Papers for

financial trading teams

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Large language models struggle to predict multi-actor event outcomes accurately

From Single-Actor Environments to Multi-Actor Event Worlds: Can LLMs Be Event World Models?

Abstract: World models help agents assess the consequences of their actions before acting. Many real systems, however, involve multiple actors who make their own decisions, can evolve even when the agent takes no action, and are only partially observed through event logs. We formalize these systems as event worlds and ask whether large language models can predict future observations from the log and the agent's proposed actions. We build complementary benchmarks from human Diplomacy games and real Polymarket trades. Our hierarchical framework compares prediction with and without other actors' actions, then tests multi-step rollouts, predictions for actions never taken, and usefulness to agents. Fine-tuning and more data improve prediction when all actors' recorded actions are given. With only the agent's action, models struggle when other actors strongly affect the outcome, and larger models and longer histories help little. This gap also limits multi-step prediction and agent decisions, with models overestimating plan success. Yet imagined experience and initialization from world models can still improve policy learning. Anticipating other actors' actions remains a central challenge for event world modeling.

Wed 21 OctArtificial Intelligence
The gist
The paper explores whether large language models (LLMs) can predict what will happen next in systems where many actors make decisions, like games or markets. The authors found that LLMs do better when they know all actors’ actions, but if they only know one actor’s moves, they have trouble predicting outcomes when others have big effects. Bigger models and longer history of events help little in these cases. Despite this, using imagined experiences from these models can still help improve decision-making in some situations.
Open → 2610.30488v1

Robust learning method for zero-sum games with limited data

Toward Online Robust Zero-Sum Markov Games with Function Approximation

Abstract: Online robust zero-sum Markov games pose a coupled operator-learning and game-solving problem: from nominal interaction data, the learner must recover a worst-case minimax Bellman operator while maintaining strategic coverage and controlling stage-game approximation error. Existing robust reinforcement-learning theory does not yet provide such an operator-learning perspective beyond specialized multi-agent settings. We develop an optimistic dual robust Bellman framework that combines dual robust Bellman fitting, optimistic minimax planning, backward-validated confidence sets, and exploiter-mix data collection. Our analysis is organized around the robust Bellman residuals induced by the candidate value class and yields a regret reduction that separates exploration complexity, robust operator-estimation error, stage-game solution error, and misspecification. Under explicit interpretable assumptions, we obtain conditional theorem interfaces for general and projected function approximation, together with closed results in several structured regimes, including tabular, epochwise cross-fit linear, self-normalized pure-online linear / finite-rank kernel, and projected RKHS truncation. Small-scale numerical experiments illustrate robustness calibration, strategic coverage, and the return-gap trade-off predicted by the framework.

Mon 19 OctMachine LearningArtificial Intelligence
The gist
Sometimes, computer programs need to learn how to make decisions in uncertain, competitive settings where they face an opponent. This paper studies a way to help such programs learn the worst-case strategies they might face, even when only limited or imperfect data is available from past interactions. The authors develop a new method that carefully balances exploring different actions, estimating risks, and planning future moves to reduce mistakes. They prove mathematically how this method works in different situations and show small experiments to confirm its behavior.
Open → 2610.26286v1

Adversarial co-evolution improves trading between trader and critic agents

ACE-Trader: Adversarial Co-Evolution for Generator-Critic Trading Systems

Abstract: Generator--critic trading systems pair an LLM-based trader that makes trading decisions with a red team agent that evaluates these decisions and provides feedback for revision. However, improving individual trading decisions does not ensure that both agents learn from experience and improve together over time. This raises an important question: how can the trader and red-Team select useful experience and learn from each other's evolving behavior? On one hand, experience that benefits decisions under one market condition may become ineffective or harmful under another. On the other hand, trading outcomes reflect both agents' actions, making it difficult to derive appropriate experience for each role. In this work, we propose ACE-Trader, an adversarial co-evolution trading framework. Specifically, we first develop separate context-dependent memory selectors that learn to identify useful experience for the current trading decision. Both selectors receive shared feedback from net-return differences between complete memory-assisted and memory-free trading pipelines, with online updates emphasizing recent outcomes. We further design a co-evolution mechanism that links role-specific memories through shared trading records. We use these records to evaluate the Red-Team's risk assessments and the effects of the Trader's revisions against realized market outcomes, then derive experience for their respective memories. By linking this experience to the corresponding records, we enable the Trader to improve its proposals based on prior critiques and the Red-Team to adjust its review focus as the Trader's behavior changes. Experiments on U.S. stock trading show that ACE-Trader outperforms all 22 active baselines in cumulative return and achieves the highest Sharpe ratio and lowest maximum drawdown across both active methods and passive benchmarks.

Mon 19 OctMachine Learning
The gist
Trading systems often pair a trader that makes decisions with a critic that checks those decisions. But just improving each decision doesn’t mean both parts get better together over time. The authors created a system called ACE-Trader where the trader and critic learn from each other by sharing useful past experiences, adjusting based on how the market changes and how each side behaves. Their experiments on US stock data showed ACE-Trader beats other systems in overall returns and risk management.
Open → 2610.25847v1

Market makers optimize revenue under inventory limits with new learning methods

No-Regret Revenue Maximization in Sequential Market Making Under Inventory Constraints

Abstract: Market makers keep trading efficient by setting bid and ask prices for buyers and sellers. In sequential market making, trades happen independently based on price thresholds, requiring the market maker to maintain balanced, low-risk inventory over T rounds. This paper studies how to maximize revenue with unknown buyer and seller valuations using regret minimization. When the market maker only sees trade success or failure, an algorithm achieves regret and inventory bounds proportional to T to the power of 2/3. When full order book data is available for failed trades, regret and inventory limits improve significantly to the square root of T. We also establish a fundamental trade-off via a lower bound, by proving that tighter round-by-round inventory limits strictly force higher minimum regret.

Sat 17 OctMachine Learning
The gist
Market makers help buyers and sellers trade smoothly by setting prices, but they must keep their inventory balanced to avoid big losses. This paper studies how market makers can better learn and adjust prices over many trades to earn more money even when they don’t know buyers’ or sellers’ exact values. The authors show that having more detailed information about failed trades helps improve performance significantly. They also prove that trying to keep inventory tightly controlled makes it harder to minimize missed revenue opportunities.
Open → 2610.24562v1

Timestamped decision tracking improves large language model accuracy

Fresh at Decision, Stale at Action: Routing New Information to LLM Decisions Already Made

Abstract: Freshness shapes an LLM decision at the moment it is made; the action it commits to is latent, carried out later, in a changing world. We therefore introduce a timestamp theory of decision validity and a decision ledger that realises it. Six timestamps define a decision; a freshness ceiling bounds what up-to-date evidence guarantees at action time, placing the remaining risk in information published afterwards. Accuracy then factorises into two capabilities: routing, which delivers that information to affected decisions before their revision deadlines, and temporal judgment, which reads it for the time of action. The ledger is persistent state holding each decision's times, evidence and background knowledge, indexed by watch conditions written at decision time. Controlled decision streams with exact labels confirm four predictions across five LLMs, from 8B to frontier scale: freshness alone yields exactly the accuracy the ceiling predicts; models act on decisive news once it arrives, so accuracy follows routing; the ledger's index sets what routing reaches, and watch conditions reach indirect dependencies at low exposure; and the remaining errors concern when a change takes effect. Freshness makes a decision right when it is made; routing and temporal judgment keep it right when it is carried out.

Tue 13 OctArtificial IntelligenceComputer Science and Game TheoryInformation Retrieval
The gist
Decisions made by large language models (LLMs) rely on fresh information at the exact time the decision is made, but actions often happen later when new information might arrive. The paper introduces a way to track the timing of decisions and the information used, so that updates can be routed to the right decisions before they are carried out. This helps ensure that the model acts on the most relevant and up-to-date information. The authors tested this idea on several LLMs and found it improved accuracy by making sure decisions were based on current evidence when actions occur.
Open → 2610.18126v1

Solving large asset selling problems with precise optimization methods

Non-convex integer programming: a deterministic solution method of the large block sale problem

Abstract: We study a non-convex, nonlinear integer mathematical program with a linear constraint. This problem originates in finance and consists of selling a large predetermined quantity of a financial asset over a fixed time horizon, while maximizing the total proceeds. The selling price is modeled using a deterministic penalty function capturing the market's adverse reaction to the selling flow. From a complexity perspective, the problem is in NP. We first solve the problem exactly using dynamic programming and propose several heuristics that yield tight numerical lower bounds. We then reformulate the problem into an equivalent factorable form and derive a tight convex relaxation using McCormick inequalities. This relaxation is embedded within an adapted Branch & Bound algorithm to find the global optimum in finite time. Numerical experiments show that our methods outperform several widely used generic solvers (LocalSolver, Scip, Baron, Couenne, and Bonmin) for selected instances, but do not yield tight upper bounds, beyond small instances.

Sat 10 OctDiscrete Mathematics
The gist
Selling a large amount of a financial asset over time can lower the price you get because the market reacts to big sales. The authors studied how to figure out the best way to sell that asset to make the most money. They developed exact and approximate methods to solve this problem, using ideas from math and computer science. Their methods work better than some existing software for certain cases, though they struggle on bigger problems.
Open → 2610.14962v1

Agent system uses past trading memory to improve financial decisions

Agent Memory with Episodic Retrieval for Financial Decision-Making

Abstract: Large language models (LLMs) have demonstrated strong capabilities in financial analysis and reasoning, inspiring recent advances in agent-based trading frameworks. While these systems show promise, prior approaches either emphasize long-horizon forecasting or operate as stateless analyzers, limiting their applicability to the demands of trading in complicated settings. To address these gaps, we introduce META (Memory Enhanced Trading Agent), the first RAG-like episodic-memory-augmented multi-agent framework for financial decision making. META integrates a family of specialized indicator agents (e.g., Trend, MACD, Stochastic, RSI, SMA, AVWAP, Heikin-Ashi) with a Decision Agent that fuses their reports, and a Memory module that retrieves and updates past trading episodes encoded as market state embeddings with outcomes and reflections. By recalling relevant experiences and adaptively reweighting signals under similar market regimes, META achieves improved directional accuracy and robustness under short-horizon evaluation. Our results demonstrate that episodic memory provides a powerful mechanism for regime-aware, interpretable, and low-latency decision-making in trading and decision making. The code of this project is released on GitHub.

Wed 23 SeptArtificial Intelligence
The gist
Financial trading is complicated and requires quick decisions based on lots of information. The authors created META, a system that remembers past trading experiences and uses that memory to make better decisions today. META combines several specialized expert agents with a decision maker and a memory module that recalls similar past market situations. This helps the system adapt to changing market conditions and make more accurate short-term predictions.
Open → 2609.28771v1

Transformer model improves long multivariate time series forecasting accuracy

SETTer: Sparse-Encoder Transformer for Long-term Multivariate Time Series Forecasting

Abstract: Long-term multivariate time series plays a significant role in many application areas such as power systems, trading, etc. However, their accurate prediction is quite difficult for conventional forecasting methods as they often exhibit high dimensionality and complex relationships. Recent works show that transformer-based approaches are quite effective for long-term forecasting thanks to their attention mechanism. However, in the presence of complex high-dimensional inputs, they show evidence of oversmoothing, limited capacity, and opacity. To this end, this paper introduces SETTer, a transformer-based model that addresses these challenges by incorporating novel techniques for decoupled self-attention and hybrid masking. The proposed techniques enable SETTer to effectively capture the dominant short- and long-term patterns across the temporal and channel dimensions. In addition, we enrich the model layers with simple explainable structures that indicate the discriminative pattern of SETTer. We show that with a single-layer transformer architecture, SETTer can effectively model long-term dependencies in the presence of varying data complexities. Extensive experiments on real-word benchmark datasets for long-term multivariate time series forecasting demonstrate that SETTer outperforms state-of-the-art models in 88% of the scenarios.

Thu 17 SeptMachine Learning
The gist
Forecasting many related measurements over long times is hard because the data is complex and high-dimensional. The authors developed a transformer model called SETTer that uses special attention and masking techniques to better find important patterns in both time and different data channels. SETTer also includes simple explainable parts that show what patterns it focuses on. Their tests on multiple real datasets show it beats many current top methods most of the time.
Open → 2609.20086v1