The global OTC derivatives market reached $846 trillion in notional value in June 2025 — up 16% in one year, the largest annual increase since 2008. That is eight times the output of all human civilisation in a year. The number is simultaneously terrifying and misleading. This letter explains what it actually means, what actually happened in 2008, what has changed since, and — crucially — what solutions are being built. Because solutions do exist. And some of them are working.
Not investment advice. Data sourced from BIS OTC Derivatives Statistics June 2025, ISDA Key Trends H1 2025, BIS Triennial Survey 2025, and cited works. All figures current as of June 2026.
The Big Short is brilliant popular finance precisely because it translates abstract financial instruments into physical intuition. A Jenga tower you can visualise. A collateralised debt obligation structured from BBB-rated tranches of subprime mortgage-backed securities, then rated AAA by agencies paid by the banks that created them, sold to pension funds as safe assets — that requires three levels of abstraction that most people never complete. The film completed them. And it did something that mattered: it made people angry enough to ask why the instruments existed at all.
The answer is that derivatives exist because they are genuinely useful. This is the point the crisis mythology obscured. A derivative is simply a contract whose value derives from something else. A wheat farmer who sells a forward contract to lock in a price before harvest is using a derivative — and protecting his family from the risk of a price collapse. An airline that buys jet fuel options to hedge against a price spike is using a derivative — and protecting its customers from sudden fare increases. A pension fund that buys currency forwards to protect its foreign investments against exchange rate moves is using a derivative — and fulfilling its fiduciary duty to its members. The instrument is neutral. What matters is who uses it, for what purpose, and whether they hold sufficient capital to honour their side of the contract if it triggers.
In 2008, AIG did not hold sufficient capital. It had written approximately $440 billion in credit default swaps — insurance-like contracts guaranteeing mortgage-backed securities — without holding the reserves to pay claims if the guarantees were called. When the housing market collapsed, they were called. AIG needed $182 billion in government rescue funds to honour its obligations. The derivative was not the villain. The villain was the absence of any requirement that the party writing the derivative hold the capital to stand behind it.
The BIS reported $846 trillion in OTC derivatives outstanding in June 2025 — up 16% in one year, the largest annual increase since 2008. This number requires immediate context, because it is simultaneously the most important number in global finance and the most misunderstood.
These three numbers are all technically accurate. They tell three completely different stories. The $846 trillion figure is the one that generates headlines. The $21.8 trillion gross market value is the one that matters for understanding what would happen if every contract needed to be replaced simultaneously. The $3–4 trillion net credit exposure is the one that matters for understanding actual default risk. The gap between $846 trillion and $3–4 trillion — a 200× compression — is the result of two mechanisms: close-out netting (bilateral agreements that net opposite positions against each other) and collateralisation (posting cash or securities as security against contracts).
The honest answer to "are derivatives safer than in 2008?" is: yes, significantly — and differently riskier. The specific mechanisms that caused the 2008 crisis have been substantially addressed. New risks have been created in the process of addressing them. Understanding both is the analytical task.
Dodd-Frank (US, 2010) and EMIR (EU, 2012) mandated that standardised OTC derivatives — primarily interest rate swaps and credit default swaps — must be cleared through central counterparties (CCPs) rather than bilaterally between banks. Today, approximately 70% of interest rate derivatives and 69.5% of CDS are centrally cleared. Initial margin posted at major CCPs reached $430.4 billion at mid-year 2025, up from $364.4 billion a year earlier.
The mechanism: when trade A and trade B are both cleared through a CCP, the CCP becomes the counterparty to both. If one party defaults, the CCP absorbs the loss rather than it cascading through the bilateral network. CCPs hold margin, default funds, and their own capital as successive layers of protection. Close-out netting — the legal agreement allowing positions to be netted against each other before default settlement — reduced gross market value exposures by 86.4% at mid-year 2025.
The residual risk: CCPs are now too big to fail in ways that individual banks were in 2008. LCH (London Clearing House), CME Clearing, and DTCC between them clear tens of trillions in derivatives daily. If one failed, the consequences would dwarf Lehman Brothers. The solution moved counterparty risk from a distributed network to concentrated nodes — better in normal times, catastrophically different in failure mode.
Not every derivative can be standardised enough for central clearing. Bespoke, customised contracts between sophisticated parties remain bilateral. For these, regulators have mandated the Uncleared Margin Rules (UMR) — requiring both initial margin (upfront collateral against potential future exposure) and variation margin (daily cash settlement of mark-to-market moves) to be exchanged between counterparties.
The UK's September 2022 LDI crisis was the first major test of whether margin rules were calibrated correctly — and it failed. UK pension funds had entered into interest rate derivative positions to match their long-duration liabilities. When gilt yields spiked, variation margin calls required immediate cash. The pension funds didn't have it in liquid form. The Bank of England had to intervene with emergency gilt purchases to prevent a cascade of forced selling. The margin rules worked as designed — they required collateral. The gap was that the collateral management practices of the pension funds were not designed for the speed and scale of the margin calls. The lesson: technically correct rules plus inadequate liquidity planning equals crisis.
One of the most dangerous features of the 2008 derivatives market was opacity. Regulators did not know who held what positions. When Lehman Brothers filed for bankruptcy, it took weeks to untangle its derivatives book. Dodd-Frank and EMIR required all derivatives trades to be reported to centralised trade repositories — the DTCC's Global Trade Repository, ICE Trade Vault, and equivalents globally.
Today, regulators have visibility into derivatives markets that would have been unimaginable in 2007. The BIS publishes the statistics cited in this letter — a direct product of the expanded reporting that post-crisis reforms mandated. ISDA has developed the Digital Regulatory Reporting (DRR) framework using Common Domain Model (CDM) — a standardised digital representation of derivatives that makes machine-readable reporting possible. Gentek AI was selected to develop a traceability tool for the DRR in 2026. The direction of travel: from periodic aggregate reporting to near-real-time granular data that regulators can use to identify concentration risks before they become crises.
One reason the $846 trillion notional figure is so large is historical accumulation — derivatives are added to the book but the offsetting positions are not always cancelled. Portfolio compression is the process of replacing multiple offsetting contracts with fewer, equivalent contracts — reducing notional without changing the economic exposure. In H1 2025, non-market-facing compression trades amounted to $1.9 trillion — 24% of all IRD turnover, up from 14% in 2022.
TriOptima (owned by LSEG) pioneered multilateral compression and has compressed over $2,000 trillion in notional derivatives over its history — the single largest reduction in systemic risk that almost nobody has heard of. Capitalab, Quantile Technologies, and AcadiaSoft operate in adjacent spaces. The process: multiple dealers submit their portfolios, an algorithm finds the set of terminations and replacements that reduce gross notional while leaving each dealer's net position unchanged. What took 2 trillion in notional out of the system while changing nobody's actual economic exposure is perhaps the most elegant systemic risk reduction ever engineered — a pure mathematical solution to a structural problem.
The most significant unsolved problem in derivatives risk management is speed. AI-driven trading books now accumulate derivatives positions faster than human risk managers can monitor them. A major dealer's derivatives book can change by billions in seconds through automated market-making, algorithmic hedging, and high-frequency positioning. The risk frameworks designed to govern these books were built for human-speed trading.
The solution being built: AI risk monitoring systems that operate at the same speed as the trading activity they monitor. JPMorgan's LOXM system, Goldman Sachs's Marquee platform, and a generation of fintech infrastructure companies (Finteum, OpenGamma, Cassini Systems) are building the real-time risk analytics that can flag dangerous concentrations before they become crises. Cassini Systems specifically focuses on margin optimisation — using AI to determine the most capital-efficient way to clear and collateralise a derivatives portfolio across multiple CCPs simultaneously.
The unresolved tension: AI systems trained on historical data have not been tested against a truly novel market dislocation. The correlations that break in a real crisis are exactly the correlations the models use as inputs. The AI risk monitoring frontier is essential and insufficient simultaneously — it makes normal operations safer without guaranteeing it catches the genuinely unprecedented event.
Greenspan's "flaw" was the assumption that derivative dealers would manage their counterparty risk rationally — that self-interest would prevent any single institution from writing more contracts than it could honour. The assumption was theoretically coherent. It was empirically catastrophic. AIG's derivatives desk wrote contracts as if it were an infinite-capacity insurer of last resort, collecting premiums that were priced as if the underlying risk did not exist at systemic scale. The rest of AIG did not know. The regulators did not know. The counterparties did not fully know.
The post-2008 reforms address the Greenspan flaw directly: mandatory clearing forces standardised contracts through CCPs that hold capital against default. Mandatory margining forces counterparties to post collateral daily. Mandatory reporting forces transparency that was previously absent. The specific mechanism that caused 2008 — unregulated, uncollateralised, bilaterally-settled derivatives written by undercapitalised counterparties — has been substantially eliminated.
What has not been eliminated: the possibility of a systemic event triggered by mechanisms not yet identified. Every financial crisis in history has been caused by a different specific mechanism — but the same underlying dynamic: leverage, opacity, and the assumption that correlations observed in benign conditions will persist in stressed ones. The derivatives market of 2026 has less of the first two than 2006 did. It has not solved the third. The next crisis, if it comes, will not look like 2008. It will look like something we haven't named yet — which is precisely what makes it a crisis.
CCP concentration is the new AIG problem — more structured, less visible. The post-2008 reforms moved derivatives risk from a diffuse bilateral network to concentrated CCPs. LCH, CME Clearing, and Eurex together clear contracts whose notional value is a meaningful fraction of the $846 trillion total. If any CCP experienced a default waterfall failure — where initial margin, default fund, and CCP capital were all insufficient — the systemic consequence would be larger than Lehman. The reforms reduced the probability of this event. They did not eliminate it. And the concentration they created has made the consequence of failure larger than before.
The April 2025 data is a warning signal that hasn't been fully processed. The 16% jump in OTC derivatives notional in June 2025 — the largest since 2008 — occurred "against the backdrop of elevated uncertainty over trade, monetary policy outlooks, and geopolitical tensions." This is the BIS's careful language for: tariff war, confused central banks, and a fragmenting world order. Derivatives markets grow fastest when uncertainty grows fastest — because that is when hedging demand is highest. The $846 trillion is not a number that appeared randomly. It appeared because the world's largest companies, banks, and governments are more uncertain about the future than at any time since 2008. That uncertainty is itself a systemic risk signal.
The AI trading book problem has not been solved — it has been deferred. The speed of modern derivatives markets — driven by algorithmic market-making, AI-driven hedging, and high-frequency positioning — has created positions and correlations that accumulate faster than existing risk governance can monitor. The UK LDI crisis of 2022 was a slow-motion demonstration of what happens when margin calls arrive faster than liquidity management systems can respond. An equivalent event in AI-driven derivatives books would not be slow-motion. It would be over before a human could react.
Long-horizon thinking on capital, technology, and the forces shaping the next decade of wealth creation. Written from first principles. Not consensus. Not noise.