Intelligence Guide
What an SRT is, how we score the likelihood a bank will do one, and how to work the dashboard. Five short sections — jump around with the contents panel.
What is an SRT Transaction?
Why banks do them, in one screen
A Significant Risk Transfer (SRT) is a synthetic securitisation: a bank moves the credit risk of a loan portfolio to outside investors without selling the loans. The loans stay on its balance sheet — only the risk, and the capital it must hold against that risk, moves off. Banks do this to lift their CET1 ratio (capital ÷ risk-weighted assets) when rapid loan growth or rising risk weights squeeze it below their comfort zone.
How We Score SRT Likelihood
One question, two very different answers
Every bank gets a 0–100 SRT likelihood score. But “will this bank do a deal soon?” is a different question depending on whether it has issued before — so the model splits the universe into two populations, each scored on what actually predicts its next deal. This split lifted out-of-sample accuracy from AUC 0.73 to 0.82.
Banks with a prior SRT deal are scored on timing. Active programmes re-issue on a cadence (typically 6–18 months), so we measure how overdue a bank is versus its own historical cadence — its due-ness. More overdue → higher score, on a smooth curve that eases off near 99 (no ties at 100).
due-ness = months since last ÷ own cadence
score = 55 + 44·(1 − e−1.2·due-ness)
Banks with no prior deal are scored on capital-stress fundamentals: CET1 level & trend, leverage headroom, loan & RWA growth, portfolio concentration, and size. They sit in a lower band because a first deal is far rarer and harder to call than a repeat.
Shared modifiers
Novelty-weighted, up to +12. Fresh capital-relief intent (provisions, capital-optimisation talk on the earnings call) counts more than data we already have.
A daily LLM sentiment read (−20…+20) of SRT-market news nudges scores up or down with market conditions.
Designated GSIBs (incl. custody banks like State Street) are tagged so they never read as “mid-size” — a label fix, not a score change.
Try It
See how each population is scored
Repeat issuers span 55–99 by due-ness alone — a bank at its typical cadence scores ~86, one 2× overdue ~96. Fundamentals are context, not the driver.
Key Terms
The vocabulary you need, in plain English
A synthetic securitisation where a bank transfers the credit risk of a loan portfolio to outside investors without selling the loans. The loans stay on the balance sheet; only the risk — and the capital requirement — moves off.
Why it matters for SRT: The instrument this platform sources. Capital relief is granted once the regulator confirms the investor's first-loss position is large enough to absorb the portfolio's expected and unexpected losses.
Common Equity Tier 1 capital ÷ Risk-Weighted Assets — the strictest measure of a bank's capital adequacy. Regulators require 4.5% minimum; banks target 12%+ to keep a buffer above stress-test thresholds.
Why it matters for SRT: The core SRT trigger. SRT reduces RWA (the denominator), lifting CET1 without raising new equity — so a low or declining CET1 is a capital-stress signal in the first-time-issuer model.
Total assets scaled by risk weight — a $1B mortgage (35% weight) counts as $350M RWA; a $1B corporate loan (100%) counts as $1B. Higher-risk books carry more RWA per dollar of assets.
Why it matters for SRT: An SRT transfers the credit risk of a reference portfolio to an investor, cutting the RWA the bank must hold capital against. Rapid RWA growth is itself a capital-strain signal.
A repeat issuer's cadence is the typical gap between its SRT deals (the median of its historical inter-deal intervals). Due-ness = months since its last deal ÷ its cadence. 1.0 means it is exactly at its usual re-issue point; above 1.0 it is overdue.
Why it matters for SRT: The engine of the repeat-issuer score. Active programmes re-issue on a schedule, so how overdue a bank is versus its OWN cadence is the strongest validated predictor of the next deal.
A validation method that scores each bank using only data that existed at the time, then checks whether it actually issued in the following quarters. Confidence intervals are bootstrapped by bank. It avoids "circular" validation where the answer leaks into the score.
Why it matters for SRT: How this model was validated: out-of-sample AUC ≈ 0.82 and a 2.8× lift in the top decile — an honest measure of predictive accuracy, not a curve-fit.
Post-2008 capital rules. Basel III set the 4.5% CET1 minimum, a conservation buffer, and a leverage ratio. Basel IV (finalised) adds an "output floor" that raises RWA for banks using internal models, phasing in through 2025–2030.
Why it matters for SRT: Basel IV's output floor is a major European SRT catalyst: as it phases in, reported RWA rises, CET1 falls, and fresh SRT demand appears.
Tier 1 capital ÷ total exposure — a non-risk-weighted backstop to CET1. Basel III requires ≥3%; US GSIBs face a 5% enhanced SLR.
Why it matters for SRT: A bank near its leverage floor has strong incentive to shrink RWA via SRT. Low leverage is one of the capital-pressure signals in the first-time-issuer band.
The junior slice of an SRT structure that absorbs the first losses on the reference portfolio, typically 5–15% of notional. The investor here takes the most risk and earns the highest return.
Why it matters for SRT: Sizing the first-loss piece is what makes the SRT test pass — it must be large enough to absorb expected and unexpected losses for the regulator to grant capital relief.
Globally Systemically Important Bank — a bank on the FSB's designated list (JPMorgan, State Street, BNP Paribas, etc.), subject to extra capital surcharges. Custody GSIBs like State Street and BNY Mellon carry unusually low RWA relative to their true scale.
Why it matters for SRT: The size proxy (RWA-based) can misread custody GSIBs as "mid-size". The model tags designated GSIBs explicitly so they read correctly, without changing the score.
The regulatory disclosures we ingest. US banks file quarterly FDIC Call Reports (FFIEC 031/041); European banks publish Basel Pillar 3 reports (semi-annual/quarterly) with CET1, RWA, and portfolio breakdowns.
Why it matters for SRT: These are the raw inputs to every capital-stress signal. US data lags quarter-end by ~1–2 months; European data updates as filings publish.
Using the Tool
Dashboard → shortlist → memo → outreach
Scan the dashboard
All ~66 banks ranked by SRT likelihood. Repeat issuers that are overdue versus their own cadence float to the top; HIGH means score ≥ 65.
The top of the list is dominated by overdue active programmes — the highest-conviction near-term candidates.
Open a bank profile
Click a bank to see 8+ quarters of CET1 and RWA history plus the Score Breakdown. For repeat issuers it shows a "Re-issuance timing" panel (months since last deal, cadence, % through cycle); for first-timers it shows the fundamentals.
A repeat issuer well past 100% of its cycle, or a first-timer with a declining CET1 trend, is the strongest confirmation.
Generate the AI memo
On the profile, click "Generate Analysis". The model (gpt-oss-120b) calls four data tools — bank metrics, deal comps, regulatory context, and the scoring model — then streams a memo covering thesis, likely asset class, deal size, timing, and risks. It cites the exact same score shown on the page.
Takes 10–30s and isn't saved — copy it before leaving. "Regenerate" for a fresh pass.
Act on it
Use the memo's asset class to target your pitch, its deal-size range to size investor capacity, and its timing to schedule outreach. Cross-check the Deals page for the bank's prior structures.
Prior deals in the same asset class are the strongest confirmation of what a repeat issuer will bring next.
Ready to start sourcing?
Open Dashboard