Can Internal Models Alone Be Used to Determine the Close out Amount?

Concept

The question of whether an internal model, in isolation, can determine a final close-out amount touches upon the most critical junctures of financial risk management. It probes the line between a firm’s proprietary view of the world and the objective, often unforgiving, reality of the market. The answer, from a systems perspective, is a definitive and resounding no. An internal model is a vital organ within the risk infrastructure, a sophisticated engine for generating a valuation.

However, it is not, and can never be, the entire system. The process of determining a close-out amount is not a simple calculation; it is a legally and commercially rigorous procedure designed to function under the duress of a counterparty default, an event that by its nature invalidates many of the assumptions upon which tranquil market models are built.

At its heart, a close-out amount represents the cost of replacing the economic equivalent of a terminated portfolio of transactions. This process is governed by contractual frameworks, most notably the ISDA Master Agreement, which acts as the operating system for the global derivatives market. Within this system, the guiding principle is not mathematical purity or model elegance.

The governing principle is “commercial reasonableness.” This standard mandates that the party determining the close-out amount (the “Determining Party”) must act in good faith and use procedures that are objectively defensible to produce a result that the broader market would recognize as fair. This immediately moves the exercise from the theoretical realm of a model to the practical realm of evidence.

The Inherent Subjectivity of Internal Valuation

An internal model, no matter how complex or well-calibrated, is an expression of a single institution’s perspective. It is built upon a specific set of assumptions, calibrated to historical data chosen by the institution, and designed to fit within its particular risk appetite and operational workflow. These models are indispensable for daily marking-to-market, for internal risk reporting, and for providing a baseline valuation. Yet, in the context of a close-out, this inherent subjectivity becomes a critical vulnerability.

A counterparty default is a stress event. Liquidity evaporates, bid-ask spreads widen dramatically, and correlations that held for years can break down in hours. The historical data underpinning an internal model may become irrelevant. The model’s output, which seemed precise the day before, can become a theoretical number untethered from any price at which a replacement trade could actually be executed.

Relying solely on such a number would be akin to navigating a storm using a map of a different ocean. It ignores the present, chaotic reality in favor of a past, orderly model.

A close-out valuation’s integrity is measured not by its model’s sophistication but by its verifiable connection to external market reality.

From Calculation to Justification

The ISDA framework implicitly recognizes this limitation. It allows for the use of internal information, but only as one of several potential inputs in a hierarchy of valuation sources. The framework is designed to compel the Determining Party to look outside its own four walls first. The primary sources of valuation are observable, external data points:

• Third-Party Quotations ▴ The most robust evidence is firm or indicative quotes for replacement transactions from active dealers in the relevant market. This represents a live, executable, or near-executable price from a disinterested party.
• Relevant Market Data ▴ If direct quotes are unavailable, the next best source is other market data. This could include the prices of similar instruments, data from information vendors, or prices from brokered markets. This data provides a tangible, albeit indirect, reference point.
• Internal Model Information ▴ Only when these external sources are not readily available or would produce a result that is not commercially reasonable can a firm turn to its internal models as the primary basis. Even then, the model’s output is not the final answer. It is a piece of evidence that must be used within a “commercially reasonable procedure.”

This structure transforms the challenge from a simple calculation to a process of justification. The Determining Party must be able to construct a defensible narrative, supported by a clear audit trail, demonstrating that it followed a reasonable process to arrive at its final number. The internal model is a part of this narrative, but it cannot be the entire story. The true system for determining a close-out amount is a procedural one, a system of evidence gathering and reasoned judgment, where the internal model serves as a powerful tool but never as the ultimate arbiter.

Strategy

The strategic framework for determining a close-out amount is fundamentally a system of risk mitigation, not just for market risk, but for legal and operational risk as well. The core strategy is to construct a valuation that is not only economically sound but also contractually resilient and defensible under scrutiny. This requires a shift in mindset from seeking a single “correct” price to engineering a “reasonable” process. The governing document, the 2002 ISDA Master Agreement, provides the blueprint for this strategy, establishing a clear, albeit flexible, hierarchy of valuation inputs.

A sole reliance on internal models is a strategically flawed approach because it prioritizes a firm’s subjective viewpoint over the objective standard required by the contract and, potentially, by a court of law. The Lehman Brothers default in 2008 provided a crucible for these principles, with numerous legal battles fought over the reasonableness of close-out calculations. The key takeaway from that era was that process and evidence trump theoretical precision. A firm that could demonstrate it had polled dealers, sought market data, and then used its internal models to supplement or interpret that data was on far stronger ground than a firm that simply presented the output of its proprietary “black box.”

The Hierarchy of Valuation Inputs

The ISDA framework does not prescribe a rigid, unyielding formula. Instead, it outlines a cascade of preferred information sources. This hierarchy forms the core of a robust close-out strategy. The Determining Party’s primary goal is to obtain the most objective, market-facing data possible under the prevailing circumstances.

1. Tier 1 ▴ Direct Market Evidence (Quotations). The gold standard is obtaining firm or indicative quotations for a replacement transaction from one or more third-party dealers. This is the most direct evidence of the cost to replace the economic substance of the terminated trades. A strategy built on this tier involves maintaining strong relationships with a panel of dealers and having a clear, documented process for soliciting these quotes immediately following a termination event.
2. Tier 2 ▴ Indirect Market Evidence (Observable Data). In many stress scenarios, especially for less liquid products, obtaining multiple firm quotes is impossible. The market may seize up. In this case, the strategy shifts to gathering other relevant market data. This could include the prices of more liquid, correlated instruments, data from consensus pricing services, or matrix pricing based on observable inputs (e.g. using the price of a 7-year swap to help value a 6.5-year swap). The strategic imperative is to anchor the valuation in objective, external data points, even if they are imperfect proxies.
3. Tier 3 ▴ Internal Substantiation (Model-Based Valuation). The internal model finds its proper strategic role at this level. When external data is scarce, unreliable, or nonexistent, the internal model becomes the primary tool for generating a valuation. However, its use must still be part of a “commercially reasonable procedure.” This means the model itself must be robust, well-documented, and regularly used for valuing similar transactions in the normal course of business. The strategy here is one of preparedness ▴ ensuring that internal models are independently validated, their limitations are understood, and their outputs can be explained and justified. The model is used not as an oracle, but as a sophisticated tool for estimating a value that cannot be directly observed.

Model Risk as a Strategic Consideration

A sound strategy must explicitly account for the inherent limitations and risks of the models themselves. These risks become magnified during the very stress events that trigger close-outs.

Relying on an internal model alone for a close-out amount is substituting a verifiable market process for a subjective internal opinion at the moment of greatest dispute.

The European Central Bank’s guide on internal models, while focused on capital, highlights several relevant weaknesses that inform a close-out strategy. Key model risks include:

• Distributional Assumptions ▴ Models often assume that market returns follow certain statistical distributions (e.g. normal distribution). During a crisis, “tail events” that the model considered astronomically unlikely can occur, rendering its output invalid.
• Correlation Breakdowns ▴ A model may be calibrated on historical data showing stable relationships between different assets. In a panic, these correlations can converge to 1 (everything goes down) or break down entirely, invalidating hedges and diversification assumptions built into the model.
• Data Gaps and Illiquidity ▴ Models are only as good as the data they are fed. In a crisis, data feeds can become unreliable, and the absence of trading makes it impossible to mark positions to a true market price. A model may continue to produce a price, but it becomes a phantom price with no basis in reality.

The following table outlines the strategic difference between a model-only approach and the contractually sound, multi-source approach.

Ultimately, the strategy is to build a valuation file that can be presented to the defaulted counterparty, and if necessary a judge or arbitrator, that tells a clear and compelling story. The story is not “this is what our model said,” but rather, “we followed a commercially reasonable process as prescribed by our contract; we sought external prices, we gathered market data, and where necessary, we used our validated internal models to create a valuation that reflects the economic reality of the market at that time.” This approach transforms the internal model from a potential liability into a powerful, integrated component of a resilient and defensible risk management system.

Execution

The execution of a close-out valuation is a high-stakes, time-sensitive procedure that demands a fusion of quantitative analysis, operational discipline, and legal precision. It is the practical application of the strategy, translating the principle of “commercial reasonableness” into a sequence of concrete, auditable actions. A firm’s ability to execute this process effectively is a direct reflection of the robustness of its entire risk management infrastructure. The internal model is a powerful instrument in this process, but its successful deployment depends entirely on the operational playbook that surrounds it.

The Operational Playbook for a Close-Out Event

Upon the occurrence of an Early Termination Date, the Determining Party must initiate a pre-defined, documented procedure. This playbook ensures that all actions are methodical, consistent, and aimed at creating a defensible final amount.

1. Event Verification and Notification ▴ The first step is the formal verification of the Termination Event (e.g. bankruptcy filing, failure to pay). This triggers the immediate assembly of a dedicated team, typically including personnel from risk management, legal, trading, and operations. A formal notice is sent to the counterparty declaring an Early Termination Date.
2. Portfolio Reconciliation ▴ The team must immediately confirm the exact portfolio of transactions to be terminated. Any discrepancies with the counterparty’s records must be identified. The system of record must be frozen to provide a definitive list of all trades and their contractual terms.
3. Initiation of Market Sounding (Tier 1) ▴ The operational team immediately contacts a pre-approved list of market makers and brokers to request quotations for a replacement portfolio. All communication ▴ emails, recorded phone lines, chat logs ▴ must be meticulously logged. The request should be for either a full replacement of the portfolio or for key representative transactions.
4. Parallel Market Data Aggregation (Tier 2) ▴ Simultaneously, the risk team begins pulling all relevant, observable market data from sources like Bloomberg, Refinitiv, and consensus data providers (e.g. Markit). This includes closing prices, volatility surfaces, credit spreads, and any other inputs that could be used to value the transactions or their components.
5. Internal Model Valuation (Tier 3) ▴ The quantitative team runs the firm’s approved internal models on the frozen portfolio. This provides an initial, internally consistent mark-to-market value. This value serves as a benchmark against which external data can be compared.
6. The Valuation Synthesis ▴ This is the most critical step. The team synthesizes all gathered information. If credible dealer quotes are received, they will typically form the primary basis for the close-out amount. If quotes are sparse or wide, the team will use the observable market data to adjust or corroborate the internal model’s output. For example, the model might produce a price for an illiquid bond, but the team will apply a liquidity adjustment based on the observed bid-ask spread of a more liquid, similar bond. Every judgment, every adjustment, must be documented with a clear rationale.
7. Calculation of the Final Close-Out Amount ▴ The synthesized valuations for each transaction are aggregated. The cost of terminating or re-establishing hedges is added, as permitted by the ISDA Master Agreement. Unpaid Amounts due prior to the termination date are calculated separately and are not part of the Close-out Amount itself, but will be included in the final payment calculation.
8. Preparation of the Calculation Statement ▴ The legal and risk teams prepare a formal Calculation Statement to be delivered to the counterparty. This statement must show the calculations in “reasonable detail,” including the quotations, market data, and internal information used. This document is the culmination of the process and the primary piece of evidence in any potential dispute.

Quantitative Modeling and Data Analysis

The synthesis of data requires a quantitative framework that can blend model outputs with messy, real-world market information. The following table illustrates a simplified valuation for a single interest rate swap, showing the divergence between a pure model price and a final, adjusted close-out value.

Predictive Scenario Analysis a Case Study in Execution

Consider the default of a hypothetical hedge fund, “Momentum Capital,” during a sharp market downturn. A mid-sized bank, “StableBank,” is the Determining Party for a portfolio of exotic equity options and interest rate swaps. StableBank’s execution of its close-out obligations provides a clear illustration of the principles in action.

On Day 1, following Momentum Capital’s failure to meet a margin call and subsequent notice of default, StableBank’s risk committee convenes. The operational playbook is activated. The portfolio is confirmed ▴ five bespoke options on a basket of tech stocks and three standard interest rate swaps. The internal models at StableBank, which are state-of-the-art and have been rigorously backtested, value the entire portfolio at a net positive of $45 million in StableBank’s favor.

The equity derivatives team immediately contacts three specialist dealers for quotes on the exotic options. The response is telling. Dealer 1 refuses to quote, citing unprecedented volatility. Dealer 2 provides a quote so wide it is effectively meaningless, a “we will trade with you but at a punitive price” signal.

Dealer 3 offers an indicative bid for the portfolio at $20 million, less than half the model-driven value. For the standard interest rate swaps, however, the rates team secures three reasonably tight indicative quotes from major dealers, with an average valuation of $10 million, very close to StableBank’s internal model value of $10.5 million.

Here, the execution process faces a critical juncture. A model-only approach would demand a close-out amount near $45 million. But the market evidence, sparse as it is, points to a vastly different number. The StableBank team documents the refusal to quote from Dealer 1 and the exceptionally wide quote from Dealer 2 as evidence of market illiquidity.

They take the $20 million bid from Dealer 3 as the most credible, albeit painful, piece of external data for the options portfolio. They decide to use this as the primary basis for the options close-out. For the swaps, they average the three dealer quotes to arrive at a value of $10 million. They then add the documented costs of unwinding the delta-hedges associated with the options, which amounts to $1.2 million. The total Close-Out Amount they calculate is $20 million (options) + $10 million (swaps) – $1.2 million (hedge costs) = $28.8 million.

In the Calculation Statement sent to Momentum Capital’s administrators, StableBank does not simply state the final number. They attach a detailed appendix. It includes the time-stamped logs of their calls to all three dealers. It shows the wide quote from Dealer 2 and a note explaining why it was deemed not commercially reasonable to use.

It presents the firm bid from Dealer 3 as the core of the valuation. It includes the printouts of the swap quotes. It shows the calculation of the hedge unwind costs with transaction receipts. Finally, it includes the output from their internal model, but it is presented as a supplementary, internal benchmark used to assess the reasonableness of the market quotes, demonstrating the significant dislocation between theoretical value and executable price in a stressed market. By executing this rigorous, evidence-based process, StableBank has constructed a close-out amount that is not based on their model alone, but is grounded in the harsh reality of the market, making it overwhelmingly defensible against any future challenge.

System Integration and Technological Architecture

Effective execution is impossible without a supporting technological architecture. This system must ensure data integrity, create an unimpeachable audit trail, and facilitate rapid analysis.

• Centralized Risk Engine ▴ The core of the system is a risk engine capable of running the approved internal models on demand. This engine must be integrated with the firm’s trade capture system to ensure it is valuing the correct, reconciled portfolio.
• Data Aggregation Layer ▴ This layer must be able to pull in and time-stamp data from multiple external sources (e.g. Bloomberg, Reuters, Markit) and internal sources (e.g. communication logs from email and turret systems). This creates a single repository of all information used in the valuation process.
• Documentation and Reporting Module ▴ A system is needed to automatically generate the bulk of the Calculation Statement. It should pull the trade details, the synthesized values, the documented rationale for adjustments, and attach the supporting evidence from the data aggregation layer. This reduces the risk of manual error and ensures consistency.

This integrated system ensures that the execution of a close-out is not a frantic, manual scramble but a disciplined, systematic process. The internal model is a critical queryable database within this larger architecture, but the value is created by the system’s ability to synthesize the model’s output with external, objective evidence to produce a single, defensible close-out amount.

Reflection

The System beyond the Model

The exploration of a close-out valuation reveals a fundamental truth about financial risk management. The sophistication of any single component, such as an internal pricing model, is secondary to the integrity of the overall system in which it operates. A model produces a number; a system produces a defensible result. The framework of commercial reasonableness forces an institution to look beyond its own reflection in the polished surface of its models and engage with the abrasive, objective reality of the market.

This process holds a mirror to an institution’s own operational readiness. Does the firm possess the relationships and infrastructure to gather external data under duress? Is there a culture of meticulous documentation that can withstand legal scrutiny?

Is the legal and risk apparatus capable of making and defending difficult judgments in real time? The close-out event is the ultimate, non-hypothetical stress test, and it evaluates the entire organism, not just the strength of one muscle.

Therefore, the knowledge gained here is a component in a larger architecture of intelligence. It is an understanding that true resilience is not found in the theoretical elegance of a model, but in the design of a robust, evidence-gathering process that anticipates failure and prizes defensibility above all. The ultimate strategic potential lies not in building a better model, but in architecting a superior valuation system that acknowledges the model’s limits and embeds it within a framework of objective, market-facing evidence.