← Back to Interaction Rule Set

CE-E01: Deal Science BIX Reference

CONFIDENTIAL. PreFi, Inc. / Purpose Technology, Inc. d/b/a Purlend.

THE CLARITY ENGINE

Science of the Deal Math • Logic • Alchemy BIX Development Reference | Purlend / PreFi, Inc. | CONFIDENTIAL

This document defines the complete technical and conceptual framework BIX must internalize to build the Clarity Engine — the borrower intelligence core of Purlend’s Digital Lending Highway. It covers the hard math of mortgage qualification, the logical decision architecture of the engine, and the hidden alchemy of deal outcomes that transform a borrower’s financial reality into an executable financing strategy.

PART I: THE MATH OF A MORTGAGE DEAL

Before the engine can propose, validate, or execute — it must calculate. Every borrower financing decision reduces to a set of mathematical relationships. BIX must model these precisely.

1. INCOME ANALYSIS — THE FOUNDATION OF QUALIFICATION

1.1 Gross Monthly Income (GMI) Calculation

GMI is the engine’s primary input variable. It drives every downstream calculation.

Income Type Calculation Method Documentation Required W-2 Salaried Annual salary ÷ 12 2 years W-2s + current paystubs Hourly Full-Time (Hourly rate × Hrs/wk × 52) ÷ 12 Same as W-2 Overtime / Bonus 2-yr avg if likely to continue Written VOE from employer Self-Employed (1099/LLC) 2-yr avg net after add-backs 2 yrs tax returns + P&L Rental Income 75% of gross rent (vacancy factor) Schedule E + lease agreements Social Security / Pension Gross award letter amount Award letter, not taxed gross-up Non-taxable SS / Disability Gross-up by 25% (Fannie/FHA) Award letter + history Commission (>25% of income) 2-yr avg; declining = lower used 2 yrs returns + YTD paystubs

CLARITY ENGINE LOGIC

Income must be stable, verifiable, and likely to continue for 3 years. The engine must flag declining income trends, recent job changes (<2 yrs), and gaps exceeding 30 days. A borrower’s stated income is never the qualifying income — it is the starting point for validation.

1.2 Self-Employed Income: The Add-Back Algorithm

Self-employed borrowers are systematically underqualified by traditional lenders. The Clarity Engine must execute the full Schedule C / S-Corp / Partnership income analysis.

HIDDEN ALCHEMY

A borrower showing $60K/yr net income on taxes may qualify at $90K/yr after legitimate add-backs. This is the engine’s first ‘unlock’ — surfacing hidden qualifying income before the borrower even applies.

2. DEBT-TO-INCOME RATIO — THE GATEWAY CALCULATION

2.1 Front-End DTI (Housing Ratio)

PITI = Principal + Interest + Taxes + Insurance (+ HOA + MIP/PMI if applicable)

Formula Fannie Mae Thresholds Front-End DTI = PITI ÷ GMI Max 28% (guideline); 36% manual underwrite Back-End DTI = (PITI + All Debts) ÷ GMI Max 36% manual / 45% with conditions / 50% DU

2.2 Back-End DTI: The Full Liability Stack

The Clarity Engine must pull and calculate ALL liabilities from credit bureau data:

CLARITY ENGINE LOGIC

The engine must evaluate each debt for elimination eligibility. Student loans on income-driven repayment plans, co-signed debts with payment history, and business debts are three high-value debt reduction pathways. Always calculate ‘DTI with removal’ as a parallel scenario.

2.3 The DTI Optimization Matrix

When a borrower fails DTI, the engine must present a structured path. The following variables can be manipulated:

Lever Effect on DTI Engine Action Pay down revolving debt Reduces monthly min payment Calculate exact paydown needed for qualification Eliminate installment debt Removes fixed monthly obligation Flag debts <10 payments remaining (exclude) Add co-borrower income Increases GMI denominator Model combined vs individual scenarios Increase down payment Reduces PITI (P&I component) Show loan amount reduction required Buy down interest rate Reduces P&I payment Calculate cost of buydown vs benefit Select different loan product ARM vs fixed rate impact Qualify on initial ARM rate (with 5/1 ARM)

3. LOAN-TO-VALUE (LTV) — THE RISK EQUATION

3.1 Core LTV Calculation

LTV Range PMI Required Rate Impact Product Eligibility ≤80% No Best pricing Conventional, jumbo, all 80.01%-90% Yes (or piggyback) Slight premium Conv. / FHA competitive 90.01%-95% Required Moderate premium Conv. 3-5% down programs 95.01%-96.5% Required Higher premium FHA (3.5% down) primary 96.5%-100% Required / funded Maximum VA (0% down), USDA

HIDDEN ALCHEMY

The 80% LTV threshold is a hard economic line. The engine must always calculate: (1) exact dollar amount needed to reach 80%, (2) whether gift funds, seller concessions, or DPA programs can bridge the gap, and (3) whether PMI cost vs rate improvement math favors one path.

4. CREDIT SCORE ANALYSIS — THE RISK CLASSIFIER

4.1 The Three-Bureau Model

4.2 Score Tier Impact Matrix

FICO Score Rate Adjustment PMI Rate Impact Product Access 760+ Best / par rate Lowest tier All conventional + jumbo 740-759 +0.125% to +0.25% Tier 2 Full conventional access 720-739 +0.25% to +0.50% Tier 3 Conventional with overlays 700-719 +0.50% to +0.75% Tier 4 Conv. / FHA competitive zone 680-699 +0.75% to +1.00% Tier 5 FHA often better economics 640-679 +1.00%+ Tier 6 FHA primary; VA if eligible 580-639 FHA minimum floor N/A conventional FHA 3.5% down only; VA

4.3 Rapid Rescore Protocol

The engine must identify the fastest path to score improvement and model the resulting rate/payment impact:

CLARITY ENGINE LOGIC

For every borrower with a score below 740, the engine must run a ‘Score Improvement Scenario’ — modeling the monthly payment savings at the next score tier against the time and cost required to achieve it. This is the borrower’s ROI of waiting vs. acting now.

PART II: THE LOGIC ARCHITECTURE

The math produces numbers. The logic determines what those numbers mean, which scenarios to run, and what actions to propose. This is the decision-making layer of the Clarity Engine.

5. THE BORROWER INTELLIGENCE LEDGER

5.1 Core Data Primitives (from URLA Form 1003)

The Clarity Engine ingests the Fannie Mae URLA Form 1003 + Lender Loan Information + Unmarried Addendum. Key data primitives the engine must store and process:

BORROWER IDENTITY PRIMITIVES

FINANCIAL PRIMITIVES

5.2 The ‘Always Approved’ State System

The Clarity Engine’s foundational architecture is not a pass/fail system. It is a state machine with one terminal state: ‘Always Approved.’ Every borrower exists in one of three states:

State Definition Engine Response

APPROVED NOW

All qualification criteria met today Execute: generate pre-approval, initiate pipeline

APPROVED WITH ACTION

Qualifies after specific, measurable steps Prescribe: issue Action Plan with timeline and milestones

PATH IN PROGRESS

Not yet qualified; engine tracks progress Monitor: automated check-ins, life event triggers, re-scoring

ARCHITECTURAL IMPERATIVE

The engine never tells a borrower ‘you don’t qualify.’ It always tells them ‘here is your path.’ This is the product truth that makes Purlend categorically different from every existing mortgage platform.

6. THE DECISION TREE: PRODUCT SELECTION LOGIC

6.1 Loan Product Priority Waterfall

The engine evaluates loan products in a structured waterfall based on borrower profile. BIX must implement this evaluation sequence:

6.2 Conventional vs. FHA Decision Algorithm

This is one of the engine’s highest-value calculations. The breakeven analysis:

FORMULA: FHA MIP vs. PMI Breakeven FHA Monthly MIP = (Loan Amount × 0.55%) ÷ 12 (for 30-yr, LTV > 90%) Conventional PMI varies by LTV, FICO, and PMI provider (0.2%-2.0% annually)

If Conv PMI > FHA MIP: FHA wins on monthly cash flow If Conv PMI < FHA MIP AND borrower reaches 78% LTV within 7 years: Conventional wins Always model: Time to PMI removal (Conventional) vs. Life-of-loan MIP (FHA, if LTV >90%)

7. SCENARIO ENGINE: THE PARALLEL SIMULATION FRAMEWORK

7.1 Required Scenario Matrix (Every Borrower)

The Clarity Engine must generate a minimum of three scenarios for every borrower interaction. The BIX architecture must support parallel scenario computation:

Scenario Parameters Purpose Scenario A: As-Is Current income, debt, credit, assets Baseline qualification state Scenario B: Optimized After credit improvement, debt paydown, income add-backs Best achievable now Scenario C: Future State With 6/12/18-month action plan executed Path to optimal deal Scenario D: Co-Borrower Add spouse/partner/family member income and debt Combined qualification Scenario E: Product Alternatives Same profile across VA/Conv/FHA/Non-QM True cost comparison

7.2 Rate vs. Point vs. Payment: The Buydown Matrix

The engine must calculate the full economics of rate buydown options:

EXAMPLE CONVERSATION OUTPUT

Borrower at $450K loan, 7.25% rate:

PART III: THE ALCHEMY — HIDDEN OUTCOMES & DEAL OPTIMIZATION

Alchemy is the Clarity Engine’s competitive moat. It’s the surfacing of non-obvious opportunities that turn a declined application into a closed loan, or a mediocre deal into an optimal one.

8. ASSET ALCHEMY: UNLOCKING DOWN PAYMENT

8.1 Non-Obvious Asset Sources

HIDDEN ALCHEMY

A borrower with $30K in a 401(k), $5K in savings, and a willing seller can often purchase a $300K home with near-zero out-of-pocket by combining DPA + seller concession + 401(k) as reserve documentation. The engine must calculate this combination automatically.

9. DEAL STRUCTURE ALCHEMY

9.1 The Optimal Offer Architecture

The Clarity Engine must coach borrowers and their agents on structuring purchase offers to maximize financing outcomes:

SELLER-FAVORABLE TERMS

BUYER-FAVORABLE EMBEDDED TERMS

9.2 The Refinance Trigger Logic

The engine must maintain ongoing monitoring of refinance opportunity for every closed borrower. Trigger conditions:

10. THE SEVEN D’s: LIFE EVENT TRIGGER TAXONOMY

The engine’s long-term intelligence layer monitors borrower life events that create financing needs. These are the Purlend ‘7 D’s’ — the trigger taxonomy that powers the pipeline.

Trigger Life Event Financing Opportunity Diapers (1st) New baby / first child Upsize purchase, nursery renovation HELOC, life insurance review Diamonds Engagement / marriage First home purchase, consolidate dual rentals, combine finances Diapers (again) Additional children Upsize again, school district relocation, cash-out for college fund Deployment Military service SCRA rate cap (6%) enforcement, PCS relocation, VA loan optimization Downsizing Empty nest / retirement Downsize purchase, reverse mortgage evaluation, investment property Divorce Marital dissolution Title buyout refi, cash-out refi for settlement, one spouse’s new purchase Death Loss of household member Estate refi, title transfer, heir purchase, reverse mortgage payoff

ARCHITECTURAL NOTE FOR BIX

The 7 D’s are data primitives in the borrower intelligence ledger. The engine must monitor signals from Revaluate (purchase propensity), Attom Data (property events), and Clear Capital (AVM changes) to detect these triggers in real time and initiate automated outreach workflows.

PART IV: BORROWER CONVERSATION MODELS

The engine must translate complex calculations into clear, actionable borrower conversations. These are the canonical conversation frameworks BIX must support.

11. CONVERSATION TEMPLATES

11.1 The ‘You’re Not Ready Yet’ Conversation (Reframe to Path)

Speaker Dialogue Borrower “I was told I don’t qualify for a mortgage. My credit isn’t good enough.” Engine Output “Your current FICO of 618 puts you just below the conventional threshold, but you’re only 22 points from FHA eligibility and 62 points from the rate tier that saves you $187/month over the life of your loan. Here’s your 90-day path: Pay your Discover card from $4,200 to $1,260 (30% utilization). That single action, based on your credit profile, projects a +31-point increase. We’ll check your score at day 45 and initiate pre-approval at day 90.” Engine Output “While you’re on the path, I’m monitoring homes in your target zip codes. When you hit 640, I’ll have your pre-approval ready within 24 hours.”

11.2 The DTI Problem Conversation

Speaker Dialogue Borrower “The bank said my debt-to-income is too high. I make $95K a year.” Engine Output “At $95K salary, your qualifying income is $7,917/month. At the $425K purchase price, your PITI is $2,940, giving you a 37.1% back-end DTI — above the 36% manual threshold. But here’s what the bank didn’t tell you: your car loan has 9 payments remaining. Under Fannie Mae guidelines, debts with fewer than 10 payments can be excluded. Without that $487/month car payment, your DTI drops to 30.9%. You qualify today.” Engine Output “I’ve also run a DU approval scenario. With automated underwriting, your file scores at 44.8% DTI — within the 45% DU threshold even without excluding the car payment. You have two clean paths to approval.”

11.3 The Self-Employed Income Unlock Conversation

Speaker Dialogue Borrower “I own my own business but my tax returns only show $52,000 net income.” Engine Output “Your Schedule C shows $52K net, but let’s run the full add-back analysis. You have $18,400 in depreciation, $4,200 in home office, and $6,100 in one-time business startup expenses that won’t recur. Your actual qualifying income is $80,700 — a 55% increase. At that income level, you qualify for a $340,000 loan, not $220,000. The 24-month average across your two returns is $78,400. That’s your qualifying number.”

11.4 The Veteran First-Timer Conversation

Speaker Dialogue Borrower “I served 6 years active duty and never knew I had a VA loan benefit.” Engine Output “Your VA entitlement is one of the most powerful financial benefits you’ve earned and it’s never expired. With your honorable discharge after 6 years of service, you qualify for a VA loan with zero down payment, no private mortgage insurance, and a rate typically 0.25%-0.50% below conventional. On a $400,000 home, that’s a $0 down payment instead of $80,000, and approximately $210/month less than a conventional FHA loan. Your disability rating also means your VA funding fee is waived entirely. Let’s get your Certificate of Eligibility and move your pre-approval forward today.”

PART V: BIX BUILD CHECKLIST & RESOURCES

12. CLARITY ENGINE BUILD CHECKLIST FOR BIX

12.1 Gear 1 — Core Calculation Engine (Sprint 1 Priority)

12.2 Gear 2 — Scenario Engine

12.3 Gear 3 — Borrower Intelligence Ledger

13. PRIMARY REFERENCE RESOURCES

13.1 Required Technical References

13.2 Data Partner Integrations (API Contracts to Build)

Partner Data Type Engine Use Case Attom Data Property records, AVM, ownership history LTV validation, equity monitoring, refinance triggers Clear Capital AVM, BPO, property condition Appraisal support, value confidence scoring Revaluate Purchase propensity scoring

7 D’s life event detection, pipeline prioritization

Credit Bureaus (3B) Full credit report + FICO Score classification, debt inventory, rapid rescore Fannie Mae DU / Freddie LP Automated underwriting engines AUS approval response, max DTI, risk classification Optimal Blue / Polly Real-time lender pricing Rate comparison, buydown pricing, product availability

13.3 Fair Lending Guardrails (Non-Negotiable)

Every calculation, scenario, and recommendation the Clarity Engine produces must pass through the fair lending compliance layer:

ARCHITECTURAL IMPERATIVE FOR BIX

The fair lending compliance layer is not optional and not an afterthought. It must be baked into the engine at the calculation level — every scenario must be tested for disparate impact before it surfaces to the borrower. This is both a legal requirement and a core brand promise of Purlend.

THE CLARITY ENGINE: CORE TRUTH

A mortgage is not a product. It is a math problem with a human life at the center. The Clarity Engine’s job is to solve that problem completely — surfacing every possible path, eliminating every unnecessary obstacle, and arriving at one outcome: the borrower gets the home.

Document Version: 1.0 | Prepared for: BIX Technology Corp Development Team | Classification: CONFIDENTIAL — PreFi, Inc. / Purpose Technology, Inc. (d/b/a Purlend)