Site Logo

STPrism

How do Key Risk Indicators relate to Black Swan events

5 min read

In my attempt to use NotebookLM to comment on and analyse various inputs (reviewing the piece through the eyes of author) I added a case study by Michael Mainelli to the collection of Nicholas Taleb. The ‘machine’ chose to incorporate the Mianelli ideas and produced this response - interesting in its own right but perhaps not the sharpest of critiques and re-framing in the ideas of the ‘voice’

Based on the provided sources, the relationship between Mainelli’s Key Risk Indicators (KRIs) and Taleb’s Black Swan events is fundamentally one of tension between predictive “scientific management” and the uncomputable nature of extreme events.

While Mainelli argues that KRIs restore “statistical correlation to its rightful place” to predict incidents[1], Taleb argues that relying on such statistical trends in complex systems makes one a “turkey”—blind to the imminent, unprecedented event that destroys the system[2].

Here is how the Mainelli case study can be interpreted through Taleb’s concepts:

1. The Turkey Problem: Predictive KRIs vs. Black Swans

Mainelli defines KRIs as metrics that “must contribute to the predictability of losses and incidents”[3]. For example, in the case study of a global commodities firm, high trading volume and high complexity were identified as indicators of future losses[4].

Taleb’s Critique:

• Retrospective Distortion: Mainelli’s approach relies on “Environmental Consistency Confidence”—the idea that the future will resemble the past[5]. Taleb argues that Black Swans are, by definition, outliers that do not appear in past samples. A KRI based on historical data behaves like the turkey who uses past safety to predict future safety, right up until Thanksgiving[2].

• The Green Lumber Fallacy: Mainelli suggests that better data leads to better prediction[6]. Taleb counters that in complex domains (Extremistan), more data often increases noise rather than signal, giving a false sense of confidence (the “Green Lumber Fallacy”)[7],[8].

2. Domains of Validity: Mediocristan vs. Extremistan

Mainelli acknowledges Taleb’s influence, noting that statistical techniques work well in “thin-tailed” distributions (Mediocristan) but are fragile in “fat-tailed” domains (Extremistan) where Black Swans occur[9].

• Mainelli’s View: KRIs are for “regular management, not extreme events”[10]. He argues that finding correlations helps manage the “known” and “known-unknown” risks[11].

• Taleb’s View: This distinction is fatal. You cannot separate “regular management” from extreme risk because in complex systems (like finance), the “extreme” event dominates the aggregate result[12]. Optimizing for the “regular” using KRIs often increases fragility to the “extreme” by removing redundancy[13].

3. Interpreting KRIs as Detectors of Fragility (Convexity)

The most constructive way to relate KRIs to Black Swans is to shift from predicting events to detecting fragility.

• Non-Linearity: Mainelli notes that KRIs are “not necessarily linear,” citing staff overtime which might be safe at moderate levels but dangerous at high levels[14].

• The Convexity Heuristic: Taleb provides the mathematical framework for this observation. Instead of using KRIs to predict when a loss will occur, they should be used to measure acceleration of harm (concavity)[15],[16].

    ◦ Interpretation: If a small increase in a KRI (e.g., trading volume or leverage) leads to a disproportionately large increase in potential error or loss, the system is fragile[17].    ◦ Action: The goal is not to predict the crash (Black Swan) but to identify that the system is in a state where a crash would be terminal, and then reduce exposure[16].

4. Correlation vs. Causality

Mainelli emphasizes using multi-variate statistics to find correlations between activities (inputs) and losses (outputs)[18].

• Catalysts as Causes: Taleb argues that focusing on these correlations confuses catalysts with causes. For example, high trading volume might be the catalyst for a breakdown, but the cause is the underlying fragility of the system[19].

• The Illusion of Control: Mainelli cites the use of “control structures” and “dashboards”[20]. Taleb argues that these often provide an “illusion of control,” leading to overconfidence. In the “Black Swan domain,” measurement errors are so high that “risk management” based on probability calculations often increases risk rather than reducing it[21],[22].

Summary Table: KRIs through the Lens of Antifragility

ConceptMainelli (Risk Case Study)Taleb (Antifragile/Fooled)
GoalPredict incidents/losses using correlations[5].Avoid ruin by reducing fragility; prediction is impossible[23].
MethodStatistical measurement (KRIs)[1].Heuristics (via negativa, barbell strategy)[24],[25].
Data UseMore data improves confidence[6].More data often increases noise and illusion of safety[26].
Risk ViewManage “known” risks; acknowledge limits for “unknowns”[11].Focus only on “unknowns” (Black Swans) because they cause ruin[27].
Use of KRIsIndicators of future incidents[3].Indicators of fragility (if they show acceleration of harm)[16].

Conclusion: To align Mainelli’s KRIs with Taleb’s philosophy, one must stop using them to forecast specific events (e.g., “losses will rise next month”) and start using them to stress-test the system for concavity (e.g., “if volume doubles, do errors multiply by ten?”). If the latter occurs, the system is fragile to Black Swans, regardless of what the historical probability suggests.

Graph View

Backlinks