2.6 What’s the magic?

Two Swaps, Proprietary Data and AI

ChinaAI delivers up to 50x regulatory capital relief for affiliate banks while maintaining robust, autonomous, risk management. This is initially achieved through a “zero-risk weighted” SPV structure, which guarantees the full notional value of the bank credit referenced by the CDO and then issues a super senior tranche as $BRICS. But it is made sustainable through the use of dynamic error-minimising AI that selects loss-minimising CDO exposures in real-time, making use of proprietary bank data.

Fully funded synthetic tokenised CDO, structure & tranching

How It Works: Two Key Swaps

1. From Originator Bank to Underwriters: The originating bank transfers the credit risk to the SPV underwritten by NASASA-Old Mutual.
2. From Underwriters to Investors: The SPV then issues $BRICS tokens to investors as a super senior tranche backed by the fully collateralized underlying Repo facility.

These back-to-back swaps mitigate the risk for the Underwriters economically. They leave the originating bank as the protection buyer and the investors as the protection sellers (or more accurately, re-insurers of bank credit risk).

Regulatory Arbitrage and Capital Relief

1. Basel III Compliance: Under Basel III risk-weighted mitigation rules, the regulatory capital treatment of the originating bank depends on the risk weight of the SPV’s investments, defined by high grade sovereign notes in this case.
2. AI Portfolio Optimization: Originating banks use the unlocked regulatory capital (trapped inside its own coffers) to fund AI-vetted credit portfolios (e.g. global trade receivables), where Basel capital charges are as much as 200x higher than warranted by expected losses. Log's of the individual company's referenced (and their predicted liabilities) may be found here.

Risk Management: How Stakeholders Benefit

1. Originating Banks (Nedbank, FirstRand, Barclays, etc.):

   1. Retain a 1%-5% equity tranche to maintain “skin in the game,” addressing potential moral hazard, even while offloading significant risk.

2. Underwriters (NASASA-Old Mutual)

   1. Take on the full notional risk of the residual portfolio (95%-99%) for the contract’s duration.

   2. Leverage ChinaAI machine-learning and statistical tools (e.g., gradient-boosting, T-Copula) to mitigate adverse selection and monitor portfolio risks. Find log's on CDO token value, following 10 000 Monte Carlo estimations of tail risk here. We elaborate on the computational methods used further below.
   3. Earn complaint governance rights over bank credit-risk mitigation tools at scale.

3. $BRICS Investors

   1. Provide as little as 2% credit protection in total, effectively re-insuring Underwriters through standardized tokens that cover the expected and unexpected losses.

   2. Earn the entire waterfall of monthly CDS premiums, plus risk-free interest — the benefits of which are delivered via airdrop, token buybacks or burning, Section 3,3 elaborates.

   3. Gain fractional access to investment grade BRICS credit assets, hitherto costly and illiquid for any one individual investor; or simply not possible without poolimg and risk-tranching.

Why Does This Arbitrage Opportunity Exist?

In short: mispriced capital charges. Basel’s broad categorization of assets fails to reflect the true risk of specific clusters, creating inefficiencies that ChinaAI exploits. For instance:

• Expected losses for trade receivables are as low as 0.01%, yet the Basel minimum charge is 2%.

• $BRICS aligns predicted pricing with actual risk while minimizing default probabilities — producing a win-win for both banks and investors. Naturally, this requires ChinaAI utilise more sophisticated risk estimation methods than is typically practised by tradtional banks.

Optimization Through Advanced Modeling and Proprietary Data

To enable sustainable, long-term, profitability, ChinaAI employs econometric and machine learning models to select bespoke portfolios with minimal default probabilities. Techniques include:

• XGBoost, Random Forest, Long short-term memory for estimates of Default Probability per Obligor
• Autoregressive models (AR, ARIMA, GARCH) for forecasts of Default Probability

• Logit/Poisson Maximum Likelihood Estimators to validate PD

• Generalized Method of Moments for Risk and Characteristic Scoring

• Levy and T-Copulas for joint risk scoring for the CDO
• Monte Carol simulation for estimates of loss distribution

These models ensure that portfolios are optimized for risk-adjusted returns, addressing inefficiencies in Basel capital requirements and providing a sustainable structure for both protection buyers and token holders. Moreover, the models are self-correcting and leverage proprietary bank data to dynamically adjust asset exposure decisions in real-time.

The result? $BRICS harmonizes regulatory compliance, risk mitigation, and financial inclusion, transforming how bank credit risk is dynamically managed and distributed globally.