Análisis profundo
Desglose paso a paso
Paso 1. Scorecard Architecture and Factor Weighting
FICO's generic scoring models partition consumer credit files into distinct scorecards based on profile characteristics such as the presence of derogatory information, the age of the oldest tradeline, and whether the file is thin or seasoned. Each scorecard contains its own regression coefficients, which means the relative weight of each factor shifts depending on which scorecard a consumer is assigned to. The commonly cited 35-30-15-10-10 breakdown is an approximation of average weights across the full population, not a fixed formula applied identically to every file.
Within each scorecard, individual predictor variables are binned into attribute ranges. For example, the payment history dimension does not simply count late payments; it evaluates recency, frequency, severity (30-day vs. 90-day vs. charge-off), and the type of account on which the delinquency occurred. A single 30-day late from eight years ago on a retail card and a 90-day late from last quarter on a mortgage occupy entirely different attribute bins and produce different point contributions even though both fall under the payment history umbrella.
VantageScore 4.0 uses a machine-learning-derived model that also segments consumers but publishes a different weight hierarchy: payment history 41%, depth of credit 20%, utilization 20%, balances 6%, recent credit 11%, and available credit 2%. Comparing FICO and VantageScore weights reveals where the models diverge in predictive emphasis, explaining why the same consumer can sit in different risk tiers depending on which model a lender pulls.
- FICO models use between 10 and 15 scorecards per bureau-specific version, each with unique coefficient sets
- The 35% payment history weight is a population average; thin-file scorecards may weight utilization more heavily
- VantageScore 4.0 collapses the five FICO factors into six categories with different percentage allocations
- Scorecard assignment is deterministic: the same file always lands on the same scorecard for a given model version
- Factor weights published by FICO are population-level averages, not individual-level guarantees
Paso 2. Payment History: Binary Derogatory Flags and Severity Gradients
Payment history is the strongest predictor of future default in both FICO and VantageScore models. The models construct a multi-dimensional delinquency profile: number of accounts with delinquencies, worst delinquency status ever reported, recency of the most recent delinquency, and the proportion of accounts that have ever been delinquent relative to total tradelines.
Severity buckets operate on a graduated scale. A 30-day late triggers the lowest derogatory flag. A 60-day late carries a meaningfully larger penalty because consumers reaching 60 days past due are significantly more likely to progress to 90+ days. Charge-offs, collections, bankruptcies, and foreclosures occupy the most severe bins. These events also trigger scorecard reassignment, moving the consumer from a clean-file scorecard to a derogatory scorecard where the entire coefficient structure changes.
Recency decay is critical: a 90-day late from seven years ago contributes far fewer negative points than the same event from six months ago. The models apply a time-decay function that diminishes the predictive power of older derogatory events. Most damage occurs in the first 12-24 months, with progressively smaller marginal effects as the event ages toward the seven-year removal window.
- A single 30-day late on an otherwise clean file can reduce a FICO 8 score by 60 to 110 points depending on starting score
- The severity gradient runs: 30-day, 60-day, 90-day, 120-day, charge-off, collection, bankruptcy, foreclosure
- Recency is weighted more heavily than frequency in most scorecard configurations
- Public record items like civil judgments were removed from credit reports in 2017 but bankruptcies remain
- Authorized user accounts report payment history, which is why piggybacking can inject positive or negative history
Paso 3. Utilization Ratio: Per-Card and Aggregate Dual Evaluation
Credit utilization is calculated at two levels in FICO models: the individual account level and the aggregate level across all revolving accounts. Both calculations feed separate predictor variables into the scorecard. A consumer carrying 90% on one card and 0% on three others scores differently than one carrying 22.5% across all four, because the per-card maximum triggers a separate high-utilization flag.
The models evaluate utilization as a snapshot of balances reported at statement closing, not in real time. Most issuers report the statement balance to bureaus, so a consumer who charges heavily but pays before the statement date can show near-zero utilization. This reporting lag makes utilization one of the most manipulable short-term levers in the scoring model.
FICO 10T and VantageScore 4.0 incorporate trended credit data tracking 24 months of balance history. Under trended data models, the algorithm distinguishes transactors who pay in full each month from revolvers who carry persistent balances. This distinction penalizes chronic revolvers even if their snapshot utilization on any given month is moderate.
- FICO models evaluate both per-card and aggregate revolving utilization as separate predictor variables
- Statement balance is the figure most commonly reported to bureaus, not the real-time balance
- Consumers with zero utilization on all revolving accounts score slightly lower than those showing 1-3% usage
- Trended data models penalize persistent revolving behavior across 24-month balance histories
- Installment loan utilization (remaining balance vs. original amount) is a separate variable with lower predictive weight
Paso 4. Credit Age: Oldest, Newest, and Average Tradeline Metrics
Length of credit history encompasses three distinct time-based variables: age of the oldest tradeline, age of the newest tradeline, and average age across all tradelines. Each feeds a separate predictor variable. The oldest account establishes maximum file depth. The newest account signals recent credit-seeking activity. The average age captures overall portfolio maturity.
Closing an old account does not immediately remove it from age calculations. Under FICO, closed accounts in good standing continue to contribute to average age for up to 10 years after closure. However, once the closed account falls off the report, the average age recalculates and may drop significantly if remaining accounts are young. This delayed effect explains why closing an old card sometimes shows no immediate score impact but causes a dip years later.
VantageScore treats account age differently by weighting the oldest account more heavily than the average. It also considers age of the oldest account by type (oldest revolving, oldest installment), meaning a consumer with a 15-year mortgage but only 2-year credit cards may score differently in VantageScore compared to FICO for the same file.
- FICO evaluates three separate age metrics: oldest tradeline, newest tradeline, and average age across all accounts
- Closed accounts in good standing remain on the report and contribute to age calculations for approximately 10 years
- Opening a new account reduces average age immediately but the newest-account variable recovers within 6-12 months
- VantageScore weights the oldest account age more heavily and evaluates age by account type independently
- Authorized user accounts contribute their full age to the user's credit file, inflating average age
Paso 5. Credit Mix: Account Type Diversity as a Risk Signal
Credit mix evaluates the diversity of account types on a consumer's file. The models categorize tradelines into revolving (credit cards, HELOCs), installment (auto loans, personal loans, student loans, mortgages), and open accounts (charge cards, utility accounts). Having experience with multiple types demonstrates the ability to manage different repayment structures.
Despite comprising only approximately 10% of FICO score weight on average, credit mix can be decisive for consumers in the 680-740 range who have optimized other factors. Adding an installment tradeline to a revolving-only file can produce a 15-30 point lift. The effect is non-linear: moving from one type to two produces a larger marginal gain than moving from three to four.
FICO's industry-specific models (Auto Score, Bankcard Score) evaluate credit mix with additional granularity. The auto score weights prior auto loan presence more heavily, while the bankcard score emphasizes revolving account management. A consumer's credit mix contribution can differ across model variants even with the same file data.
- The three primary account categories are revolving, installment, and open (charge cards, utility accounts)
- Moving from a single account type to two types produces the largest marginal credit mix benefit
- FICO Auto Score and FICO Bankcard Score weight account type experience differently than generic FICO models
- Mortgage tradelines carry a higher prestige signal in some scorecard configurations
- A consumer with five credit cards and no installment loans has weaker credit mix than one with two cards and an auto loan
Paso 6. New Credit: Inquiry Mechanics and Application Velocity Detection
The new credit dimension evaluates recent application activity through hard inquiry count and recently opened accounts. Hard inquiries signal credit-seeking behavior, which statistically correlates with elevated default risk. The number of accounts opened in the past 6-12 months is a separate variable from inquiry count.
FICO employs a deduplication window for rate-shopping. Multiple inquiries for mortgage, auto, or student loans within a 45-day window (30 days in older versions) count as a single inquiry. This deduplication does not apply to credit card inquiries. VantageScore uses a 14-day window that applies to all inquiry types including credit cards.
New account openings carry more weight than inquiries because they represent actual new credit exposure. A consumer can have five inquiries but only one new account if four applications were denied. The models treat these differently: inquiries signal intent to borrow, while new accounts signal real changes to credit capacity and repayment obligations.
- Hard inquiries remain on the report for 24 months but FICO only scores the impact for the first 12 months
- FICO's rate-shopping window is 45 days for mortgage, auto, and student loan inquiries in FICO 8+
- VantageScore deduplicates all inquiry types within a 14-day window, including credit card applications
- Each additional hard inquiry beyond the first typically reduces the score by 3-5 points with diminishing impact
- Newly opened accounts carry more scoring weight than inquiries because they represent actual credit exposure
