all six validated cases at 8.1 years β still nearly a decade, and a useful reminder that even the platform's 'quickest' precedent moved on a multi-year timescale.
β¦ Genesis
Product framing
Comparative, not predictive
Compliance layer
On every output surface
Historical cases
6, fully disclosed
It would be easy to market historical precedent matching as a forecasting tool. Signal Intelligence deliberately doesn't, because the distinction between comparison and prediction is the entire basis of the product's credibility.
π Historical Precedents
Standard
Evidence before claim
Risk addressed
Reputational + commercial
Domains
6, structurally independent
It's tempting to claim that a signal in one domain predicts behaviour in another. Signal Intelligence treats that kind of claim as something that requires genuine evidence before it's presented to anyone β sophisticated buyers especially.
π Historical Precedents
Newsletter access
Independent of platform tier
Platform access
Independent of newsletter
Detection method
Variant ID only
Trading Card Lens Weekly is built and billed entirely independently from Signal Intelligence platform access. This separation is a deliberate architectural decision, not an oversight.
π Historical Precedents
Domain-specific logic in core
Zero
MASTER, the platform's core infrastructure, contains zero domain-specific logic. Every signal, regardless of source, flows through identical normalisation, scoring, and historical comparison β by design.
β¦ GenesisHPI
Confirmed matches today
12 entities
OpenAI's 2017β2019 signal profile produced a peak opportunity score of 0.89 and a 6.8 year lead time β the shortest lead time among the six cases, and the most frequently matched precedent in the AI & Startups domain today.
π PokΓ©monHPI
Replay rule
Zero hindsight permitted
Top HPI match
Shopify 2014
Umbreon VMAX Alt Art β widely known by its community nickname 'Moonbreon' β is one of the entities validated through the platform's strict historical replay engine, which permits zero hindsight data.
β¦ GenesisHPI
Current top match
49.8% (Ethereum)
SpaceX's pre-recognition signal profile from 2008 currently produces the strongest single HPI match across the entire platform β 49.8% similarity to Ethereum's live signal data today, the highest of any active comparison.
β¦ GenesisHPI
Signal type
Infrastructure-building
Tesla's validated case carries a 9.7 year lead time and a 0.88 peak opportunity score β built during a period when the company's long-term viability was, by most public accounts, genuinely uncertain.
π Historical Precedents
Cases in library
6, all eventual successes
Failed-pattern cases
0 currently
Disclosure policy
Proactive
All six validated cases in the HPI library are success stories. That's not an oversight β it's a structural limitation worth naming clearly, especially to institutional buyers evaluating the platform seriously.
π Historical Precedents
Core constraint
Zero hindsight data
Methodology
Repeatable, auditable
Historical replay isn't just a phrase used to describe how the six validated cases were built. It's a specific, repeatable methodology with a hard constraint: only information available before a chosen date is permitted into the analysis.
π Historical Precedents
Governing principle
No new modules before onboarding proven
Modules already shipped
Multiple
Discipline applied
Every sprint
Building infrastructure is genuinely enjoyable work. Shipping monetisable product before the previous layer is proven is a discipline failure β and this platform has been built around actively guarding against that exact failure mode.
β¦ GenesisHPI
Archetype examples
Infrastructure, platform, protocol
Cases sharing infra shape
SpaceX, Tesla
Cases sharing platform shape
Shopify, OpenAI
Not every validated historical case shares the same signal shape. Distinguishing the underlying pattern archetype β not just the similarity score β is what makes a comparison genuinely informative rather than a single flattened number.
π Historical Precedents
Pricing range
Β£9βΒ£79, one-time
Sequencing
After first paying customers
Top candidate
Historical Similarity Checker
A future roadmap of smaller, standalone one-time-purchase tools is planned β anchored to the platform's proprietary HPI and entity database, not built as generic wrappers around any off-the-shelf model.
π Historical Precedents
Public output
Directional only (rising/stable/cooling)
Gated output
Full playback + precedent name
Design intent
Useful, not exhaustive
Any future public-facing tool built on this platform's data will follow one deliberate design rule: reveal enough directional information to be genuinely useful, while keeping full historical playback and precedent detail behind paid tiers.
π Historical Precedents
Currently viable domain
PokΓ©mon only
Requirement
Real producer module
A planned licensing model would let third-party niche sites embed live CIA and HPI scoring badges directly β currently viable only in domains where a real producer module already exists.
β¦ GenesisHPI
Timeline structure
T-48 to T-0
Pattern archetypes
8 named types
Comparison basis
CIA stage matching
A similarity score tells you how closely two signal profiles resemble each other. It doesn't tell you where you currently sit inside that resemblance β early, late, or somewhere in the middle of a multi-year pattern. That's the question Genesis is built to answer.
π Historical Precedents
Core engine logic
Identical across domains
Domain-specific code
Lives only in modules
Live domains
3 of 6 planned
The core thesis behind this entire platform is a single sentence: the engine is domain-agnostic, only the lenses change. Unpacking what that actually means in practice reveals why it matters more than it might first sound.
π Historical Precedents
Subscription examples
Platform tiers, newsletter
One-time examples
Future satellite tools
Test applied
Outcome vs ongoing value
Every product decision on this platform follows one simple test: if the value is a single outcome, it's priced once. If the value is ongoing fresh data, it's a subscription. Mixing the two creates confused products and confused pricing.
π Historical Precedents
Current channels
Content, community, newsletter
Paid channels
Not currently in use
Target communities
X, Reddit, Discord, TCG forums
Content, community, and a Β£49/month newsletter are the current distribution strategy. Paid acquisition channels are deliberately not part of the current plan β a sequencing decision, not a permanent one.
π Historical Precedents
Scaling effect
Compounding, not linear
115 entities in Crypto & DeFi today. 110 in AI & Startups. As these numbers grow, the CIA layer doesn't just track more entities β it gets structurally more valuable, because comparison and corroboration both improve with scale.
π Historical Precedents
API access tier
Enterprise only
Enterprise price
Β£5,000/month
Rationale
Infrastructure-level access
Programmatic access to signals, entities, CIA scores, and HPI analyses is available exclusively on the Enterprise tier at Β£5,000/month β a deliberate gate, not an arbitrary upsell.
π Historical Precedents
Starter cadence
Weekly digest
Professional cadence
Priority real-time signals
Underlying data
Identical across both
The gap between Starter's weekly digest and Professional's priority real-time signals isn't an arbitrary feature split β it reflects a genuine difference in how each tier's likely user actually wants to consume intelligence.
π Historical Precedents
Constraint type
Deliberate, not technical
The domain-agnostic engine could, in principle, support dozens of intelligence lenses. The current roadmap deliberately limits scope to six β a constraint chosen for quality, not a limitation of the architecture.
π PokΓ©mon
Components
Detector mix + corroboration
Reading priority
Decomposition over raw score
An opportunity score of 0.82 is a single number. The decomposition behind it β which detectors fired, how many sources corroborated, how recent the activity is β is where the actually useful information lives.
⬑ Crypto
CIA dimensions
Frequency, recency, diversity
Risk addressed
Stale-signal false positives
Weighting
Recency decays older signals
An entity with frequent historical signal activity that has gone quiet recently scores differently from one with the same total signal count but sustained recent activity. Recency weighting exists specifically to capture that difference.
π Historical Precedents
Claim type
Structural, not promotional
Mechanism
CIA layer accumulation
What it doesn't claim
Guaranteed returns
The phrase appears directly on the platform's own pricing section. Unpacking exactly what it claims β and what it deliberately doesn't claim β is worth doing explicitly rather than taking the marketing language at face value.