Static pool analysis

Static pool analysis tracks a fixed cohort of loans from origination through their entire life. This is the most important analysis for understanding actual performance because it eliminates the distortion of new originations constantly entering the pool.

Your current portfolio is a dynamic pool—loans enter as they originate and exit as they pay off or default. New originations dilute seasoned performance data, making it hard to see true loss patterns. Static pool analysis solves this problem.

The dynamic pool problem

Consider a portfolio growing at 20% monthly. Each month, new originations dominate the mix:

Month	Pool UPB	New originations	% Pool < 3 months old
January	$30M	$5M	45%
February	$34M	$6M	52%
March	$39M	$7M	56%

By March, over half the pool hasn’t had time to demonstrate performance. The delinquency rate looks low because most loans are too young to go delinquent. You might think performance is improving when you’re actually just growing faster.

Static pools isolate this effect by tracking cohorts of loans born at the same time, through their entire life.

Building static pool data

A static pool is a group of loans originated in the same period (typically a month or quarter), tracked from origination through payoff, default, or current status. Unlike a dynamic pool, a static pool never adds new loans.

Cohort definition

Define your cohorts by origination period:

Q1 2022 Cohort: All loans originated Jan-Mar 2022
  - Original count: 500 loans
  - Original UPB: $12.5M
  - Tracked monthly for 24+ months

Q2 2022 Cohort: All loans originated Apr-Jun 2022
  - Original count: 550 loans
  - Original UPB: $13.8M
  - Tracked monthly for 21+ months

Q3 2022 Cohort: All loans originated Jul-Sep 2022
  - Original count: 480 loans
  - Original UPB: $12.0M
  - Tracked monthly for 18+ months

Granularity choice: Monthly cohorts provide more precision but require more data management. Quarterly cohorts are standard for most analysis. Annual cohorts smooth noise but may miss within-year trends.

Monthly tracking metrics

For each cohort, track monthly from origination:

Month on book	Remaining count	Remaining UPB	Cum defaults	Cum losses	Cum prepays
0	500	$12.5M	$0	$0	$0
1	498	$12.4M	$0	$0	$20K
2	495	$12.3M	$0	$0	$45K
…	…	…	…	…	…
12	420	$9.8M	$250K	$180K	$1.2M
24	310	$6.5M	$680K	$510K	$3.4M

Remaining count/UPB: Loans still active in the cohort Cumulative defaults: Total dollar amount defaulted through this month Cumulative losses: Net losses after recoveries Cumulative prepayments: Principal paid early (beyond scheduled amortization)

Data requirements

To build static pool analysis, you need either:

Historical tape archive: Monthly loan tapes saved over time, allowing reconstruction of cohort performance
Loan-level performance history: Each loan’s full payment history from origination
Pre-built static pool reports: Some originators provide these directly

If you only have the current tape without history, you can’t build true static pool analysis. You can estimate historical performance from the current snapshot, but this is less accurate.

Cumulative loss curves

The cumulative net loss (CNL) curve is the primary static pool visualization. It shows total losses as a percentage of original pool balance over time since origination.

Building the curve

For each cohort at each month on book:

CNL = Cumulative Net Losses / Original Pool Balance

Month on book	Q1 2022 Losses	Q1 2022 CNL	Q2 2022 Losses	Q2 2022 CNL
6	$150K	1.2%	$193K	1.4%
12	$437K	3.5%	$538K	3.9%
18	$650K	5.2%	$786K	5.7%
24	$750K	6.0%	$897K	6.5%
30	$800K	6.4%	-	-

Reading loss curves

Curve shape by asset class:

Different asset classes have characteristic loss curve shapes:

Consumer unsecured: Front-loaded losses. Steep rise months 6-18, flattening by month 24-30. Losses occur early because there’s no collateral buffer
Auto loans: Similar front-loaded pattern. Peak loss velocity around months 12-18. Curve flattens as surviving loans demonstrate payment discipline
Mortgage: Back-loaded losses. Slower initial rise with foreclosure timelines extending curves. May not flatten until month 60+
Equipment: Depends on collateral type. Titled equipment resembles auto; soft collateral resembles consumer

Terminal loss: Where the curve flattens tells you expected lifetime loss. If Q1 2022 flattens at 6.4% by month 30, that’s your loss assumption for similar pools with similar seasoning.

Curve velocity: The slope at each point. Steep slope = losses accelerating. Flattening slope = losses decelerating. A curve that re-steepens after flattening signals something changed (macro event, collection practice change).

Vintage comparison

The power of static pool analysis is comparing cohorts to identify trends. Plotting multiple cohorts on the same chart reveals:

Building a vintage comparison

Months on book	Q1 2022 CNL	Q2 2022 CNL	Q3 2022 CNL	Q4 2022 CNL
6	1.2%	1.4%	1.5%	1.8%
12	3.5%	3.9%	4.2%	4.8%
18	5.2%	5.7%	6.1%	-
24	6.0%	6.5%	-	-

At each age point, compare across vintages. The pattern here is clear: each successive vintage performs worse at the same age.

Interpreting vintage trends

Worsening vintages (later cohorts perform worse):

Credit loosening in underwriting
Macro environment deterioration
Origination volume pressure compromising quality
Product drift to riskier segments

Improving vintages (later cohorts perform better):

Credit tightening
Macro improvement
Process improvements in underwriting
Portfolio mix shift to better segments

Stable vintages (similar performance across cohorts):

Consistent underwriting discipline
Stable macro environment
Mature, controlled origination process

Quantifying vintage differences

The Q4 2022 cohort is running 80 bps higher than Q1 2022 at month 12 (4.8% vs. 3.5%). At the same age, Q4 2022 is performing 37% worse than Q1 2022.

This finding should trigger investigation:

Did underwriting criteria change between Q1 and Q4?
Did the macro environment change?
Was there channel or product mix shift?
Is the sample size large enough to be statistically meaningful?

Vintage analysis catches deterioration before it appears in current pool metrics. Current pool metrics blend all vintages together; vintage analysis isolates each one.

Seasoning adjustments

Newer vintages haven’t experienced their full loss curve. A 6-month-old pool with 1.5% CNL isn’t comparable to a 24-month-old pool with 6% CNL. You need seasoning adjustments to project where newer vintages will end up.

The loss development method

Use mature vintage curves to project newer vintage terminal losses.

Step 1: Determine the loss emergence pattern from mature vintages

From the Q1 2022 cohort (mature):

CNL at 6 months: 1.2%
CNL at 24 months (terminal): 6.0%
% of terminal losses emerged by month 6: 1.2% / 6.0% = 20%

Step 2: Apply emergence pattern to newer vintage

Q4 2022 cohort shows 1.8% CNL at 6 months.

If Q4 2022 follows the same emergence pattern (20% of losses by month 6):

Projected terminal CNL = 1.8% / 20% = 9.0%

This suggests Q4 2022 is on track for 50% higher terminal losses than Q1 2022.

Building a loss emergence table

Months on book	% of terminal loss emerged
3	8%
6	20%
9	35%
12	52%
18	78%
24	92%
30	98%

This table comes from averaging multiple mature cohorts. Apply it to project any vintage based on current age.

Example projection:

Cohort	Current age	Current CNL	% Emerged	Projected terminal
Q1 2023	12 mo	4.1%	52%	7.9%
Q2 2023	9 mo	2.9%	35%	8.3%
Q3 2023	6 mo	1.7%	20%	8.5%
Q4 2023	3 mo	0.8%	8%	10.0%

The projections show continued deterioration through 2023 vintages.

Limitations of seasoning adjustment

Assumes pattern stability: If the macro environment changes or collection practices shift, historical emergence patterns may not apply.

Sensitive to terminal loss estimate: If mature vintages haven’t truly flattened, your emergence percentages are wrong.

Sample size matters: Small cohorts produce noisy estimates. A 50-loan cohort showing 2% CNL at month 6 could easily be 1% or 3% with different loan selection.

Use seasoning adjustments for directional guidance, not precision estimates. A projected 8% terminal CNL might actually land anywhere from 6% to 10%.

Vintage mix analysis

Understanding how vintage composition affects current pool metrics helps you interpret snapshot data.

The vintage contribution framework

Your current pool is a weighted average of vintage performances. Breaking down contributions reveals dynamics:

Vintage	% of current pool	Current DQ rate	Contribution to pool DQ
Q1 2022	15%	8%	1.2%
Q2 2022	20%	7%	1.4%
Q3 2022	25%	6%	1.5%
Q4 2022	40%	4%	1.6%
Pool	100%	-	5.7%

The Q4 2022 vintage shows only 4% DQ—but it’s 6 months old. Q1 2022 shows 8% DQ at 18 months seasoning. The lower pool DQ rate is partly driven by the young vintage mix, not credit quality improvement.

Stress testing vintage assumptions

Scenarios for how newer vintages might perform:

Base case: Q4 2022 follows Q3 2022 pattern

Projected Q4 2022 DQ at 18 months: 6.5%
Pool DQ impact: +1.0% when Q4 2022 seasons

Adverse case: Q4 2022 follows Q1 2022 pattern (worst vintage)

Projected Q4 2022 DQ at 18 months: 8.0%
Pool DQ impact: +1.6% when Q4 2022 seasons

Severe case: Q4 2022 performs 20% worse than any prior vintage

Projected Q4 2022 DQ at 18 months: 9.6%
Pool DQ impact: +2.2% when Q4 2022 seasons

This frames the range of outcomes as the pool seasons.

Static pool for prepayments

The same analysis applies to prepayments (CPR curves by vintage).

Why prepayments matter for static pools

High prepayment rates:

Shorten WAL
Reduce total interest income
May indicate refinancing (rate-sensitive) or payoff (good credit exits)

Low prepayment rates:

Extend WAL
May indicate credit deterioration (can’t refinance)
Affect reinvestment assumptions in revolving facilities

Cumulative prepayment curves

Months on book	Q1 2022 CPP	Q2 2022 CPP	Q3 2022 CPP
6	5%	4%	3%
12	15%	13%	10%
18	28%	25%	-
24	38%	-	-

CPP = Cumulative prepayment percentage (prepayments / original balance)

Prepayments are accelerating over time (Q1 2022 at 28% vs. Q3 2022 at 10% at month 18). This could be:

Rate environment changes (refinancing cheaper)
Portfolio seasoning (better borrowers exit earlier)
Product changes

Presenting static pool analysis

Executive summary format

Static Pool Analysis: XYZ Consumer Portfolio

Analysis covers Q1 2022 through Q4 2023 originations. Key findings:

Terminal CNL for mature vintages (Q1-Q2 2022) stabilizing at 6.0-6.5%

2023 vintages tracking 20-30% higher than 2022 at equivalent ages

Loss emergence pattern front-loaded: 52% of losses by month 12

Projected terminal CNL for 2023 vintages: 7.5-8.5%

Recommendation: Adjust loss assumptions for 2023+ production from 6.0% to 8.0%. Monitor Q1 2024 vintage for continuation of adverse trend.

Visual presentation

Include:

CNL curve chart with multiple vintages overlaid
Vintage comparison table at fixed ages (6, 12, 18, 24 months)
Emergence table used for projections
Sensitivity analysis showing range of terminal loss projections

Common mistakes to avoid

Comparing different ages without adjustment: “Q4 2023 at 2% CNL is better than Q1 2022 at 6%” is meaningless—one is 6 months old, the other 24 months.

Over-projecting from small samples: A 100-loan cohort with 3 defaults isn’t statistically reliable for fine-grained projections.

Ignoring macro context: A 2020 vintage (COVID) may not be comparable to 2019 or 2021 vintages even with seasoning adjustments.

Using only terminal loss: The path matters. Two cohorts with 6% terminal loss but different timing have different NPV impact.

Key takeaways

Static pools isolate vintage performance. They eliminate the dilution effect of new originations that masks true performance in dynamic portfolios.
Loss curves reveal patterns. The shape, velocity, and terminal point of loss curves are asset-class specific and highly informative.
Vintage comparison catches deterioration early. If each successive vintage performs worse, something changed—underwriting, macro, or mix.
Seasoning adjustments enable comparison. A 6-month cohort isn’t comparable to a 24-month cohort without adjusting for emergence.
Use projections directionally, not precisely. Seasoning adjustments have inherent uncertainty. Frame results as ranges.
Prepayments matter too. The same static pool framework applies to CPR curves and prepayment behavior.