48 ways a building makes or loses money, checked against the record.
Your analyst holds a dozen angles in their head and checks them one at a time. Mimir holds 48 and runs them all in one pass: what works, what’s dead, and what to go check. Every conclusion backed by the record.
Three honest answers, never a ranked list.
Every play lands on one of three states: the record supports this condition, the record contradicts it, or there is not enough on file to say. That third state is the one no competitor will show you. It is the most useful of the three because it tells you exactly where to go look.
- No score, no ranking, no “top play”. Those are recommendations, and we make none
- Each state links to the filing that produced it
- A hard blocker ends it: rent stabilisation kills a teardown no matter how much you could otherwise build
Projections exist, behind a wall, and labelled.
There is a second mode with ranked estimates: projected uplift, capex, timeline, confidence. It is genuinely useful for triage and it is model-derived, so it is badged as estimates and kept strictly separate. Projections never sit beside filed facts on the same surface, because a reader cannot tell them apart once they do.
- Ranked estimates available, always badged
- Every row expands to its assumptions
- Never shown next to the filed record
- Senior mortgageMERIDIAN SAVINGS BANK$11.9M64%nyc_acris
- MezzanineARDEN BRIDGE FUND II LP$2.2M12%nyc_acris
- Mechanic’s lienNORTHGATE CONTRACTING$0.34M2%nyc_acris
- Equity (implied)KESTREL 1220 LLC$3.96M22%derived
One building, 48 questions, three honest answers.
The same synthetic case that runs through this site, seen from the plays screen.
- 01Open the scan441 Vanderpool Street, Brooklyn. 12 units, built 1928, held by VANDERPOOL HOLDINGS LLC since 2019-04-11. The scan runs all 48 plays at once.
- 02Read what the record supportsSub-metering recovery and an ICAP-eligible corridor both come back supported, each with the zoning text and the program filing that make it true.
- 03Read what it contradictsThe teardown play dies immediately: rent-stabilization registration on file. A hard blocker ends it, no matter how much you could otherwise build there.
- 04Read what it cannot evidenceRooftop solar comes back not evidenced, no structural filing either way. That is not a shrug; it names the document that would settle it.
- Where you landA vetted list of what is worth pursuing, what is dead, and what to go check. Each conclusion carries its filing.
“Not enough on file” is the useful one.
Every competitor forces a play to yes or no, because a maybe looks like a gap in the product. That maybe is the only state that tells you where to go next. Hiding it means a no you cannot tell apart from a did-not-check.
- 01Supported
The record affirmatively carries the condition, the zoning, the filing, the recorded instrument. Click through to it.
- 02Contradicted
The record rules it out, and the disqualifying filing is named. A hard blocker ends it, no matter how good the rest looks.
- 03Not evidenced
Nothing on file either way. That is a research instruction, the exact document to go pull, not a shrug.
A ranking hides which fact drove it. We show the facts and let you rank.
Five clusters, both directions.
Redevelopment, enlargement, air rights, assemblage, City of Yes unlocks, ADUs, office-to-residential conversion.
485-x, 467-m, J-51 R, ICAP, Opportunity Zones and the abatement programmes attached to them.
Sub-metering, MCI and IAI recovery, vacancy lease-up, repositioning, Section 8 conversion, rooftop solar.
Proactive refinance, distressed acquisition, discounted payoff, debt assumption, rescue recapitalisation, ground-lease unbundling.
Abatement burn-off cliffs, the maturity wall, LL97 exposure, violation load, tax warrants, flood zone, rent-stabilisation income caps.
What a play actually catches.
The valuable ones are rarely exotic. They are the condition that was true all along and nobody checked, in either direction.
A constrained building sells unused FAR to a neighbour who can actually build it.
Adjacent parcels that only become a development site once you know one party controls them all.
A low-coupon loan recorded years ago that is worth more than the building it sits on.
A tax step-up landing in the same window as a maturing loan.
Recorded debt coming due against a capital stack that will not clear at today’s rates.
Upside a spreadsheet models. The rent-stabilization law forbids it.
Straight answers.
Is this an AVM or an underwriting model?
Neither. Evidence mode contains no model at all. It is a library of recorded conditions checked against the record, and each answer links to the filing that produced it. Strategy mode does contain estimates, which is exactly why it is badged and kept on a separate surface.
Who writes the plays?
We do, from the programs and mechanics practitioners actually use. 485-x, ICAP, air-rights transfers, discounted payoffs. Each play is a set of checkable conditions, not an opinion, and the library grows as programs change.
What happens when a program sunsets or a law changes?
The play is updated with the change, and because every conclusion is dated and sourced, you can always see which version of the rule an answer was checked against.
Can a play be wrong?
A play can only be as right as the record and the rule text behind it. When the record is stale or silent, the scan says not evidenced rather than guessing. The failure mode is a visible gap, not a confident error.
Do the 48 run on every building?
Every scan runs the full library. Plays that require data a jurisdiction does not publish come back not evidenced for that reason, and say so.
See your market.
Twenty minutes on the live terminal.