Quant Letter
The week in quantitative finance research: 90 papers from arXiv, SSRN and RePEc, screened and ranked, 7–13 Nov 2026.
This week: generalization and robustness in execution and hedging; and systematic evidence failures in published quant research. Standouts: Microstructure Invariance For Execution Policies shows trained policies transfer across assets and venues. RL Hedging for Options cuts tail losses versus delta. Evidence Audit of 91 LLM Trading Papers catalogs severe methodological flaws—51% ignore costs, 98% have look-ahead bias.
If you read five papers this week, read these.
An execution policy trained once on 954 stocks in normalized units generalizes to unseen assets, exchanges, and crypto with minimal performance loss.
Pre-registers and audits 91 LLM trading papers, finding 51% ignore trading costs, median test window is 12 months, and 98% fail to control for look-ahead bias in model training.
A systematic audit of 1,820 economics and finance papers finds 10% of claims show discrepancies between code and description, with 49% containing potentially serious errors affecting central results.
Topics drawing unusually many papers this week.
6.4% of this week's 689 new papers, 1.5x its share over the previous 7 weeks (4.3%).
1.9% of this week's 689 new papers, 2.1x its share over the previous 7 weeks (0.9%).
4.5% of this week's 689 new papers, 1.5x its share over the previous 7 weeks (3.0%).
What the quant community starred on GitHub this week, from a panel of 10,589 developers who follow financial machine learning (270 of the 857 we checked starred something). Counts only: we never publish who starred what.
Personal finance app for managing money and budgets.
★ 14.5k+3,657 this weekRuby
Self-hosted AI agent managing budget, investments, and trades.
★ 437+418 this weekPythonNew repo
Forward-testing harness for trading crypto perpetuals on Bybit.
★ 145+145 this weekPythonNew repo
Free open source order flow screener for spot and perpetual trading.
★ 184+151 this weekTypeScriptNew repo
FinRL-X: AI-native modular infrastructure for quantitative trading.
★ 3,938+27 this weekPython
Backtesting engine for quantitative trading and portfolio analysis.
★ 648+41 this weekPython
Production-grade charting library for brokers and trading platforms.
★ 197+8 this weekTypeScript
Foundation model for the language of financial markets.
★ 47.7k+655 this weekPython
Agent skills framework for building AI agent capabilities.
★ 8,098+8,081 this weekNew repo
Local emulator of AWS APIs for development and testing.
★ 6,507+6,510 this weekRustNew repo
Python framework for semantic graph-based reasoning over data.
★ 12.5k+3,088 this weekPython
Web best practices guidance for coding agents.
★ 9,906+7,280 this week
Understand how software works and build your own.
★ 31.1k+12.3k this weekRustNew repo
Open-source plugin for designing and reviewing software with agents.
★ 18.4k+5,239 this weekPythonNew repo
Now published in International Conference on Learning Representations · featured 3 Oct 2025, 4 days after release; 30 citations
Now published in International Conference on Learning Representations · featured 11 Jun 2025, 1 day after release; 13 citations
Now published in Finance and Stochastics · featured 12 Dec 2024, 2 days after release; 6 citations
Filter by venue, topic or fanfare.
No papers match these filters.
Quantitative finance and ML-for-finance preprints
30 papers · 29 figures
An execution policy trained once on 954 stocks in normalized units generalizes to unseen assets, exchanges, and crypto with minimal performance loss.
Fanfare 4Trading, Microstructure & ExecutionHieu Tran and Hong YuPDF
A reinforcement learning agent trained on historical S&P 500 options data reduces tail losses by 3.1% versus delta hedging, reaching 8.9% in volatile downturns.
Fanfare 3Trading, Microstructure & ExecutionJacky Chen et al.PDF
A no-go theorem proves that propagator estimation cannot identify asymptotic impact decay because diffusive prices force fine-tuned decay artificially.
Fanfare 2Trading, Microstructure & ExecutionShunta Fujiwara et al.PDF
Adding inventory and cash management penalties to reinforcement learning market makers improves resilience and adaptability compared to pure profit-maximizing approaches.
Fanfare 2Trading, Microstructure & ExecutionSiddharth Nand and Bora GuneyPDF
Neural networks approximate the solution map of optimal execution under transient price impact, reducing financial costs from kernel misspecification.
Fanfare 2Trading, Microstructure & ExecutionRyan Donnelly et al.PDF
A pre-registered evaluation finds zero evidence that five frozen time-series foundation models beat a no-change forecast on index futures returns across 160 tests.
Fanfare 3Econometrics & ForecastingRishabh UpadhyayPDF
A tabular foundation model pretrained on synthetic data outperforms tree-based and language models at predicting next-year earnings changes.
Fanfare 3Econometrics & ForecastingEd deHaan and Allen H. HuangPDF
VAPOR jointly estimates vector autoregressive dynamics and factor-structured precision to reveal direct dependencies in high-dimensional macroeconomic time series.
Fanfare 3Econometrics & ForecastingPeter DunsonPDF
The study conditions self-supervised learning on market factors to improve prediction of returns, exposure, risk, and liquidity across minute-level stock data.
Fanfare 3ML & AI MethodsYuqi Wang et al.PDF
Combining linear and nonlinear predictors as specialists improves rank IC by over 13% on daily Chinese stock returns versus neural networks alone.
Fanfare 3ML & AI MethodsDanning LiPDF
A generative model pretrained on crypto order-flow data and fine-tuned for microprice forecasting outperforms gradient boosting and transfers zero-shot to unseen assets.
Fanfare 3ML & AI MethodsFedor ShabashevPDF
![Directional skill vs lead time: corr (\mu_{h}, return over (t{+}\text{gap},t{+}\text{gap}{+}h]) against lead gap (as a](figures/arxiv-06-3023ff5e.png)
A 2.68 million-parameter model with grouped-query attention reduces inference latency and matches larger baselines on financial time-series forecasting across multiple assets.
Fanfare 2ML & AI MethodsChiang Qin Zhi et al.PDF
Evaluates LLM agents on 408 investment research tasks, finding that half of workflows passing result checks contain temporal violations, revealing a 50.8% gap between output agreement and temporal validity.
Fanfare 3ML & AI MethodsDong Bo et al.PDF
Introduces FinCutBench, a monthly updated benchmark measuring parametric knowledge boundaries in financial LLMs, tracking 9,000 questions from 2023 to 2026 to detect look-ahead bias.
Fanfare 3ML & AI MethodsSiyuan Li et al.PDF
A financial-replay benchmark reveals that persona conditioning shifts LLM behavior but does not reliably produce stable, coherent persona-specific policies.
Fanfare 1ML & AI MethodsYuyan Wang et al.PDF
Moving backtest corrections inside autonomous searches exposes a variance-dilution effect; gating with safeguards eliminates spurious discovery across 53 audited runs.
Fanfare 3Portfolio & AllocationEray GençayPDF
Walsh decomposition reveals how interactions among signals drive turnover and persistence; most single-signal portfolios beat quoted spreads but few multi-signal combinations do.
Fanfare 3Portfolio & AllocationZiyao Wang et al.PDF
Develops a robust procedure using radially weighted alpha estimation to identify skilled mutual funds, controlling false discovery rates even under heavy-tailed returns and cross-fund dependence.
Fanfare 3Portfolio & AllocationPing Zhao et al.PDF
Presents a continuous-action reinforcement learning method that learns to internalize investment policy constraints without post-hoc correction, reducing mandate violations from 96% to 4.8%.
Fanfare 3Portfolio & AllocationZeynep Yıldız et al.PDF
A continuous-time reinforcement learning framework using Thompson Sampling balances wealth and information gain, with exploration endogenously tied to posterior uncertainty.
Fanfare 2Portfolio & AllocationKexin Chen et al.PDF
The research identifies five seasonality anomalies across bond, commodity and equity futures; slower seasonal patterns produce positive net returns after costs while fast patterns do not.
Fanfare 3Asset Pricing & FactorsAndrew Butler et al.PDF
The study shows that investors learning about both news and valuation multipliers from earnings surprises respond less on average but more sharply near zero when multiplier uncertainty is high.
Fanfare 3Asset Pricing & FactorsThomas Graeber et al.PDF
Fitting nonlinear time series momentum models to individual markets rather than globally produces modestly better out-of-sample returns despite smaller sample sizes.
Fanfare 2Asset Pricing & FactorsAndrew Butler et al.PDF
Introduces FirstSwap, a decentralized exchange giving traders early price guarantees before consensus via a scaled admissibility protocol, improving welfare and reducing arbitrage losses.
Fanfare 3Crypto & DeFiWilliam Diamond et al.PDF
Faster blocks lower priority-fee competition and shift rents to arbitrageurs; Base Flashblocks data confirms liquidity peaks at interior speeds.
Fanfare 3Crypto & DeFiRuizhe Jia and Shihao YuPDF
A micro-macro-micro framework combining numerical states with linguistic market interpretations and cross-asset communication improves coordinated trading decisions in perpetual futures.
Fanfare 2Crypto & DeFiJiashun Wang et al.PDF
Orthogonal methods remove bias in downside-risk backtests when estimating high-dimensional Value-at-Risk and Expected Shortfall forecasts.
Fanfare 2Risk, Credit & BankingZaichao Du et al.PDF
The research derives bounds on conditional value-at-risk when outcome data is hidden, enabling statistical certificates for tail-risk claims across predictors.
Fanfare 2Risk, Credit & BankingEl Mustapha MansouriPDF
Discrete-time volatility models preserving polynomial moments reduce implied volatility root mean squared errors by 8–41% versus benchmarks across options spanning 1996–2023.
Fanfare 3Derivatives & VolatilityMaciej Augustyniak et al.PDF
Encoders trained on earnings calls with yearly cutoffs raise rank correlation of return forecasts by about half and improve earnings-surprise prediction.
Fanfare 2LLMs & TextAlessio BriniPDF
New working papers in finance, economics and ML
30 papers · 1 figures
Pre-registers and audits 91 LLM trading papers, finding 51% ignore trading costs, median test window is 12 months, and 98% fail to control for look-ahead bias in model training.
Fanfare 4ML & AI MethodsHeng Zhang
An AI-based proxy advisor using firm context and text embeddings beats traditional advisors by 6 to 7 percentage points, earning 10.3% annualized alpha on contested board decisions.
Fanfare 4ML & AI MethodsSean S. Cao et al.
A practitioner's guide proposes a reference architecture for deploying agentic AI systems in quantitative research, balancing flexibility and control through two reasoning components around a deterministic execution core.
Fanfare 3ML & AI MethodsAnna-Helena Mihov and Philip Treleaven
A systematic audit of 1,820 economics and finance papers finds 10% of claims show discrepancies between code and description, with 49% containing potentially serious errors affecting central results.
Fanfare 4Econometrics & ForecastingChristoph Huber et al.
A sample of 1,500 sports and politics prediction markets shows prices are well calibrated with no favourite-longshot bias, though calibration slightly worsens seven days before events.
Fanfare 2Econometrics & ForecastingJoseph Glasson
Tests 15 pre-registered option strategies and published signals on Nasdaq-100 options with real execution costs, finding none produce tradable edges after correcting for multiple comparisons.
Fanfare 4Derivatives & VolatilityDharm Desai
The paper derives term premia and convexity from rebalancing strategies rather than theory, showing that volatility smile materially affects convexity value and linking the result to empirical swap estimates.
Fanfare 2Derivatives & VolatilityVladimir Piterbarg and Wouter Van der Helm
Describes a pre-registered protocol using sequentially adjusted significance levels to judge trading strategies, applied to two Euro Stoxx 50 futures strategies that both fail strict statistical tests.
Fanfare 4Trading, Microstructure & ExecutionDaniele Frongia and Fabio Crescenzi
The study tracks commodity traders through explosive price episodes and finds managed money follows trends but stops buying as long-horizon trends strengthen, suggesting extrapolation does not amplify bubbles.
Fanfare 3Trading, Microstructure & ExecutionRafael Baptista Palazzi
Using exogenous variation in maintenance margins across futures, the research finds higher margins reduce trading volume by 2.67%, cut liquidity by 7.51%, and raise volatility by 2.59%.
Fanfare 3Trading, Microstructure & ExecutionGabriel Bracons
The research derives a closed-form lower bound on trading costs that depends only on return and order-flow covariances, showing that Kyle's lambda achieves optimality for multi-asset liquidity provision.
Fanfare 3Trading, Microstructure & ExecutionAndreas Park and Katya Malinova
Automates asset pricing research using agents to propose and test stochastic discount factors, finding one with out-of-sample Sharpe ratio of 3.38 and 0.86 rank correlation across periods.
Fanfare 4Asset Pricing & FactorsAlejandro Lopez-Lira et al.
Robust econometric methods reduce the fraction of 102 nontraded factors classified as priced from 39% to 4%-13%, suggesting prior tests overstated their pricing ability.
Fanfare 4Asset Pricing & FactorsAmedeo Andriollo et al.
Machine learning finds no robust evidence that sustainability characteristics predict stock returns after multiple-testing corrections, contradicting prior research.
Fanfare 4Asset Pricing & FactorsEnrica Archetti et al.
Models a computational arms race where many weak signals drive return predictability and explain why uncertainty about signal coefficients sustains an equity premium.
Fanfare 4Asset Pricing & FactorsSergei Glebkin et al.
Incorporating upper and lower tail information from machine learning models increases Sharpe ratios for long-short equity portfolios across ten prediction models from 1992 to 2024.
Fanfare 3Asset Pricing & FactorsYifei Wang
Professional forecasters systematically overestimate Treasury yields and stock returns when expecting monetary tightening, neglecting how policy changes alter the price of risk.
Fanfare 3Asset Pricing & FactorsPeter Vaporakis
Research shows that sorting currencies on long-run interest-rate differentials yields a carry strategy with higher Sharpe ratios and significant alpha versus traditional carry.
Fanfare 3Asset Pricing & FactorsMohammed Mehdi Kaebi and Igor Ferreira Batista Martins
Financial institutions excluded from factor studies show meaningful factor premiums; including them with adapted definitions improves risk-adjusted portfolio returns.
Fanfare 3Asset Pricing & FactorsGuido Baltussen et al.
A scenario-analysis framework reveals strong cross-state and cross-sector heterogeneity in climate risk, with portfolio-level valuations declining roughly 4.0% under current policy.
Fanfare 3Asset Pricing & FactorsVincent Bouchet and Nicolas Schneider
The study maps the financial network supporting AI buildout, revealing concentrated debt exposures and network propagation risks that could cause losses reaching $16.87 trillion in a low-productivity state.
Fanfare 4Risk, Credit & BankingStijn Van Nieuwerburgh and Zijin Su
The paper quantifies that reasonable modeling choices for Value-at-Risk and Expected Shortfall create capital gaps of a third between identical portfolios at different banks.
Fanfare 3Risk, Credit & BankingGregor N. F. Weiss
Following ChatGPT's release, analyst forecast divergence fell sharply and AI-assisted reports show stronger textual similarity, reducing information diversity in markets.
Fanfare 4LLMs & TextShawn X. Huang et al.
Surveys applications of large language models to return prediction, risk measurement, and investor beliefs from financial text, highlighting look-ahead bias and evaluation pitfalls in existing research.
Fanfare 3LLMs & TextMohammad Gheidi
Narrative distinctiveness drives fees and flows while portfolio uniqueness predicts future alpha, suggesting fund prospectuses serve marketing while holdings signal performance ability.
Fanfare 3LLMs & TextGe Yu
A 10 basis point QE-type surprise raises real house prices 3% in the US and 1.6% in the euro area, lifts technology stocks, and shifts wealth toward the middle class without raising stock indices broadly.
Fanfare 4Macro-Finance & RatesNuoya Chen
Data on Treasury buybacks reveal dealers reduce inventories in targeted maturity segments, with larger declines among specialists and balance-sheet-constrained dealers facing binding leverage constraints.
Fanfare 3Macro-Finance & RatesCourtney Wiegand et al.
A yield-curve slope measure based on random-walk beliefs outperforms the expectations hypothesis in forecasting Treasury returns and economic value out of sample.
Fanfare 3Macro-Finance & RatesHoyong Choi
The dollar depreciated after U.S. tariff announcements not from treasury flight but from equity rotation from domestic to foreign stocks, revealed by fund flow analysis.
Fanfare 3Macro-Finance & RatesRashad Ahmed et al.
Centralized model portfolios buffer fund flows but weaken incentives for active management, reducing performance persistence and dampening rewards for outperformance.
Fanfare 3Portfolio & AllocationPietro Lazzaretto
Working papers curated by RePEc's NEP field reports
30 papers · 22 figures
Positive residual stock shocks reverse slower than negative shocks, and this asymmetric reversal explains short-term predictability better than idiosyncratic volatility alone.
Fanfare 4Asset Pricing & FactorsFederico Baldi-Lanfranchi et al.PDF
Expected excess returns on government bonds correlate 67% with bond-stock betas and account for more than half of long-term yield moves since the mid-1980s.
Fanfare 3Asset Pricing & FactorsMatteo Leombroni et al.PDF
The study finds that AI patenting activity predicts higher stock market returns across 61 countries, with strongest effects in the EU, UK, and G7 economies.
Fanfare 3Asset Pricing & FactorsGuglielmo Maria Caporale et al.PDF
Analysis of Dutch residential appraisals shows they copy realized transaction prices rather than provide independent valuations, raising collateral risk for borrower and lender.
Fanfare 2Asset Pricing & FactorsRamon F. A. de Punder and Mathijs R. G. DijkstraPDF
A retrospective reviews behavioral finance's progress over four decades, covering earnings beliefs, return beliefs, prospect theory, and emerging directions in cognitive foundations and AI.
Fanfare 2Asset Pricing & FactorsNicholas C. BarberisPDF
Lengthening government debt maturity raises term premia and affects inflation differently under monetary versus fiscal dominance, with lasting price-level consequences.
Fanfare 4Macro-Finance & RatesArtur Doshchyn et al.PDF
A heterogeneous-agent model shows quantitative easing and liquidity facilities outperform operation twist at the zero bound and increase wealth inequality most through portfolio shifts.
Fanfare 3Macro-Finance & RatesCharles de Beauffort and Ansgar RannenbergPDF
Fixed-floating interest rate parity violations in agency debt are driven by money market fund demand for floating notes rather than arbitrageur constraints.
Fanfare 2Macro-Finance & RatesMatthias Fleckenstein et al.PDF
Analyzing security-level data shows low real rates cut debt ratios substantially, while inflation surprises mattered little, and central-bank holdings initially reduced but later raised consolidated debt.
Fanfare 2Macro-Finance & RatesJef Boeckx et al.PDF
Currency depreciations relative to the dollar stimulate borrowing from abroad and economic growth for peggers, driven by lower uncovered interest parity deviations rather than expenditure switching.
Fanfare 2Macro-Finance & RatesMasao Fukui et al.PDF
Structural VARs and a DSGE model establish that post-pandemic inflation in the US and Euro Area was predominantly driven by strong demand from fiscal expansion and accommodative policy.
Fanfare 3Econometrics & ForecastingDomenico Giannone and Giorgio PrimiceriPDF
Prediction market data on macroeconomic outcomes show Fed rate hikes affect expectations through standard transmission channels, not through an information effect.
Fanfare 2Econometrics & ForecastingEric T. Swanson et al.PDF
The study links price dispersion in food and gasoline to household uncertainty and shows contractionary policy compresses this uncertainty, suggesting optimal policy should vary by sector.
Fanfare 2Econometrics & ForecastingPhilip Schnattinger and Prachi SrivastavaPDF
Incorporating professional forecasts into GARCH models improves predictions of joint tail risks for growth, inflation, and unemployment, enabling better scenario-based risk assessment.
Fanfare 2Econometrics & ForecastingManuel Schick and Anne OpschoorPDF
Professional forecasters hold heterogeneous narratives about inflation's impact on stocks, causing disagreement in expectations and divergent portfolio choices despite common signals.
Fanfare 2Econometrics & ForecastingChristian Conrad et al.PDF
Proposes computationally efficient estimators for multivariate stochastic volatility models using moment equations and shrinkage methods, avoiding numerical optimization while maintaining reliability.
Fanfare 1Econometrics & ForecastingMD Nazmul Ahsan and Jean-Marie DufourPDF
A DSGE model reveals that interest-rate policy transmission to output remains robust as non-bank finance grows, but quantitative easing effectiveness depends critically on non-bank sector composition.
Fanfare 3Risk, Credit & BankingNicolò Bandera et al.PDF
A regression discontinuity design shows that SBA disaster lending preserved 9.8% of jobs in affected counties, operating primarily through reduced firm exit rather than new hiring.
Fanfare 2Risk, Credit & BankingFederico Esposito and Saverio SimonelliPDF
The link between sovereign and bank risk remains economically meaningful for less-capitalized banks with large sovereign holdings, especially during market stress and in fiscally constrained countries.
Fanfare 2Risk, Credit & BankingCharlotte Hoopes et al.PDF
Digital financial inclusion boosts bank profitability across Sub-Saharan Africa but shows weaker effects on stability, with impact concentrated in high-rate, low-inflation environments.
Fanfare 2Risk, Credit & BankingEric Kofi Boadi et al.PDF
Teams outperform solo managers by 11.5 basis points during outflows through threat-rigidity cohesion, but underperform by 36 basis points during inflows due to social loafing.
Fanfare 3Portfolio & AllocationCristhian Rodriguez-Revilla et al.PDF
Passive ownership raises return correlations and reduces achievable portfolio diversification by 2.1%, shifting risk from firm-specific to market-wide components.
Fanfare 3Portfolio & AllocationI. GuflerPDF
Investment funds buy depressed stocks of geopolitically exposed firms seeking yield, stabilizing valuations during shocks while raising portfolio concentration.
Fanfare 3Portfolio & AllocationDaniel Marcel te Kaat et al.PDF
Tracking 6.8 million Reddit users reveals distinct entry routes to stock investing from budgeting, crypto, or gambling forums, with composition predicting market price and volume reactions.
Fanfare 2Trading, Microstructure & ExecutionOlga Balakina and Jonathan SeimPDF
Single-stock trading pauses on the Tokyo Stock Exchange reduce subsequent volatility by roughly a quarter by allowing aggressive limit orders to offset initial imbalances.
Fanfare 2Trading, Microstructure & ExecutionAkitada Kasahara and Masahiro YamadaPDF
Analysis of 114 billion consumer transactions finds only 2.9% of Americans pay for AI, with low switching costs and retention rates of 20-45%, suggesting strong competitive pressure ahead.
Fanfare 2ML & AI MethodsMark L. Egan et al.PDF
Combining a five-factor model with gradient boosting and technical indicators for hedging, the strategy outperforms index buy-and-hold on Japanese stocks while maintaining interpretability.
Fanfare 2ML & AI MethodsDaiya Mita and Akihiko TakahashiPDF
A latent flow-matching framework generates implied volatility surfaces enforcing no-arbitrage constraints and closely reproduces empirical distributions with 90.8% surface validity.
Fanfare 2Derivatives & VolatilityDusica Bajalica et al.PDF
Accounting for bank runs in GARCH models improves in-sample and out-of-sample performance for modeling gold price returns and volatility using annual data since 1694.
Fanfare 2Derivatives & VolatilityRuipeng Liu et al.PDF
A fine-tuned language model reveals the ECB holds predominantly negative views of non-banks since the financial crisis, driven by stability and regulatory concerns.
Fanfare 2LLMs & TextAnisa Tiza Mimun and Ami DalloulPDF
