About Crescent.
Crescent is the computational pricing layer for healthcare. We build, run, and publish from a single team. This page describes what the firm is, what we believe, and where we are going.
the firm
A research-and-execution firm.
Crescent is a research-and-execution firm with one focus: computing the correct price for every healthcare claim in America. The firm builds the engine, runs the diligence engagements, and publishes the research from a single team. There is no separation between engineering and engagement delivery and research output. Each is produced by the people doing the others.
The firm is headquartered in New York. Crescent Intelligence, Inc. is privately held.
the thesis
On the absence of a pricing layer.
Every important industry in America has a pricing layer except the largest one.
Fixed income has Bloomberg and TRACE. Insurance has Verisk. Mortgages have Optimal Blue. Real estate has CoStar. Healthcare — 5.6 trillion dollars, the largest sector of the U.S. economy — has nothing.
The reason is not that the problem is unimportant. The reason is structural. Healthcare is the only major asset class whose price must be computed rather than observed.
Every other asset class has a price that emerges from one of three mechanisms. Market discovery, where buyers and sellers clear in real time. Contractual definition, where the price is a short formula on public inputs. Commodity benchmarking, where a public reference sets the rate. All three are observation problems. You look up the price.
Healthcare is different. The price of a knee arthroscopy with modifier 59 performed in POS 24 on a Medicare Advantage patient under a specific payer contract is a deterministic function of roughly five hundred interacting rules spanning the CFR, the MPFS, NCCI edits, MUE limits, local LCDs, contract addenda, sequestration adjustments, and patient deductible status. The price exists. The formula exists. The answer has never been computed at scale, because the computation requires encoding the entire regulatory corpus, normalizing private payer contracts, and resolving rule interactions across a five-hundred-dimension state space. Nobody has done it.
So the entire industry runs on approximations of a quantity that has an exact answer. Revenue cycle vendors guess at it with machine learning. Normal quality-of-earnings practices sample 100 claims out of 200,000 and extrapolate. Private credit lenders underwrite to face value with a haircut. M&A buyers pay for cash flows they cannot verify. Every transaction in healthcare finance is mispriced because nobody has done the math.
The contrarian truth: healthcare is not a service industry. It is the largest mispriced receivables market in the world. And the entire industry’s reliance on probabilistic estimation of a deterministic quantity is a sixty-year category error that ends the moment the pricing layer exists.
The historical analogues are clear. Each pricing layer in the developed world created the asset class it priced. The U.S. unlocked mortgage capital with the conforming loan and the GSE pricing standard. It unlocked corporate capital with public exchanges and audited financial statements. It unlocked consumer credit with the FICO score. None of these existed when the asset classes they price were younger; each was built by an institution willing to invest the years required to encode a massively complex system. Each was worth, eventually, more than the companies it served.
Healthcare receivables are the largest unpriced asset class remaining in the developed world. Crescent is the pricing layer.
We are not building a diligence tool. We are not building a billing product. We are not building a healthcare AI company. We are building the missing financial infrastructure of the largest sector of the economy — and the first company to build pricing infrastructure for an asset class is the company that owns the asset class it creates.
operating beliefs
How we think about the work.
The firm operates on a small number of positional claims that shape what we build, how we engage, and what we publish. Each is load-bearing.
01
Healthcare payment is a computable function, not a probability problem.
The realized payment for a healthcare claim is the deterministic result of a finite rule system applied to a finite state space. The accuracy ceiling of the ML approaches the industry has used for two decades is structural, not architectural. The right method is not better sampling. It is no sampling at all.
02
Outputs must be defensible to the law and contract that made them correct.
A computed amount without a citation chain is indistinguishable from a model prediction. Every adjustment the engine produces traces back to the controlling authority. The trace is the difference between an opinion and a proof. We treat traceability as a first-class architectural feature, not a compliance afterthought.
03
Research is the firm’s lifeblood.
We publish methodology notes and field notes because the work requires us to think rigorously about problems no one else is thinking about rigorously. The research corpus is the public face of an internal intellectual project. It is not a content marketing program.
04
Build, run, and publish from a single team.
The engineering, engagement delivery, and research outputs are produced by the same people. Separation between “engineering” and “client services” and “thought leadership” is the structure that produces mediocre output in three directions. Our structure produces unified output in one.
where we are going
The pricing layer is the endgame.
Reimbursement quality of earnings engagements are the firm’s first commercial product. The engagements produce the data, the relationships, and the credibility required to build something larger.
The long-term project is the pricing layer that every counterparty in healthcare references when it needs to know what a payment is worth. Lenders underwriting healthcare receivables. Asset managers pricing healthcare-backed securities. Operating partners running portfolio companies. Researchers studying payer behavior at scale.
The pricing layer is durable infrastructure. Verisk in property and casualty insurance. BlackRock’s Aladdin in asset management. CoStar in commercial real estate. These are the analogous positions. None of them existed when the asset classes they price were younger; each was built by an institution willing to invest the years required to encode a massively complex system.
Crescent is that institution for healthcare. The work is multi-year. The trajectory is non-linear. The destination is the substrate on which healthcare capital allocates.
leadership
Seven years of hands-on healthcare billing and adjudication experience starting in a family medical practice. One year of healthcare private equity diligence experience. Published research on cancer and machine learning, presented at the American Society for Radiation Oncology. Prior founder of an education technology company, sold in 2024.
If you want to work with us, send a note to the engagement desk. If you want to work for us, send a note to careers.