The Returns tab keeps the five result cards, the decisive margin and the limitation notes, and folds the snapshot, tables, decomposition, the eight tests, the grade and the lender view behind one Show full analysis control, which Why this verdict opens and print always includes. The inputs that move the verdict (price, hold, revenue, exit cap, LTV and rate) lead and stay open; the other groups fold with a count and open themselves for a flagged, sample or edited input, or a figure taken or computed from a document. IRR swings read in pts, and the verdict word leads the plate. No figure changes.
What shipped, and when.
A day-by-day record of what landed in Underwrite, kept by hand. Page 2 of 23.
Want to verify? The six files the decision depends on are rendered from the live source at /methodology/engine.
Filter by engine, UI, AI, ops, marketing or fixes.
The desk keeps the verdict on screen and folds the rest behind one control, the IC memo reads as a decision memo in chapters, the charts are redrawn, three engine paths are corrected, commercial expense recoveries arrive as an optional input, the home page is shorter and the pricing page names one plan. Card checkout is not open yet. The canonical figures do not move: 15.34% IRR, 2.44x and PURSUE at a 6.50% exit cap, 13.51% IRR, 2.19x and PASS at 7.00%.
From 1280px, once the desk is open, a slim title bar carries every desk action and a Site menu, and the inputs are a column that scrolls on its own beside the plate, so typing a 7.00% exit cap and reading PASS at 13.51% fit on one screen at 1280 by 720. The parser can sit in a Source documents column beside the plate, so a parsed figure's quoted sentence and the verdict share the screen: it is open by default from 1600px, opens itself on a first visit from 1440px, and the title bar toggle opens or closes it at any width from 1280px. Below 1280px the desk keeps its single column.
The charts share one tested palette, show exact figures on hover and use round axis ticks and fixed text sizes. NOI plots through the hold with the sale year's NOI labeled and DSCR through the loan term, the tornado names the 15.00% hurdle, IRR by LTV and by hold are lines against the hurdle, the stress presets are a dot plot, the waterfall is a cumulative equity line with its payback year, and the sensitivity heat table steps above and below the hurdle with a legend. Monte Carlo says how many of the 1,000 trials give no IRR and draws its histogram on the trials' own range, and value labels are placed clear of lines, points and each other.
Compare shows a pooled IRR from summed cash flows only when every deal has one, names a deal with none and adds a hurdle strip; the one-bar asset mix is gone. The macro page states one refresh time instead of three, the lending conditions meter is a neutral track with a marker, and the rate stack is a table.
The Full memo, which showed all 178 sections, now reads in chapters (Deal at a glance, Operations, Capital, Risk, Returns detail, Rent roll, Diagnostics), with the seeded reference overlays in one closed appendix at the end. It has one Executive Summary, the grade is one muted line under it and not a second verdict, IRR moves read in pts and percentages print at two decimals.
The DSCR Stress section names its tests from the engine: the verdict tests the 1.00x coverage floor, the 1.20x lender floor is advisory and does not gate it, and the 1.25x stress floor is a stricter test of its own. Overlay text no longer names data vendors or claims integrations the product does not have, two overlays that printed $NaNK and NaN bps say the figure could not be estimated, the co-invest sweep prints n/a where it has no figure, and a few figures no longer run into the word after them. The link preview image labels its bar Grade score; it was labeled Deal Strength, the name of a different figure in the memo.
Each cause of a missing project IRR now reads as one plain line on the plate, the memo, the tornado and the sensitivity grid, with a new cause: the equity gets back less than it puts in over the hold. The plate says the sale does not cover the loan only when the sale year's cash flow is below zero and the sale, net of selling costs, is short of the balances it pays off.
When a refinance fell in the exit year, such as a hold half a year past the loan term, the sale never repaid the new loan: the worked example on that structure read 19.07% and PURSUE against 15.34% for the same deal without the refinance. The sale now repays the new loan, so the equity flows match the same deal with no refinance. Every other refinance is unchanged.
The co-invest exit counted operating years twice and moved money from the GP's promote to the LP: on the worked example at 10% co-invest the LP received $6,451,587.21 against an entitlement of $6,327,886.28. Each party is now paid what it has not yet received of its entitlement. Only the co-invest split moves.
A 7-year class held 10 years recovered 10/7 of its basis. The accelerated class now stops depreciating at its life. This changes after-tax figures only on cost segregation deals held past the class life; the worked example does not move.
A new input group, off by default, lets commercial tenants recover their pro rata share of expenses on occupied space under NNN, modified gross (base year or expense stop) or a fixed CAM charge, with an optional admin fee and a cap on controllable CAM growth, credited to EGI as its own line. Share links, saved deals, the workbook, TRACE and the methodology page carry it, and /trust now counts eight registered limitations. The worked example is unchanged.
A complete operating statement that lists no reserve line now keeps the desk's reserve (2.00% on a new deal, marked Sample assumption) rather than reading it as 0.00%, and the line beside the verdict names it. A reserve line the document states still applies.
A small state tax line, under $10,000 a year, is listed as not taken with a note to check it instead of being dropped as income tax. A stated retail rent that the tenant rows do not add to is named with both figures and the gap, and a period note on a verified sentence shows in quiet ink instead of the amber that means check by hand.
- marketing
The home page has five sections instead of eleven: the hero, a 30-second proof, how it works, Check the work and pricing with the founder note. The second button reads Try the worked example, and labels are 14px in sentence case.
Fields are 16px so the phone does not zoom, buttons and button-shaped links are 44px targets, wide tables and boxes show a swipe cue, chip rows wrap and the macro grids fit. A shared link lands with its verdict word on the first screen at 360 by 780, the stress dot plot puts each scenario's name and figure above its track, and the memo footer wraps.