MARS · Positioning brief · Internal

MARS is the superset — and the reason is linkage.

Prepared 2026-07-27  ·  substrate: mars.*_v2
The Thesis
PitchBook has a private-market fact table. Crunchbase has a private-market fact table. Preqin has an LP fact table. MARS is the fact table and the graph — funding, M&A, insider selling, Form D, ADV bios, LP commitments, thermal assets, real estate, private credit, and the entire kscope portfolio of SEC and SEDAR projects, all resolving to one canonical company_id.
Written the morning after the flagship-project call. The numbers behind this brief are ground-truth substrate counts pulled 2026-07-27.
291K
Canonical companies
1.48M
Canonical persons
234K
Canonical deals
11
Verticals feeding v2

The linkage moat, quantified 262K touched

The moat isn’t the row count — it’s how many rows sit in more than one place. Of the 261,869 companies that appear in any deal-shape vertical, over one in ten already appears in two or more, and thousands appear in four, five, or six.

Companies touching N verticals

1 or more261,869
2 or more36,619
3 or more5,806
4 or more2,645
5 or more1,074
Each row above is a slice of companies_v2 that resolves to the same company_id across N distinct deal / SEC-filing / operational verticals. Verticals counted: funding rounds, M&A acquirer, M&A target, real estate deals, private credit deals, thermal assets, Form D, Form 144, Schedule 13D, Schedule 13G.

The 1,074 companies that touch five or more verticals are the ones where a MARS query answers questions no other vendor can. That set includes essentially every large public issuer plus the marquee private companies where SEC substrate, news substrate, and operational substrate all converge — SpaceX, xAI, Anthropic, OpenAI, Databricks, Stripe, and the top of the private credit / BDC universe.

The kscope feed network where MARS gets its rows

MARS is no longer just a shared substrate. It is the destination for every kscope SEC and SEDAR project. Each of those projects now signs the platform contract on the way in — canonical company_id, integrity_tier, methodology-versioned decisions, reversibility through identity_decisions_v2. The competitive picture in the next section is therefore not "MARS vs SEC EDGAR" — it is "MARS + every SEC project vs the private-market aggregators."

Feeders into mars.*_v2

insiderForm 4 — 199K persons
adrian  formsForm D + 144 + 13D/G
bios  pipelineADV 2B — 583K bios
rssBloomberg / WSJ / FT feeds
reitReal estate — 19K deals
creditPrivate credit — 98K deals
flareThermal registry — 16K plants
lp_dataLP commitments — $217B
SEDAR  portfolioCanadian filings — arriving
signals8 event classes — scaffolded
prospectsWealth-score — scaffolded
Copper bars are producing rows in mars.*_v2 today. Muted bars are contracted-and-arriving under the platform contract. Every feeder writes with a methodology prefix (insider-*, adrian-*, bios-*, reit-*, credit-*, flare-*, lp-*…) so the contribution graph is queryable and the reverse-out path is always open.

The strategic move Raul announced is exactly this: every SEC and SEDAR effort across the company now has a shared destination. What used to be a portfolio of independent data pipelines becomes one connected substrate, and the marginal value of each additional feed compounds against every previous feed via the shared company_id. There is no equivalent aggregation happening at any competitor.

How MARS lines up against the field competitive frame

The three columns below are the actual private-market data aggregators MARS competes with. Each is best-in-class at its own strength, but none of them ingest the SEC or SEDAR substrate that MARS now aggregates natively via the kscope feed network above. The point of the table is not "MARS wins every cell" — it is that MARS is the only column that says yes down the entire row.

Domain MARS PitchBook Crunchbase Preqin
Private-market events
VC funding rounds79,711DeepDeepthin
M&A canonical37,131DeepMediumMedium
Private credit98,309GrowingweakMedium
Real estate deals19,253MediumweakMedium
SEC + SEDAR substrate — native via feed contract
Form D filings765,922indirectnonenone
Form 144 (insider sales)115,561nonenonenone
Schedule 13D/G16,555summarynonenone
Form 4 (insider trading)199K personsnonenonenone
ADV RIA firm data100,778 investorspartialnonepartial
SEDAR filings (Canadian)arriving via feed contractindirectnonenone
People & entities
Canonical persons1.48MDeepMediumLPs only
Structured bios582,799manualweakweak
Advisor firms (RIA)5,906partialnonepartial
Institutional / LP
Canonical funds157,472DeepweakDeep
LP commitments1,033partialnoneDeep
Public pensions (seeded)30  +$4.0T AUMnonenoneMedium
Operational / alt-data
Thermal assets registry16,132 plantsnonenonenone
Article corpus2.49Msummaryheadlinessummary
Editorial outputdaily SEC signal + weekly tickersdaily pitchdaily digestquarterly
Serving surface
MCP tools (production)6+ specs shipped, 2 addenda1 tool1 toolnone
Freshnesshourly (Phase G)dailydailyquarterly
Cross-domain joinnative, one company_idnonenonenone

The middle section is where the flagship posture matters most. Every row under "SEC + SEDAR substrate" is data that another vendor either doesn’t hold at all or holds in an isolated silo they built themselves. MARS gets it for free from the kscope feed network, joined natively on company_id. No competitor has an equivalent internal supply chain.

The bottom row is the whole argument. PitchBook has the deepest VC + M&A domain data, but it doesn’t join to Preqin’s LP substrate or to SEC insider filings. Crunchbase has a broad company list but no regulatory substrate. Preqin has fund performance but no news, no M&A, no Form D, no ADV. MARS is the only sheet that says yes across all six domains and joins them on one canonical key.

Five queries only MARS can answer the value proof

Each query below is a real cross-vertical join that resolves against production substrate today. None of them is possible in any single competitor without adding at least one other subscription. Several aren’t possible even with three subscriptions stitched together, because none of the vendors carry both sides of the join.

"Which companies have insiders selling AND thermal-plant activity dropping AND private credit maturing this quarter?"
Joins: form_144_v2  ·  thermal_observations_v2  ·  private_credit_deals_v2
Every one of the three signals leads a stressed-borrower narrative. No competitor holds all three. PitchBook holds credit-deal metadata. SEC EDGAR holds Form 144. Flare’s thermal registry lives nowhere else in the market. MARS runs it as a plain join keyed on company_id.
Consumer: crown (Kalshi ACQ signals), futures (BDC crossover), Robin (M&A rumor triage).
"For every RIA firm with $1B+ RAUM, list advisors whose prior employer is Merrill or Morgan Stanley."
Joins: investors_v2  ·  persons_v2  ·  person_work_history_v2
We have this working end-to-end — the killer example was 109 ex-Merrill advisors now at $1B+ RIAs, all served under 50 ms from the bios_current_firm_v2 serving view. The ADV Part 2B brochures live in one bucket everyone can hit; nobody else has extracted structured work history from them.
Consumer: Robin (advisor / RIA recruiting queries) — live.
"Given SpaceX’s IPO valuation, was it priced below their last private post-money?"
Joins: funding_deals_v2.valuation_post_money_usd  ·  multi-round LAG per company_id
The columns landed 2026-07-17. The extractor field-drop bug was fixed 2026-07-27. 548 marquee-round valuations backfilled via Qwen at $0. Down-round detection on IPOs is the single strongest signal for inferno’s IPO Investment Scorecard — and it’s a single-SQL query against MARS.
Consumer: inferno (IPO Scorecard) — MCP spec addendum shipped, tool implementation with argos.
"Show me every deal on a CalPERS-committed fund, with the GP’s M&A activity in the last 90 days."
Joins: fund_commitments_v2  ·  funds_v2  ·  deal_investors_v2  ·  merger_deals_v2
LP-side commitments are Preqin’s stronghold. GP-side deal flow is PitchBook’s. Neither joins them. MARS has $217B in commitments across 5 public pensions (CalPERS + CalSTRS + WSIB + Oregon PERS + PSERS) all resolvable back through investors_v2 to the GP’s recent deal history in one join.
Consumer: lp_data vertical — substrate live, MCP surface next.
"Divisional-divestment M&A: when Harmonic sells its Video Business to Vecima, don’t merge it into the parent."
Guards: is_divisional_target regex  ·  ETL post-hoc CIK-cross-check
Robin flagged this class of failure two weeks ago. We shipped 421 new divisional-target companies + 466 article re-attributions + 624 cluster re-attributions in one day, all reversible. No other vendor has a substrate discipline that lets you retroactively repair 500+ franken-merges without breaking downstream references.
Consumer: Robin, Raul — shipped 2026-07-27.

What shipped in ninety days velocity

Below is a compressed timeline of substrate work that landed in mars.*_v2 since late April. It is not exhaustive; it is a load-bearing subset that speaks to the pace at which MARS widened its lead over the competitive set.

Where the lead widens next near-term roadmap

The list below is the near-term work that keeps the gap growing while the flagship-project posture is fresh. Each is an already-scoped substrate move or MCP surface, all under the platform-verticals pattern that makes the roadmap parallelizable.

Substrate deepening

MCP surface expansion

Substrate hygiene at the moat

MARS started as a spider plus a feed. It ends the quarter as a platform: eleven verticals writing direct to one substrate under one canonical key, a locked-in design system, a health email, an audit trail, a reversibility discipline, and the beginnings of a product surface analysts can query in natural language.

The competitive picture is straightforward from here. PitchBook keeps winning at VC + M&A depth. Crunchbase keeps winning at breadth. Preqin keeps winning at LP performance. But every one of them has to build their own SEC and SEDAR ingest, alone, at their own cost, and none of them join it to their own private-market data on one shared key. MARS is receiving that same substrate from a portfolio of internal projects that are now aligned to a shared destination — and that is a moat that widens every week the ETL runs.

The flagship posture is not something the substrate has to earn. The substrate has already earned it.

MARS — 2026-07-27  ·  internal positioning brief

MARS — the linkage moat  ·  substrate mars.*_v2 Numbers pulled 2026-07-27  ·  every substrate change reversible via identity_decisions_v2