This workbook ranks your real Loan Estimates on the question you are actually asking: what will borrowing this money cost me by the year I expect to sell or refinance? It stops mortgage insurance at the 78% month your own schedule reaches — naming the earlier 80% month, when you may ask for it — finds the month an offer that cost more at the closing table overtakes the one that cost least, and puts a dollar figure on what a slipped closing would cost against your rate lock.
Three lenders send you three Loan Estimates. They are the same federal form, three pages each, laid out identically — and after twenty minutes with them side by side you still cannot say which one is cheapest. That is not a failure of attention. It is what happens when the rate, the points, the lender credit, the closing costs and the mortgage insurance all pull in different directions and the summary number everyone reaches for, the APR, quietly assumes you keep the loan for thirty years.
Why the lowest rate isn’t the cheapest loan
Three things, and only three, decide what borrowing costs: what you hand over at the closing table after any lender credit, the interest you actually pay before you get out, and the mortgage insurance you actually pay before it stops. Not the monthly payment, which says nothing about how much of it is principal coming back to you as equity — so a lower payment can still mean a more expensive loan. Not the APR, which spreads your closing costs across the full loan term whether or not you keep the loan that long.
Enter each Loan Estimate once, and the workbook totals those three at three, five, seven and ten years, at full term, and at the horizon you type in. Each block ranks the offers on its own. When the winner changes between them, it says so and names the year it changes — because that, not the headline rate, is the finding.
One thing it will not pretend to know: an adjustable-rate offer is amortized at its starting rate for the whole comparison, because what it becomes after its first adjustment depends on an index nobody can quote you today. Its totals are therefore a best case, and the workbook says so on the tab that prints them — right beside a computed row telling you how far your horizon runs past that offer’s fixed-rate period, which is the fact that decides whether the best case is close to certain or a guess.
The month your mortgage insurance actually stops
This is where a comparison goes quietly wrong. Leave mortgage insurance out and a small down payment looks cheaper than it is. Run it flat for thirty years and it looks far worse. Neither is what happens: it stops.
The workbook builds a 360-month schedule for every offer and solves two months from it — the first month the scheduled balance reaches 80% of the original value, when you may request cancellation, and 78%, when the lender must terminate it on its own under the federal Homeowners Protection Act — automatic, but only if you are current on the loan. It then counts the charge only that far — to the 78% month, unless you ask at 80% and get it. Those are different months for every offer, because they depend on the rate, and the gap between the two is months of a payment that buys you nothing at all if you never ask.
When the cheap rate finally pays for itself
An offer that charges you more at the table in exchange for a lower rate is behind from day one and catches up later. The only thing worth knowing is when — and whether you will still own the loan by then.
The workbook runs both running totals forward month by month and reports the exact month each offer overtakes the one that cost least to close, then checks that month against your own horizon. If you plan to be gone before it arrives, the money you paid up front to buy the rate down never comes back, and it tells you so in a sentence. That is the whole of the points-versus-credits argument, answerable in one number as soon as you are honest about the horizon.
And what a slipped closing would cost you
A shorter lock at a better rate is a better deal right up until your closing slips. Lock extension fees are the surprise cost of this whole process, and nothing in the rate you were quoted accounts for them.
Put in how many days until your expected closing, how long a slip typically runs, and how likely you think one is, and every offer carries a priced expected extension cost — a real dollar figure sitting on the comparison beside the rate.
Weighted where the numbers run out
Money is only part of it. Lock length, a float-down option, an appraisal waiver, whether there is a named person answering the phone, how long underwriting really takes — those decide plenty and no rate comparison we could find scores them at all.
Eight criteria, weighted by you. Four of the eight the workbook answers itself: what the loan costs at your horizon, the cash you need at the table, what a slipped closing would cost, and rate certainty — which it reads from the loan product, setting each offer’s fixed-rate period against the horizon you entered. It hands you those four as suggested ratings on the same 1-to-5 scale, and you can overrule any of them.
And when the top two land inside the noise of ratings a person typed by hand, it says too close to call rather than inventing a winner.
Neutral by construction
Plenty of the mortgage calculators a borrower lands on belong to somebody hoping to lend them the money. This one names no lender, endorses none, and earns nothing from which offer wins. It fetches no live rates, connects to no lender, and predicts nothing about where rates go next — every figure in it comes from the Loan Estimates you type in. The ranking is your weights over your own Loan Estimates and nothing else.
Own it, don’t rent it: it is bought once, there is no account, and nothing renews. Keep the file — a refinance is this same comparison with one of the offers already in your name, so when rates move you put your current loan in as a column and score the new quotes against it.
Try the method first, free
Want to see how it thinks before you buy? The free Mortgage / Loan-Offer Comparison Scorer takes two offers, works out what borrowing each one costs by the year you expect to get out, and scores them across five of the criteria — in your browser, no signup, with a starter spreadsheet and a printable scorecard to take away. The workbook is the full version: five offers, all eight criteria with weights you set, and the mortgage-insurance, crossover and rate-lock engines the free tool deliberately leaves out.
What you’re comparing against
A blank spreadsheet gets you a table. The free shopping worksheets we looked at get you a place to write the numbers down, and nothing that computes. A one-shot AI prompt gets the monthly payment right and then, in the answers we tried, had no amortization schedule to derive a cancellation month from, no crossover grid, no price on the lock, and picked a winner using weights it chose for you.
The workbook arrives with four fictional example offers already filled in, so the whole method is running the moment you open it — then you overwrite them with your own. In that example the cheapest way to borrow turns out to be the adjustable-rate offer, amortized at its starting rate: a best case, not a projection, and exactly the kind of result the rate-certainty criterion exists to price.
A decision aid, not advice. This is a planning template, not financial, mortgage, tax or legal advice, and no substitute for your own Loan Estimate, your Closing Disclosure, or a licensed professional. The example offers and every figure attached to them are fictional. Two things it deliberately does not model, and says so on the tabs that would otherwise imply it: an adjustable rate after its first adjustment, which it amortizes at the starting rate and flags as a best case, and property taxes, hazard insurance and HOA dues, which you owe identically whichever lender you pick. The 80% and 78% thresholds are the federal Homeowners Protection Act rules for most borrower-paid mortgage insurance; FHA mortgage insurance premiums follow different rules. Confirm every figure against your own Loan Estimate before you lock.
