01
Loan by loan
Every analysis starts from individual loans, from origination to payoff, assignment or claim. Nineteen years of buying, hedging and managing loans taught us the answer usually sits in a few hundred of them.
The questions are hard: what a book will cost to carry and when it will leave, what a servicer's curtailments are really costing, what servicing rights are worth, what is actually driving speeds. The data behind them is harder. Roland Street answers them plainly, from loan-level data, for the people who build, hold and stand behind the credit: specialty finance companies, investment managers, banks, and anyone who needs an independent view. Home equity is a particular depth: HECM and proprietary reverse mortgages, home equity investments and HELOCs.
One client at a time · Critical thinking · Always a real person
Day one
Exposure starts at origination, not at a threshold
Loan by loan
Every answer is built from individual loans, not summaries
19 years
Buying, hedging and managing loans as an investor
Home equity
Reverse mortgages, home equity investments and HELOCs
Five parties hold a stake in the value and amount of credit in a mortgage. Each one's exposure begins the day the loan is originated and runs until payoff, default or claim. The risks attached to the loan are the same for all of them; the product decides which ones matter most.
Show the risks for
Every mortgage
Default
The borrower stops paying, or stops meeting the terms of the loan.
Prepayment
Balance leaves early, through refinance, sale or payoff.
Collateral value
What the home is worth at the moment it matters.
Servicing cost
Advances, timelines and process, from the day the loan is boarded.
Reverse mortgages · HECM and proprietary
Crossover
The balance compounds until it passes the value of the home.
Buyout and assignment
Issuers buy loans out of HMBS pools and fund them until they assign or resolve.
Mortality and move-out
The loan ends on a life event, not a payment schedule.
Tax and insurance default
The borrower's remaining obligations, and the usual path to default.
Home equity investments
Home-price path
The return is a share of the house, so the path of its value is the return.
Exit timing
Sale, refinance, buyback or a life event, each on its own clock.
Senior liens
The investment sits behind every lien on the home.
Valuation at exit
Settlement depends on what the home is judged to be worth at the end.
HELOCs
Draw behavior
The balance moves with what the borrower draws on the line.
End of draw
Payments step up when the draw period ends.
Second-lien severity
Recovery depends on what is left after the first lien is paid.
Rate reset
The rate follows the index, and the payment follows the rate.
Home equity products depend on home prices and life events as much as on rates, and few people have looked at them loan by loan. This is where we have spent the most time.
A reverse mortgage balance grows instead of shrinking, and the loan ends on a life event, not a payment schedule. On a HECM the insurance fund stands behind the crossover; on a proprietary loan the lender or bondholder does.
See these risks on the diagram ↑
FHA
A HECM balance compounds until it passes the value of the home, and the Mutual Mortgage Insurance Fund pays the difference at claim.
Crossover projections by vintage, product and geography, built on the same loan-level population the fund is standing behind, not on an actuarial average of it.
Ginnie Mae
When an HMBS issuer can't fund its buyouts, the guarantor inherits the portfolio and every obligation attached to it.
Issuer concentration and liquidity-stress views: buyout pipeline against issuer size, non-assignable share, tax-and-insurance default trend. The indicators that move before an issuer does.
Issuers & servicers
Buyouts begin the day a loan is boarded. Loans that will not assign stay on the balance sheet.
Loan-level forecast of buyouts by month across the life of the book, assignability screening, facility sizing and covenant headroom, and a plan for the loans that will not assign.
Accountants & auditors
HMSR and buyout-loan values rest on timing and cost assumptions few can test independently.
Documented assumption sets with portfolio evidence for the audit file, and the sensitivities an audit committee can read.
Bond investors
Assignments, foreclosures and claims are what drive HMBS speeds.
Monthly CPRs by cohort, split into assignment, voluntary payoff and mortality or due-and-payable. The assignment piece is arithmetic; the rest is behavior. A blended number hides which one moved.
Proprietary reverse
On a proprietary reverse mortgage, crossover is a loss to the lender or the bondholder, not an insurance claim.
Crossover, longevity and home-price projections loan by loan, for originators, warehouse lenders and securitization investors.
The investor's return depends on what the home is worth and when the homeowner exits. Both are projections; neither shows up in a payment history.
See these risks on the diagram ↑
Valuation
Value moves with the home-price path and the exit date at the same time.
Valuation loan by loan across home-price paths and exit timing, with the assumptions written down.
Timing
Exits come from sale, refinance, buyback or a life event, each on its own clock.
Exit-timing projections by vintage, geography and homeowner profile, checked against what has actually exited.
Collateral
The investment sits behind every lien on the home and depends on its condition and market.
Lien position, combined loan-to-value and market concentration, loan by loan.
The balance moves with the borrower's draws, the rate resets with the index, and the position usually sits behind a first lien.
See these risks on the diagram ↑
Utilization
Draws follow rates, home prices and the end of the draw period.
Utilization and draw behavior by cohort, and what happens to payment and credit at the end of the draw period.
Severity
Recovery depends on what is left after the first lien is paid.
Loss severity by combined loan-to-value, geography and home-price path.
Speeds
Paydowns, payoffs and line closures look alike in a blended number and mean different things.
Monthly speeds split by cause, by cohort, from loan-level data.
Judgment first: the right way to look at a position, the risks that matter and the traps underneath the data. You call for the judgment. Our technology makes it repeatable.
01
For new lenders and specialty finance companies
Build it right the first time: product and asset design, origination and data systems, and the models and reporting capital partners ask for, built so the loan-level record is right from the first application.
02
For investment managers and banks
Enter the market with your eyes open: underwriting, valuation and monitoring frameworks for mortgage and home equity assets, what to buy, what to pay, what to watch, and the traps in each. Analytical depth for teams without a modeling group of their own.
03
For boards, CFOs, auditors and lenders
A second set of eyes: independent assumption sets and valuations for loans, servicing rights and buyout and claim assets, and a written opinion on a model, a pool or a trade, written by the person who did the work.
04
For issuers, servicers, insurers and guarantors
Loan-level forecasts of buyouts and advances, facility sizing and covenant headroom, assignability screening on HMBS, and timeline and curtailment recovery on FHA claims. For those who stand behind the credit, independent issuer and portfolio surveillance.
05
For lenders, investors and platforms
Valuation, performance and risk work on HECM and proprietary reverse mortgages, home equity investments and HELOCs: products whose cash flows depend on home prices and life events as much as on rates.
06
Optional · alongside an engagement
Once the work is built, it can run every month on your population: buyouts and repurchases, CPR decomposition, endorsements, claims and curtailment. Sent as documents on a schedule we manage.
Engagements are scoped one at a time. There is nothing to log into.
Most analysis in this market runs on summaries. Ours starts from individual loans, projects each one forward through its whole life, and keeps every result, so the work can be rolled up, split apart or retrieved later.
01
Every analysis starts from individual loans, from origination to payoff, assignment or claim. Nineteen years of buying, hedging and managing loans taught us the answer usually sits in a few hundred of them.
02
Results roll up to the portfolio and split back down along any line: vintage, product, servicer, geography, status, or any combination. Every number at the top traces to the loans underneath it.
03
Assumptions are explicit, documented and tested against what actually happened. What is mechanical is modeled as arithmetic and what is behavior as behavior, so a forecast can be held to account month after month.
04
Loan-level data and every projection are stored by month in our own databases, never overwritten. Any past month can be pulled as it stood, and the answer arrives on a page with the loan detail behind it.
Charlotte, NC · working nationally · small client roster, by design
Roland Street is a boutique, and every engagement is done by the principal personally. Clients pay for judgment — the best way to look at a position, the real risks in it and the traps underneath the data — and for answers that are clear without being over-qualified. We are not a software company and not an investment bank. We read loans.
Principal — Capital Markets
Two decades on the institutional side of mortgage finance. Started as a mortgage analyst and then a trader at a bulge bracket, built a similar trading platform at a boutique bank; then managed mortgage and MSR portfolios at an operating company, buying, hedging and selling them. Taught himself to code, run databases and build predictive models before there was AI to do it, because the answers were needed faster than they could be requested. The job was always the same: take something genuinely complicated, understand what drives it, and turn it into one clear picture a decision-maker can act on.
Thirty minutes, no pitch, no slides. We will tell you honestly whether we can help, and if we cannot, who might.
Or email rich@rolandstreet.com.