ML-QuantSubscribe

Quant Letter

October 2026, Week 5

The week in quantitative finance research: 90 papers from arXiv, SSRN and RePEc, screened and ranked, 24–30 Oct 2026.

Papers
90
New this week
90
With figures
52
High fanfare
4
Subscribe free
View as MarkdownOpen in ClaudeOpen in ChatGPTOpen in PerplexitySite index for AI (llms.txt)

From the editor

This week centers on AI agent fragility in financial markets, real-time benchmark design preventing memorization, and cryptographic signals predicting equities. AI Equity Research Without Future Data reveals forecast accuracy well below overall performance; Fragility Of Synchronized Agentic Wealth Management quantifies market destabilization from concentrated AI decisions; and Weekend Crypto Volatility Forecasts Markets demonstrates novel cross-asset predictability.

Top picks

If you read five papers this week, read these.

  1. arXiv · No. 101

    AI Equity Research Without Future Data

    Investor's Last Exam benchmarks AI agents on frozen financial databases, finding leading models score 68.1 overall yet only 52.5 on forecast accuracy.

    Fanfare 4ML & AI Methods
    Scatter plot showing forecast errors for Tesla across three AI models relative to zero-credit limits.
  2. SSRN · No. 102

    AI Monoculture Paradox in Markets

    Fanfare 4ML & AI Methods

    When traders adopt more precise but common AI signals, synchronized order flow can outweigh forecast gains, reducing collective profits despite higher individual precision, a paradox that taxes can help resolve.

  3. RePEc · No. 103

    150 Years of US Inflation Expectations

    Fanfare 4Econometrics & Forecasting
    Three time series comparing inflation, narrative intensity, and grain futures spreads from 1870-2020.
  4. RePEc · No. 204

    US Monetary Policy Spillovers in Commercial Real Estate

    Fanfare 4Macro-Finance & Rates
    CRE price responses to US QE shock within and outside the US across multiple variables
  5. arXiv · No. 205

    Fragility Of Synchronized Agentic Wealth Management

    Fanfare 3ML & AI Methods
    The agentic wealth-management layer

What's rising

Topics drawing unusually many papers this week.

  1. Alternative data

    3.8×

    1.8% of this week's 488 new papers, 3.7x its share over the previous 5 weeks (0.5%).

  2. Climate and ESG

    1.4×

    12.3% of this week's 488 new papers, 1.4x its share over the previous 5 weeks (8.6%).

    In this issue

  3. Graph neural networks

    2.2×

    1.8% of this week's 488 new papers, 2.2x its share over the previous 5 weeks (0.8%).

    In this issue

GitHub radar

What the quant community starred on GitHub this week, from a panel of 10,589 developers who follow financial machine learning (291 of the 859 we checked starred something). Counts only: we never publish who starred what.

Quant repos rising

  1. goldmansachs/gs-quant5 quants

    Python toolkit for quantitative finance, derivatives and risk management.

    ★ 13.8k+504 this weekPython

  2. Jakeschincariol/day-trader-skill4 quants

    Eleven Claude skills for day trading with backtesting and scanning.

    ★ 3,464+2,568 this weekPythonNew repo

  3. QuantMLResearch/quant-handbook4 quants

    Quantitative research handbook.

    ★ 162+163 this weekPythonNew repo

  4. simonro/Edge-GEX-Lab3 quants

    Local live map of option gamma exposure for indices and stocks.

    ★ 51+50 this weekPythonNew repo

  5. Jakeschincariol/founder-skill2 quants

    Eleven Claude skills testing business viability before launch.

    ★ 2,217+653 this weekPythonNew repo

  6. nautechsystems/nautilus_trader2 quants

    Production-grade Rust trading engine for algorithmic and crypto trading.

    ★ 31.1k+410 this weekRust

  7. gloom-sh/gloomberb2 quants

    Finance terminal for market data, portfolio tracking and stocks.

    ★ 2,577+65 this weekTypeScript

  8. Hitheshkaranth/OpenTerminalUI2 quants

    Trading terminal UI for market data, charting and backtesting.

    ★ 376+19 this weekTypeScript

What quants are playing with

  1. Zagitta/gpu-unlock11 quants

    Unlocks higher tensor throughput on GeForce GPUs for AI workloads.

    ★ 13.4k+12.6k this weekCNew repo

  2. togg53192-cmd/jailbreaks10 quants

    List of jailbreaks.

    ★ 10.1k+6,563 this week

  3. fabio-rovai/open-ontologies8 quants

    Plans, applies and rolls back changes to production ontologies with audit proofs.

    ★ 2,213+898 this weekRust

  4. JetBrains/thinkrail8 quants

    Lightweight IDE for AI coding agents with customizable skills.

    ★ 6,387+5,359 this weekTypeScript

  5. pollen-robotics/microduck8 quants

    Tiny biped duck robot.

    ★ 13.1k+2,596 this weekRust

  6. elayadesign/ai-design-skills8 quants

    No description available.

    ★ 12.7k+9,915 this week

We called it

The full track record →

  1. Now publishedFourier–Laplace transforms in polynomial Ornstein–Uhlenbeck volatility models

    Now published in Finance and Stochastics · featured 8 May 2024, 5 days after release; 8 citations

  2. 250 citationsHybridRAG: Integrating Knowledge Graphs and Vector Retrieval Augmented Generation for Efficient Information Extraction

    Passed 250 citations · featured 15 Aug 2024, 6 days after release

  3. 50 citationsSea Change in Software Development: Economic and Productivity Analysis of the AI-Powered Developer Lifecycle

    Passed 50 citations · featured 28 Jun 2023, 2 days after release

All papers

Filter by venue, topic or fanfare.

Venue
Topic
Fanfare
90 papers

arXiv

Quantitative finance and ML-for-finance preprints

30 papers · 29 figures

ML & AI Methods3

01

AI Equity Research Without Future Data

Investor's Last Exam benchmarks AI agents on frozen financial databases, finding leading models score 68.1 overall yet only 52.5 on forecast accuracy.

Fanfare 4ML & AI MethodsHuangfei Jiang et al.PDF

Scatter plot showing forecast errors for Tesla across three AI models relative to zero-credit limits.
Figure 9: Forecast errors despite correct supporting analysis. Absolute Tesla forecast errors divided by each metric’s zero-credit error limit. Every marker lies beyond…
02

Fragility Of Synchronized Agentic Wealth Management

An agent-based model shows conversational AI wealth managers concentrate decisions into about two independent views, making markets 1.47 to 1.90 times more fragile than robo-advice.

Fanfare 3ML & AI MethodsSaurabh KohliPDF

The agentic wealth-management layer
Figure 1: The agentic wealth-management layer. Clients converse with agents built on a few foundation models (shares measured by StatCounter Global Stats, 2026 ). Every…
11

Multi-Agent Factor Generation with Diversity Regularization

A multi-agent framework with a log-determinant penalty on agent correlations generates diversified alpha factors and improves information coefficient by 1.1 percent.

Fanfare 3ML & AI MethodsTiantian Wu and Peng YangPDF

Multi-agent factor generation architecture with diversity penalty regularization flow diagram
Figure 1: Each agent independently maps stock features 𝑋𝑡to a standardized factor vector; the outputs form the factor matrix 𝐹𝑡(𝜃), from which the diversity module compu…

LLMs & Text3

03

Knowledge-Graph-Enhanced Financial LLMs

FinLLM-KG injects SEC filing facts into sentiment analysis at inference time, improving portfolio performance through explainable company-level context without retraining.

Fanfare 3LLMs & TextGiorgos Iacovides et al.PDF

Cumulative returns of sentiment-based 35% long–short portfolios under three conditions: Baseline (article text only),
Figure 2: Cumulative returns of sentiment-based 35% long–short portfolios under three conditions: Baseline (article text only), Summary (article text with the latest 10-…
04

LiveSearch: A Live Benchmark For Financial Agents

A live benchmark stores financial questions as executable code so answers update daily against real data, preventing memorization and revealing that 70% of daily queries change their correct answer.

Fanfare 3LLMs & TextSinuo Wang et al.PDF

Sankey diagram showing how 109 questions change across evaluation dates, tracking periods, raw values, and gold answers.
Figure 4. From new periods to new answers. All 109 questions over September 14–16, 2026. Band widths count questions and trace the joint paths from reported observation…
28

Multimodal Learning for Long Financial Documents

A transformer framework integrates semantically enriched table embeddings with unstructured text via cross-attention, showing structured data dominates but text adds incremental signal.

Fanfare 2LLMs & TextXiaoyu HuPDF

Structured Table From Apple Inc
Figure 2: Structured Table From Apple Inc. 10K

Derivatives & Volatility4

05

Weekend Crypto Volatility Forecasts Markets

Weekend cryptocurrency volatility predicts Monday variance in equities, commodities and bonds 11-22% better post-2024, following US Bitcoin ETF approval in late 2023.

Fanfare 3Derivatives & VolatilityPedram Ebadollahyvahed and Nneka UmeorahPDF

S&P 500 Monday volatility forecasts: realized versus HAR models with weekend data
Figure 5 : Monday realised volatility for the S&P 500 against the HAR and HAR-plus-weekend forecasts over the last 40 Mondays of the sample. Both forecast series are one…
14

Deep Learning for American Option Pricing

Operator splitting scheme with physics-informed neural networks prices five-dimensional American options under transaction costs and stochastic liquidity with validated boundary accuracy.

Fanfare 2Derivatives & VolatilityDong Yan et al.PDF

American put option values across asset prices showing PINN and FDM agreement with transaction cost sensitivity.
Figure 3: American put values for different transaction-cost parameters \kappa_{TC} . The inset magnifies the neighborhood of the strike, S_{0}\in[99,101] , and displays…
17

Smooth Threshold GARCH For Volatility Asymmetry

A GARCH variant with an estimated threshold captures volatility responding asymmetrically to positive versus negative returns, allowing positive returns to have no volatility impact.

Fanfare 2Derivatives & VolatilityJonas Fiedler Frederiksen and Frederik VilandtPDF

NICs comparing volatility response of STGARCH versus T-GARCH, GARCH, GJR, and A-GARCH models
Figure 1: NICs of the T-GARCH (top-left), GARCH (top-right), GJR (bottom-left) and A-GARCH (bottom-right). The red curve indicates the NIC of the STGARCH. Here we have s…
23

Neural Stratification Factors For Derivative Pricing

The research establishes continuity results for conditional expectations in stratified Monte Carlo, enabling neural networks to construct admissible stratification factors for path-dependent claims.

Fanfare 2Derivatives & VolatilityAleksandar Arandjelovic and Patrick CheriditoPDF

Conditional probabilities of autocall and downside-protection events
Figure 2: Conditional probabilities of autocall and downside-protection events.

Crypto & DeFi2

06

DeFi Yield Trading Derivatives And Tail Risk

A large-scale Pendle dataset spanning 340 markets reveals stablecoin flight-to-quality dynamics, and a tail-risk loss function improves yield forecasting and enables profitable derivatives strategies.

Fanfare 3Crypto & DeFiHuaiyu Jia et al.PDF

Stacked area chart showing Pendle yield trading market evolution from 2024-2026 with marked events.
Figure 2. Evolution of Pendle Yield Trading Market Dynamics
07

Implied Fee Rates in AMM Liquidity Pools

A no-arbitrage framework recovers implied fee rates for liquidity providers in Uniswap pools, revealing time-varying and maturity-dependent compensation that varies significantly with pool design.

Fanfare 3Crypto & DeFiDiogo Duarte et al.PDF

Implied fee-rate surface for ETH-USDC showing variation by moneyness and time to maturity
Figure 3 : Implied fee-rate surface for the ETH-USDC pair on March 1, 2022, plotted as a function of moneyness M=p_{b}/S_{0} and time to maturity T .

Risk, Credit & Banking4

09

Systemic Risk Networks Among European Banks

Using recursive network analysis of equity returns, the research identifies distinct transmitting and receiving roles among European banks and constructs measures of systemic impact and vulnerability.

Fanfare 3Risk, Credit & BankingMinh Duc Duong et al.PDF

Systemic risk quadrant map: banks plotted by impact score versus vulnerability score, sized by market cap.
Figure 8: Systemic impact versus systemic vulnerability. Marker size increases with market capitalization, and dashed lines are the sample medians. All banks are labeled…
12

Tail-Risk Scenarios via Generative Flow Networks

Generative Flow Networks systematically sample diverse and realistic financial crisis scenarios, outperforming reinforcement learning and MCMC baselines in scenario coverage and realism.

Fanfare 3Risk, Credit & BankingAdir Serruya and Gilad KatzPDF

Learned action distribution for the Close_ndx variable at each of the 12 timesteps
Fig. 2 : Learned action distribution for the Close_ndx variable at each of the 12 timesteps. The distribution shifts markedly across the horizon, showing that the time i…
13

Non-Linear Governance Premium in FDI Flows

Gradient boosting reveals that sovereign ESG improvements confer zero capital advantage until institutional quality crosses a discrete threshold, cutting forecast errors 39% during tail events.

Fanfare 3Risk, Credit & BankingSourav Bose and Taoufik BouraouiPDF

SHAP feature importance plot showing governance and institutional quality drivers of FDI inflows with marginal impact values
Figure 6: Global Drivers of FDI Inflows (Out-of-Sample XGBoost SHAP)
29

LLM-Enhanced Graphs for Financial Fraud Detection

Role-aware prompts and LLM-derived semantic enrichment of transaction graph attributes improve detection of fraud in heterogeneous card-merchant networks.

Fanfare 2Risk, Credit & BankingYixuan Chen et al.PDF

Transaction graph showing global context (yellow dashed circle) and local path evidence (black edges) around target node.
Figure 2. Global and local context for a target transaction. Global context averages the transaction neighbors under each schema, while local context captures the select…

Econometrics & Forecasting5

15

Transformer Sentiment and Volatility Forecasting

Transformer-based news sentiment does not improve volatility forecasts beyond persistence and market indicators in walk-forward tests across eleven stocks and multiple horizons.

Fanfare 2Econometrics & ForecastingSzymon Lis et al.PDF

Bar chart showing three-way distribution of sentiment labels: neutral, bearish, bullish in news corpus.
Figure 4.3: Distribution of sentiment labels in the analyzed textual corpus.
20

Time-Varying Quantile VAR Models And Inference

The research develops impulse response analysis for quantile VARs and finds that downside asymmetry in credit spreads and industrial production diminishes at longer forecast horizons.

Fanfare 2Econometrics & ForecastingJiti Gao et al.PDF

Plots of \rho_{\tau,m}(u) for \tau=0.5,0.8
Figure 1: Plots of \rho_{\tau,m}(u) for \tau=0.5,0.8
22

Compiling Implied Volatility Surfaces With Survey Sampling

The study combines survey sampling and causal activity clocks to forecast NVIDIA implied volatility surfaces, finding that stratified allocation outperforms simple random sampling for cross-sectional accuracy.

Fanfare 2Econometrics & ForecastingDeepak Bastola and Eli PrushanskyPDF

Granular NVDA Equity and Options Panel ( 2018\text{--}2025 ) and Full-Chain OPRA Index Harmonization
Figure 1: Granular NVDA Equity and Options Panel ( 2018\text{--}2025 ) and Full-Chain OPRA Index Harmonization. Panel A: Empirical 3D NVDA liquidity-weighted implied vol…
27

Semiparametric Real-Time GARCH with Nonstationarity

A two-step estimator combines kernel smoothing with GARCH to capture gradual variance changes while allowing short-run volatility to respond to current information.

Fanfare 2Econometrics & ForecastingMuyi Li et al.PDF

Returns and estimated long-run and short-run volatility components for natural gas futures.
Figure 5: U.S. natural gas futures returns and estimated volatility components over the period from May 14, 2013, to April 21, 2022.
30

Dynamic Tail Intensities in Cross-Sectional Returns

Spline-mixture models detect persistent left and right tail dynamics in equity returns independent of volatility, improving one-percent quantile loss forecasts in most panels tested.

Fanfare 2Econometrics & ForecastingYasumasa Matsuda and Keigo SatoPDF

Portfolio & Allocation4

16

Sparse Autoencoders For Crypto Portfolio Diversification

Sparse autoencoders extract interpretable factors from cryptocurrency returns and produce lower-volatility portfolios through better asset diversification, though without outperforming simple contrarian strategies.

Fanfare 2Portfolio & AllocationWeijia Han and Lisha QuPDF

Sparse autoencoder pipeline: returns to sparse code to factor selection to portfolio evaluation
Figure 1. Pipeline from the daily crypto return cross-section \mathbf{x}_{t} through a sparse code and reconstructed-return ranking to a contrarian quintile basket. Unde…
18

Measuring Leveraged ETF Performance Over Time

The study shows that measuring leveraged ETF performance with arithmetic mean compounding can rank investments backwards and proposes using realized terminal wealth and expected log returns instead.

Fanfare 2Portfolio & AllocationChris Murray and Marco SammonPDF

ULPIX and UOPIX net returns, cost adjustments, and underlying-index benchmark
Figure 3: ULPIX and UOPIX net returns, cost adjustments, and underlying-index benchmark. All paths include reinvested distributions and start at one dollar on March 31,…
19

Robust Quadratic Optimization With Wasserstein Ambiguity

A tractable mixed-integer reformulation solves distributionally robust portfolio optimization over a Wasserstein ball, unifying chance-constrained and value-at-risk formulations.

Fanfare 2Portfolio & AllocationImmanuel M. Bomze et al.PDF

Wishart Ensemble perturbation model with {\mathsf{Q}}_{\mathrm{nom}}={\mathsf{O}} . Out-of-sample performance and
Figure 3 : Wishart Ensemble perturbation model with {\mathsf{Q}}_{\mathrm{nom}}={\mathsf{O}} . Out-of-sample performance and conservatism gap for different choices of \t…
24

Mean-Variance Portfolios with RL Under Regime Switches

The study develops a model-free reinforcement learning algorithm that learns optimal no-shorting portfolio policies from trajectory data in regime-switching markets by decoupling constraints from policy exploration.

Fanfare 2Portfolio & AllocationMinghui Zhang et al.PDF

Algorithm convergence showing portfolio weights and performance metrics across iterations.
Figure 1: Iterations of \phi,w,\mathbb{E}[x_{T}] , and \operatorname{Var}[x_{T}] in Algorithm 4.1 for the target expected return z=1.05 . The first two panels show the t…

Trading, Microstructure & Execution3

21

Market Making Under State-Dependent Hawkes Dynamics

A parametric volume model with state-dependent Hawkes order flow and quote-depth multipliers informs optimal placement policies that outperform baseline quoting in realized terms.

Fanfare 2Trading, Microstructure & ExecutionJonathan Chávez Casillas and Li GanPDF

Cumulative intra-day reward comparing five market-making strategies over one trading day.
Figure 11 : Cumulative intra-day reward across the five strategies. The two closed-form policies and the myopic-greedy baseline accumulate in the order of \$10^{5} over…
25

Optimal Quote Placement via Hawkes Limit Order Books

A dynamic programming framework characterizes inventory-dependent optimal quotes for market makers facing Hawkes-driven order flow, tested on real limit-order-book data.

Fanfare 2Trading, Microstructure & ExecutionJonathan Chávez Casillas and Li GanPDF

Market maker quotes, inventory levels, and cumulative profit over a trading session.
Figure 8 : Implemented strategy on November 7, 2014. Top: ask and bid quotes and the mid-price. Middle: inventory and the \pm 5{,}000 -share thresholds. Bottom: cumulati…
26

Passive Market Impact Calibration with Queue Dynamics

A framework decomposing price into universal propagator and queue-sensitive corrections improves model fit and predicts passive limit-order impact better than classical propagator models.

Fanfare 2Trading, Microstructure & ExecutionYadh Hafsi et al.PDF

Queue sensitivity function showing decay with confidence bands and model fits.
Figure 6 : Sensitivity \kappa(q) against the best queue (AES units): sum-of-powers fit with a two-standard-error band, and affine and logarithmic fits. Dotted lines: med…

Also notable2

08

Language Models in Healthcare Equity Rankings

Language models improve one-year stock rankings for healthcare firms by 48%, especially when scientific texts cover trials or regulatory records, but contribute less at other horizons.

Fanfare 3Asset Pricing & FactorsHaotong Fu et al.PDF

Bar chart showing ranking IC across horizons (1m, 3m, 12m) with multiple model variants
Figure 2. Six of the rankings of Table 2 by horizon (the market-plus-text rows are omitted), price- and liquidity-neutral rank IC on the evaluation rows with text. Addin…
10

Political Pressure on Fed Moves Interest Rate Expectations

Analysis of four decades of news finds dovish political pressure lowers near-term rate expectations by roughly one basis point per event, while hawkish pressure shows no effect.

Fanfare 3Macro-Finance & RatesMichela Baretti et al.PDF

Three-Stage LLM Classification Pipeline
Figure 1 : Three-Stage LLM Classification Pipeline

SSRN

New working papers in finance, economics and ML

30 papers

ML & AI Methods3

01

AI Monoculture Paradox in Markets

When traders adopt more precise but common AI signals, synchronized order flow can outweigh forecast gains, reducing collective profits despite higher individual precision, a paradox that taxes can help resolve.

Fanfare 4ML & AI MethodsJose Blanchet et al.

14

AI and Cognitive Capital in Markets

A model shows that cheaper AI substitutes for trader expertise and erodes cognitive skills, yet price informativeness remains stable until unexpected AI outages expose fragility in human expertise.

Fanfare 3ML & AI MethodsAndrea Barbon and Fayçal Drissi

29

Fraud Detection Under Temporal Shift

A multi-view learning framework for fraud detection integrates transaction, behavioral, temporal and graph features while avoiding look-ahead bias, with value depending strongly on the dataset.

Fanfare 2ML & AI MethodsMichael Mireku Kwakye and Afees Odebode

Derivatives & Volatility5

02

Upsilon: Joint Spot-Vol-Time Sensitivity

The study defines Upsilon, an effective sensitivity for zero-days-to-expiration index options that captures the predictable, asymmetric joint dynamics of spot, implied volatility and time decay over their six-hour life.

Fanfare 3Derivatives & VolatilityJonathan Klip

18

Five-Minute Adverse Selection Measure

Validating against event-contract settlement data, the five-minute midquote captures only 68.7% of true price impact and significantly misprices thin markets, challenging the convention.

Fanfare 2Derivatives & VolatilityKedar Sahu

20

Correlated Random-Level Volatility Model

A six-parameter volatility model with one Gaussian factor prices European options exactly and captures implied-volatility skew better than SSVI or Heston, though short-maturity slopes remain too steep.

Fanfare 2Derivatives & VolatilitySuren Islyaev

22

Rebuilding Volatility Surfaces from Public Data

A pipeline reconstructs equity volatility surfaces from free options data and assesses reliability; a short delta-hedged straddle harvests the variance risk premium, profitable in pre-registered tests but with realistic frictions limiting the edge.

Fanfare 2Derivatives & VolatilityYasanji Ratnaike

23

Arbitrage-Free Neural Operators for Options

Proposes a neural-operator architecture that certifiably outputs arbitrage-free European call-price surfaces, addressing failure of standard neural operators to enforce no-arbitrage constraints.

Fanfare 2Derivatives & VolatilityAnastasis Kratsios et al.

Trading, Microstructure & Execution6

03

Recognition Clocks and Liquidation Batches

Research on decentralized lending shows that collateral revaluation rules predictably determine when and how many positions are liquidated together, with large price moves triggering batch liquidations far exceeding single-position boundaries.

Fanfare 3Trading, Microstructure & ExecutionFernando Hernandez-Atienza

11

Private ECB Meetings and Trading Profits

Private bilateral meetings between ECB officials and financial institutions correlate with higher trading profits, suggesting selective communication conveys economically valuable central-bank information.

Fanfare 3Trading, Microstructure & ExecutionRainer F. H. Haselmann et al.

12

Analyst Calls and Forecast Accuracy

Mobile-call data between brokerages and firms reveals that brokerage-initiated calls improve forecast accuracy, while firm-initiated calls reduce it, suggesting information gathering versus lobbying.

Fanfare 3Trading, Microstructure & ExecutionDi Lu et al.

19

Daily Mean Reversion in Global Equities

Systematic trading strategies exploiting significant negative autocorrelation across six major equity indices achieve a pooled Sharpe ratio of 1.640 with regime filtering and downside-risk controls over 2002-2026.

Fanfare 2Trading, Microstructure & ExecutionGiovanni Bertola

21

Limit Order Book Depth-Volatility Anomaly

Shallow-book imbalance shocks predict higher future volatility and positive returns, while deep-book shocks predict lower future volatility and negative returns, a pattern robust across Chinese stock constituents since 2014.

Fanfare 2Trading, Microstructure & ExecutionLin Fu et al.

25

Adverse Selection in NSE Options Market Making

Analysis of live market-making on Indian options shows losses to adverse selection within one second, but regime-detection signals that pull quotes in stress cut losses by forty-six percent for NIFTY in early tests.

Fanfare 2Trading, Microstructure & ExecutionKethan S E

LLMs & Text2

04

GenAI and Earnings Price Formation

The study shows that greater ChatGPT exposure in countries accelerates how quickly stock prices incorporate earnings-announcement information, with effects strongest where investors are sophisticated and data is transparent.

Fanfare 3LLMs & TextWai Ho Chan et al.

13

LLM Stock Ratings and News Topics

Research using FinBERT embeddings finds that firm-specific and fundamental news topics explain LLM stock ratings far better than sentiment alone, doubling the model's explanatory power.

Fanfare 3LLMs & TextVictoria Kiszczuk

Asset Pricing & Factors5

05

Macroeconomic Risk and Equity Premium

The research identifies trend-growth risk as the largest component of the equity risk premium through persistent shocks in a structural VAR, with implications depending on whether yield changes come from monetary or fiscal sources.

Fanfare 3Asset Pricing & FactorsCesar Garcia Perez de Leon

06

State-Dependent Panic Risk Pricing

A real-time panic index combining price stress, investor holdings and drawdown pain reveals that exposure to panic innovations carries a negative return premium, larger during high-panic periods and persistent for weeks.

Fanfare 3Asset Pricing & FactorsBo Su and Guohao Tang

16

Common Ownership Across 49 Countries

Study of 49 economies finds common ownership roughly doubled or tripled since 2005, concentrated in large firms and increasingly driven by the Big Three asset managers globally.

Fanfare 3Asset Pricing & FactorsMiguel Anton et al.

17

Asset Performance Across Inflation Regimes

Classifying 107 years of data into real-time inflation and growth states, the study finds sixty-forty portfolios earn negative real returns in both overheating and stagflation conditions.

Fanfare 3Asset Pricing & FactorsPanashe Bera et al.

24

Decomposing Returns with Grouped Predictors

A VAR that groups 25 predictors by persistence into three composite state variables recovers stable estimates of discount-rate and cash-flow news without specification dependence.

Fanfare 2Asset Pricing & FactorsAizhan Anarkulova et al.

Crypto & DeFi2

07

ETFs and Crypto Volatility

Analysis of eight staggered spot cryptoasset ETF launches finds that late-2025 altcoin ETFs reduced daily volatility by about eighteen percent, though crash propensity remained unchanged.

Fanfare 3Crypto & DeFisheeba kapil and HARDEK S. KAPIL

10

CoVaR Contagion Forecast Calibration

Conformal prediction corrects CoVaR understatement of crypto-treasury contagion by two- to fourfold, reducing simulated distress-day loss frequency from 30% to 7%.

Fanfare 3Crypto & DeFiLamia Sebai

Econometrics & Forecasting2

15

Earnings Prediction Markets vs Analysts

Polymarket earnings contracts incrementally improve forecasts by incorporating public information missed by analysts, though only in the final trading day before announcements.

Fanfare 3Econometrics & ForecastingTengfei Zhang

30

Official Inflation Data and Turkish Expectations

Türkiye survey respondents shown the official 30.6-32.6% inflation rate lower their expectations by 3.2 percentage points, though the effect is selective by age and political affiliation.

Fanfare 2Econometrics & ForecastingOguz Gurerk and Kadircan Çakmak

Portfolio & Allocation3

26

Replicating Low-Volatility Indices

Rebuilding low-volatility indices from public data shows that a rank-based method replicates closely but an optimization-based method fails due to index-specification ambiguity, not covariance estimation.

Fanfare 2Portfolio & AllocationSoham Parundekar

27

Distribution Timing in Currency Portfolios

Adding skewness and kurtosis to mean-variance allocations in currency markets produces economically meaningful gains that depend critically on the investment mandate and ability to adjust gross exposure.

Fanfare 2Portfolio & AllocationMattia Alfero et al.

28

Multi-Factor Investing in Borsa Istanbul

A five-characteristic long-only strategy for Turkish stocks earns 4.4% annual alpha against the cap-weighted index, but equal-weighted scaling erodes the advantage, suggesting gains come from breadth and risk management rather than characteristics.

Fanfare 2Portfolio & AllocationKemal Onat Dursun

Also notable2

08

Treasury Settlements Drive Repo Pressure

Treasury debt settlements, not reserve scarcity alone, explain recent repo-market pressure, with 100 billion in settlements raising the SOFR-IORB spread by 4.3 basis points.

Fanfare 3Macro-Finance & RatesIllia Kyslytskyi

09

Bank Systemic Risk Without Market Prices

Stressed expected loss measured from balance-sheet data captures systemic risk for unlisted banks, covering 38% of European bank assets that market-based SRISK cannot reach.

Fanfare 3Risk, Credit & BankingEric Jondeau

RePEc

Working papers curated by RePEc's NEP field reports

30 papers · 23 figures

Econometrics & Forecasting6

01

150 Years of US Inflation Expectations

A narrative intensity measure from New York Times articles since 1870 predicts inflation better than commodity-based measures and reveals that the 1930s deflation was anticipated earlier than consensus suggests.

Fanfare 4Econometrics & ForecastingElissa-Ana-Maria Iorgulescu et al.PDF

Three time series comparing inflation, narrative intensity, and grain futures spreads from 1870-2020.
Figure 3: Predictive Power of Narrative Intensity and Grain Spot-Futures Spreads.
13

DSGE Models in Central Banks

A comprehensive review traces how DSGE models became standard policy tools and assesses their value for connecting theory, data and policy experiments despite known limitations.

Fanfare 3Econometrics & ForecastingChristophe Cahn et al.PDF

Timeline showing evolution of DSGE models from macroeconomic to representative-agent to HANK toolkits
14

Financial Inclusion Determinants

Meta-analysis of 5,172 estimates finds robust evidence for technology and infrastructure as drivers of financial inclusion, while publication bias distorts reported effects across studies.

Fanfare 3Econometrics & ForecastingSamuel Fiifi Eshun and Evžen KočendaPDF

Funnel plot showing publication bias in financial inclusion effect estimates across studies.
Figure 4 - Funnel plot of partial correlation coefficients: test for publication bias
22

Energy Shocks and Household Inflation Expectations

Using German survey data, the paper finds that oil and gas supply shocks shift inflation expectations rightward and increase disagreement, with households overestimating inflationary effects.

Fanfare 2Econometrics & ForecastingMaximilian Boeck et al.PDF

Effects of energy supply shocks in 2022
Figure 7: Effects of energy supply shocks in 2022.
24

Assessing Monetary Policy with Language Models

The paper employs large language models and semi-structural models to evaluate central bank communication and its role in shaping policy expectations.

Fanfare 2Econometrics & ForecastingDiego Vilán and WanTing XuPDF

27

State-Dependent Pricing and Expectation De-Anchoring

The research finds that price-adjustment behavior and de-anchored household expectations interact super-additively to raise inflation costs, compounding risks when both channels activate.

Fanfare 2Econometrics & ForecastingJuha Kilponen and Mika KortelainenPDF

Super-additivity of state-dependent pricing and de-anchoring: four mechanisms with markup shock impulse responses
Figure 5: Super-additivity of SDP and de-anchoring: four-way mechanism comparison

Macro-Finance & Rates8

02

US Monetary Policy Spillovers in Commercial Real Estate

Research reveals that US quantitative easing raises global commercial real estate prices and credit, with spillback effects amplified through trade linkages and cross-border investor ownership.

Fanfare 4Macro-Finance & RatesDennis Bonam et al.PDF

CRE price responses to US QE shock within and outside the US across multiple variables
Figure 3: Response to a US quantitative easing shock, within and outside the US
03

US Monetary Policy and Global Risk Premia

A general equilibrium model shows US monetary tightening erodes currency-mismatched arbitrageurs' wealth, raising global risk premia and generating spillovers to international asset prices.

Fanfare 3Macro-Finance & RatesRohan Kekre and Moritz LenelPDF

impulse responses to U.S. monetary tightening
Figure 8: impulse responses to U.S. monetary tightening
05

Global Financial Cycles and Firm Investment

The study shows that rising risk premia trigger large investment contractions, especially among risky firms, through a channel linking global price of risk to financing costs via firm debt sensitivity.

Fanfare 3Macro-Finance & RatesCaitlin Hegarty et al.PDF

Decomposition of Emerging Market Corporate Borrowing Costs
Figure 2: Decomposition of Emerging Market Corporate Borrowing Costs
09

Monetary Policy Transmission in an AI Economy

A model shows monetary policy amplifies more when AI automates some workers but not others, because shielded workers' higher wages generate additional demand.

Fanfare 3Macro-Finance & RatesAlina Ermolaeva and Hannah Magdalena SeidlPDF

Stacked bar charts decomposing consumption response by income source for non-AI and AI economies
Figure 6: Decomposition of the consumption response, in percent of each economy’s steady-state consumption. Each bar feeds one equilibrium price path through the house-…
11

Macroprudential Policy Against Inflationary Shocks

Pre-emptive, countercyclical macroprudential tools strengthen banking resilience and dampen external shocks in a monetary union, while procyclical tightening worsens outcomes.

Fanfare 3Macro-Finance & RatesDiana Lima et al.PDF

12

Sentiment and Sovereign Bond Markets

A framework shows how domestic political polarization shifts both portfolio demand in sovereign bond markets and government incentives in crises, creating a sentiment-driven constraint on confrontational foreign policy.

Fanfare 3Macro-Finance & RatesStéphane Auray and Frédéric LossPDF

16

Climate Spillovers in Sovereign Ratings

Global value chains transmit climate risk to sovereigns: ignoring foreign climate spillovers underestimates total risk by 30% and raises measured borrowing costs by 11 basis points.

Fanfare 3Macro-Finance & RatesLukas Rischen and Achim HagenPDF

Temperature impacts decompose total, domestic, and foreign climate risk on sovereign ratings.
Figure 2: Temperature impacts on sovereign credit ratings
28

Demand Shocks and Monetary Policy Trade-Offs

The study shows that in New Keynesian models, more persistent demand shocks do not always boost output more; monetary policy stance determines whether permanent income or real interest rate effects dominate.

Fanfare 2Macro-Finance & RatesFABRICE COLLARD et al.PDF

Impact Multipliers and Persistence
FIGURE 2. Impact Multipliers and Persistence

Trading, Microstructure & Execution2

04

NDF Order Flow and Exchange Rate Transmission

Transaction-level data from centrally cleared NDFs shows offshore order flow moves both forward and spot rates, with a $1 billion long-USD order depreciating the spot by roughly 10 basis points.

Fanfare 3Trading, Microstructure & ExecutionGerardo Ferrara et al.PDF

Market structure diagram showing offshore investors, NDF dealers, and arbitrageurs connected by order flow arrows.
19

Cross-Border Payment Rails Compared

Fintech money-transfer operators have compressed cross-border payment costs relative to banks; stablecoins do not eliminate frictions across corridors and payment values.

Fanfare 2Trading, Microstructure & ExecutionWenxin Du et al.PDF

Median costs by fintech MTOs decline steeply with transfer amount across four tier levels.
Figure 1: Median Costs by Fintech MTOs in USD-corridors

Risk, Credit & Banking5

07

Loan Officer Pricing Power in Mortgages

Mortgage intermediaries capture over 60% of $1.5 billion in GSE fee waivers; loan officers with histories of pricing expensive mortgages pass through less than 20% of savings.

Fanfare 3Risk, Credit & BankingNeil Bhutta et al.PDF

Pass-through rates over time showing sharp drop after policy change, with confidence intervals
Figure 2: Dynamic estimates of pass-through. Regression includes county, day, and lender fixed effects, and borrower controls; borrower controls include race, gender, ag…
10

Credit Markets with New Screening Technologies

When screening errors correlate through shared AI systems, lenders specialize into distinct market segments with uniform rates or uniform quality but dispersed rates.

Fanfare 3Risk, Credit & BankingMaryam Farboodi et al.PDF

17

Stress Vulnerability Indices for Banking Sectors

Two new indicators track euro-area banking vulnerabilities using stress tests: one measures capital depletion under tail risk, the other assesses earnings capacity to absorb losses.

Fanfare 2Risk, Credit & BankingCyril Couaillier et al.PDF

Stacked area chart decomposing stress vulnerability index components over time by color.
Figure 4: Decomposition of the SVI through time
18

Insurer Demand for Corporate Bonds and Policy

UK monetary tightening raises domestic insurer demand for corporate bonds but reduces foreign demand, creating asymmetric effects driven by regulation and liability structures.

Fanfare 2Risk, Credit & BankingMarcel Brambeer et al.PDF

UK and US insurers' gilt and corporate bond holdings comparison, 2019 Q4.
Figure 23 UK and US insurers’ gilt holdings versus their UK corporate bond holdings
30

Monetary Policy and Firm Productivity

Using Austrian firm data, tighter monetary policy reduces productivity-enhancing investment and firm total factor productivity, with stronger effects among highly leveraged firms.

Fanfare 2Risk, Credit & BankingKarin Klieber et al.PDF

Causal diagram showing monetary policy stance effects on firm productivity through multiple channels.
Figure 2: Direct Effects of Monetary Policy Stance and Shocks on TFP

Portfolio & Allocation2

08

Shareholder Diversification and Investment Resilience

Firms with home-biased shareholders cut investment 20% more than diversified peers during the pandemic, revealing equity concentration weakens corporate resilience to shocks.

Fanfare 3Portfolio & AllocationElena Banu et al.PDF

Sectoral distribution of EA shareholders
Figure 2: Sectoral distribution of EA shareholders
23

Monetary Policy Investment Responsiveness Across Firms

The research documents that most firms respond little to interest rate changes, but some show strong sensitivity; substantial heterogeneity cannot be easily explained by standard firm traits.

Fanfare 2Portfolio & AllocationThomas Drechsel et al.PDF

Asset Pricing & Factors4

15

EU ETS Carbon Leakage via Trade

Causal evidence shows the EU emissions scheme drives an 8% rise in intermediate-goods imports from non-regulated countries, with 11% higher embodied carbon in manufacturing.

Fanfare 3Asset Pricing & FactorsLewoye Bantie Baylie et al.PDF

Event-study plot showing 8% rise in imports from non-ETS countries after EU ETS introduction
Figure 2: Event-study estimates of intermediate imports from non-ETS exporters into ETS- regulated importer sectors
20

Carbon Neutrality Pledges and Stock Returns

Market analysis of 887 carbon neutrality announcements reveals statistically significant negative abnormal returns over 12 months, suggesting investors view pledges as signals of future cost burdens.

Fanfare 2Asset Pricing & FactorsMichal VyletelkaPDF

Market reaction after the standard earnings call (Group 0) and earnings call connected with carbon neutraliy pledge
Figure 1: Market reaction after the standard earnings call (Group 0) and earnings call connected with carbon neutraliy pledge (Group 1)
25

Sovereign Ratings and Fiscal Fundamentals

Sovereign credit ratings overweight debt relative to primary balances, understate the role of growth, and exhibit large unexplained country-specific effects.

Fanfare 2Asset Pricing & FactorsMr. Olivier J Blanchard et al.PDF

26

Gold Safe Haven Under Shifting Conditions

Geopolitical risks increase gold returns more strongly during loose financial conditions via credit and liquidity channels, with the largest impact in 2024–26.

Fanfare 2Asset Pricing & FactorsRafael Guerra and Gabriel ZelpoPDF

Impulse response functions showing geopolitical risk impact on gold returns under tight versus loose credit conditions.
Figure 9: Cumulative IRF of GPR shocks on gold price returns conditional on credit and liquidity risk conditions. Note: Shocks are orthogonal and uncorrelated with finan…

ML & AI Methods2

21

Deep Learning for Central Bank Forecasting

Pretrained time-series foundation models match or outperform benchmarks on selected inflation and nowcasting targets but add limited gains for labor-market aggregates or irregular series.

Fanfare 2ML & AI MethodsPaulo JúlioPDF

29

AI Assistants and Investor Belief Reinforcement

AI-generated investment replies are systematically more bullish than human replies on Reddit; instruction-tuning amplifies bullishness beyond the base model's inherent bias.

Fanfare 2ML & AI MethodsSwaminathan Balasubramaniam et al.PDF

AI-human bullish gap: instruction-tuned model versus human replies
Figure 1: AI-human bullish gap: instruction-tuned model versus human replies. Panel A plots mean pA(Post+Reply) against pA(Post) in 25 equal-width bins; Panel B plots th…

Also notable1

06

Stablecoins and Bank Deposits as Substitutes

Stablecoin growth slows roughly 3.75 percentage points per standard-deviation rise in deposit rates, with effects three times larger using instrumental variables and stronger for USDC than USDT.

Fanfare 3Crypto & DeFiIñaki Aldasoro and Rashad AhmedPDF

OLS and IV estimates of stablecoin growth response to deposit rate increases over time.
Figure 6: OLS (left) and IV (right) estimates plotted against average cumulative sta- blecoin growth (dashed line). The solid line and shaded bands show the implied grow…

Read the paperOpen full-size image

    Type to search. Try rough volatility, LLM agents or FinGPT.

    ↑↓ move↵ openesc closeFull search page