Daily Summary for 2026-11-04
daily
Transcript
Introduction to the show: ident: Quantitative Finance. The day in quantitative finance: models of markets, risk and prices.
Ruth: It's the fourth of November, twenty twenty-six, and this is the day's research.
Warren: 3 new papers came out today.
Ruth: I'm Ruth.
Warren: And I'm Warren, here with Iris.
Ruth: We'll take the day in one pass, then pull out the papers we're staying with.
Warren: Start us off with Efficient Monte Carlo Valuation of Corporate Bonds in Financial Networks.
The day's research: Ruth: We are tracking risk transmission from crypto service providers using an Austrian registry to track Bitcoin, Ether, USDC, and USDT activities.
Warren: That registry allows researchers to see roughly thirty billion dollars in USD intermediated globally rather than just domestically.
Iris: This measurement shows how different groups react during shocks like the Terra-Luna collapse or FTX bankruptcy.
Ruth: It reveals that stablecoins do not uniformly act as a safe haven, which contrasts with corporate bond valuation work.
Warren: That paper tackles systemic risk in financial networks by proposing Bi-Level Importance Sampling with Splitting to estimate bond prices efficiently.
Iris: The latter paper uses a fictitious system approach to decouple an institution's default condition from the rest of the network.
Ruth: That contrasts sharply with the Austrian study, which uses direct transaction-level measurement for risk transmission patterns.
Warren: There is research exploring implied interest rates in crypto markets using data from established derivatives markets.
Iris: This work constructs yield curves for Bitcoin and Ethereum by leveraging inverse options and futures contracts.
Ruth: The results show Bitcoin's implied rates remain close to zero, while ETH fluctuates around a two percent staking yield.
Warren: That suggests different underlying dynamics between the two assets, doesn't it?
Iris: Indeed, the Austrian study focuses on how providers interact globally and their behavior during shocks.
Ruth: The corporate bond paper proposes a simulation method to accurately price bonds by handling complex network effects causing cascading losses.
Warren: That method is computationally advanced, allowing for linear complexity in terms of network size.
Iris: The derivative-implied rates study reveals different yield patterns for Bitcoin and Ethereum based on those contracts.
Ruth: So we have a view on global intermediation risk, bond pricing simulation, and asset specific interest rate dynamics.
Lucky paper: 2602.12770: Ruth: So we're looking at the Efficient Monte Carlo Valuation of Corporate Bonds in Financial Networks paper, and it tackles how cascading losses propagate through interinstitutional exposures.
Warren: It notes that standard Monte Carlo methods fail to capture rare but critical default events because they can't handle higher-order network effects or scale well with network size.
Iris: The authors address this by proposing a Bi-Level Importance Sampling with Splitting method, which seems to be a novel way to approach the valuation problem.
Ruth: They introduce this fictitious system approach to decouple an institution's default condition from the rest of the network, which is supposed to make it computationally tractable.
Warren: That decoupling strategy is what allows them to reformulate the default condition into a one-dimensional threshold condition on external assets for a specific bank.
Iris: And they use this characterization to decompose the valuation problem into two nested sampling processes: exponential tilting for outer-level sampling and optimal importance sampling for inner-level sampling.
Ruth: The complexity of this BLISS method is stated as being at most linear in the number of banks, which is a big improvement over current state-of-the-art algorithms.
Warren: That linear complexity is significant because existing rare event simulation methods face exponential growth in computational time as the network size increases.
Iris: The theoretical analysis shows that BLISS achieves asymptotic optimality in both regimes: the large asset regime and the small volatility regime.
Ruth: They prove asymptotic optimality by showing that for a specific ratio of moments, like lim inf m to infinity log Ee
(Gamma A m)two: over log E
(Gamma zero m)two: , it equals one plus two A squared over one plus A squared.
Warren: That formula gives a concrete way to define the performance bound when assets are correlated in the large asset regime.
Iris: The supplementary numerical results confirm that BLISS performs significantly better than Conditional Importance Sampling and naive Monte Carlo methods in terms of both relative error and running time for their toy example.
Ruth: They also show that BLISS maintains constant running times regardless of network size, whereas Conditional Importance Sampling experiences exponential growth when the network expands.
Warren: That's a key difference, so the efficiency gain seems to be directly tied to this linear complexity scaling.
Iris: The paper points toward future research directions like extending BLISS to value more complex securities, such as credit derivatives or collateralized debt obligations.
Ruth: Incorporating dynamic network structures where interbank exposures evolve over time would enhance the realism of the model and enable term-structure pricing for corporate bonds.
Warren: That would make the valuation even more realistic when dealing with things that change constantly in a financial system.
Iris: And they also suggest adapting this methodology to stress testing and regulatory capital assessment for evaluating tail-risk scenarios across large portfolios efficiently.
Lucky paper: 2509.03964: Ruth: Let's talk about Where Are the Rates? Derivative-Implied Interest Rates in Cryptocurrency Markets.
Warren: So this paper uses data from highly developed cryptocurrency derivatives markets to construct yield curves, which is a big shift since fixed-rate lending and bonds are almost nonexistent now.
Iris: It starts by saying traditional finance needs bond prices across a range of maturities, but that's hard in new currencies.
Ruth: The paper then uses inverse options and inverse futures with fixed maturities to replicate the payoff of zero-coupon bonds.
Warren: They use call put parity to turn it into a synthetic zero-coupon bond price equation and then use an inverse futures contract to replicate the bond in the cryptocurrency.
Iris: They find that when computing annualized implied yield curves for Deribit USD and BTC based on BTC inverse option data, there's a downward-sloping structure in the USD rates during volatile periods.
Ruth: But they also found that the implied rates for Bitcoin remain close to zero across all maturities, which is consistent with Bitcoin not intrinsically bearing interest.
Warren: That contrasts with Ethereum, where the implied rates fluctuate around zero throughout the entire observation period despite its non-negligible staking yield of around two percent.
Iris: The study notes that the differences between USD rates inferred from BTC and ETH inverse option data are frequently positive, with an average spread of approximately one percent that decreases with the tenor.
Ruth: That paper, Where Are the Rates? Derivative-Implied Interest Rates in Cryptocurrency Markets, really shows how different underlying dynamics exist between those two assets when using these specific derivatives to infer rates.
Warren: It's interesting because it suggests that even with a staking yield of two percent on Ethereum, the model still sees its implied rates staying near zero compared to Bitcoin.
Iris: The paper uses Random Sample Consensus, or RANSAC, to remove outliers from the option data by fitting a regression model.
Ruth: So while they are using this sophisticated mathematical approach, they are still relying on historical option data and then applying that to construct these annualized implied yield curves for Deribit USD and BTC.
More episodes
- The day in conversation for 2026-11-19
- Daily Summary for 2026-11-19
- The day in conversation for 2026-11-18
- Daily Summary for 2026-11-18
- 2601.05274-On the use of case estimate and transactional payment data in neural networks for individual loss reserving
- 2009.00868-Loss-Given-Default Modeling by Post-Last Passage Time Process
- 2606.30070-Financial Resilience Evaluation: From Conditional Expectations to Dynamic Convex Risk Measures
- 2212.12687-Measuring price impact and information content of trades in a time-varying setting
- Daily Summary for 2026-11-17
- 2401.08323-Dynamic portfolio selection under generalized disappointment aversion