arXiv subjects

Fuat Alican

Publications and source records attributed to Fuat Alican.

13 recordsLinked to original sources

arXiv2026

Beyond Picking Winners: Correlation-Driven Tail Risk in Venture Capital Portfolio Construction

We propose a Gaussian-copula-based framework that learns deal-level dependence directly from observed joint success frequencies across founder, geography, and market attributes. Holding marginal deal success probabilities fixed, deal-level correlation preserves expected portfolio outcomes but shifts the portfolio distribution toward heavier right tails and higher kurtosis. In portfolio simulations, correlation reduces the probability of modest success counts while sharply amplifying extreme upside outcomes, especially in structurally concentrated portfolios. Our findings suggest that extreme venture capital outcomes may partly reflect correlation-induced tail amplification rather than solely higher average deal quality, with potential implications for portfolio construction and risk management. We note that the observed dataset reflects selected deals with observable outcomes, which inflates apparent success rates relative to the true population base rate; however, the core finding that correlation reshapes the distributional shape while leaving the mean unchanged is structurally robust to the level of marginal success probabilities.

q-fin.PM↗

arXiv2026

CoFEE: Reasoning Control for LLM-Based Feature Discovery

Feature discovery from complex unstructured data is fundamentally a reasoning problem: it requires identifying abstractions that are predictive of a target outcome while avoiding leakage, proxies, and post-outcome signals. With the introduction of ever-improving Large Language Models (LLMs), our method provides a structured method for addressing this challenge. LLMs are well suited for this task by being able to process large amounts of information, but unconstrained feature generation can lead to weak features. In this work, we study reasoning control in LLMs by inducing cognitive behaviors for improving feature discovery. We introduce CoFEE (Cognitive Feature Engineering Engine), a reasoning control framework that enforces cognitive behaviors in how the LLM reasons during feature discovery. From a machine learning perspective, these cognitive behaviors act as structured inductive biases over the space of candidate features generated by the model. These behaviors have been exploited with success in ML models, and include backward chaining from outcomes, subgoal decomposition, verification against observability and leakage criteria, and explicit backtracking of rejected reasoning paths. In a controlled comparison, we show that enforcing cognitive behaviors yields features with higher empirical predictability than those under unconstrained vanilla LLM prompts. CoFEE achieves an average Success Rate Score that is 15.2% higher than the vanilla approach, while generating 29% fewer features and reducing costs by 53.3%. Using held-out feature evaluation, we assess whether cognitively induced features generalize beyond the data used for discovery. Our results indicate that, in our evaluated setting, reasoning control is associated with improvements in quality and efficiency of LLM-based feature discovery.

cs.AI↗

arXiv2026

From Stochastic Answers to Verifiable Reasoning: Interpretable Decision-Making with LLM-Generated Code

Large language models (LLMs) are increasingly used for high-stakes decision-making, yet existing approaches struggle to reconcile scalability, interpretability, and reproducibility. Black-box models obscure their reasoning, while recent LLM-based rule systems rely on per-sample evaluation, causing costs to scale with dataset size and introducing stochastic, hallucination-prone outputs. We propose reframing LLMs as code generators rather than per-instance evaluators. A single LLM call generates executable, human-readable decision logic that runs deterministically over structured data, eliminating per-sample LLM queries while enabling reproducible and auditable predictions. We combine code generation with automated statistical validation using precision lift, binomial significance testing, and coverage filtering, and apply cluster-based gap analysis to iteratively refine decision logic without human annotation. We instantiate this framework in venture capital founder screening, a rare-event prediction task with strong interpretability requirements. On VCBench, a benchmark of 4,500 founders with a 9% base success rate, our approach achieves 37.5% precision and an F0.5 score of 25.0%, outperforming GPT-4o (at 30.0% precision and an F0.5 score of 25.7%) while maintaining full interpretability. Each prediction traces to executable rules over human-readable attributes, demonstrating verifiable and interpretable LLM-based decision-making in practice.

cs.LG↗

arXiv2026

Probabilistic Modeling of Venture Capital Portfolio Outliers

In this paper, we define probabilistic measures for venture portfolio performance based on individual outlier probability for each investment and the dependence across investments. This work is inspired by loan portfolio modeling against default risk used in banking. In mathematical terms, we calculate the probability distribution of the sum of N non-homogeneous Boolean outcomes (investments becoming outliers) that are correlated through common factors such as overall market conditions and sector effects. Specifically, we implemented a latent-factor model in which each investment's success is the exceedance of a Gaussian latent variable composed of idiosyncratic returns and returns from interpretable shared factors (stock markets, industry sector indices, geography and founder type). The formulation follows a simulation approach to preserve heterogeneous deal-level success probabilities and uses empirically estimated correlation matrices. When applied to synthetic portfolios, our model reveals that expected outlier counts alone are insufficient statistics for evaluating venture portfolios. Portfolios with identical expected outcomes can exhibit drastically different levels of reliability and risk when various levels and forms of correlation are embedded. Diversification improves the probability of achieving a minimum number of outliers by reducing exposure to common shocks, but at the cost of lower upside, underscoring a fundamental tradeoff between reliability and magnitude of clustered successes. The framework provides a practical bridge between deal-level outlier probability assessment and objective-aware portfolio construction.

cs.CE↗

arXiv2025

LLM-AR: LLM-powered Automated Reasoning Framework

Large language models (LLMs) can already identify patterns and reason effectively, yet their variable accuracy hampers adoption in high-stakes decision-making applications. In this paper, we study this issue from a venture capital perspective by predicting idea-stage startup success based on founder traits. (i) To build a reliable prediction model, we introduce LLM-AR, a pipeline inspired by neural-symbolic systems that distils LLM-generated heuristics into probabilistic rules executed by the ProbLog automated-reasoning engine. (ii) An iterative policy-evolution loop incorporates association-rule mining to progressively refine the prediction rules. On unseen folds, LLM-AR achieves 59.5% precision and 8.7% recall, 5.9x the random baseline precision, while exposing every decision path for human inspection. The framework is interpretable and tunable via hyperparameters, showing promise to extend into other domains.

cs.AI↗

arXiv2025

VCBench: Benchmarking LLMs in Venture Capital

Benchmarks such as SWE-bench and ARC-AGI demonstrate how shared datasets accelerate progress toward artificial general intelligence (AGI). We introduce VCBench, the first benchmark for predicting founder success in venture capital (VC), a domain where signals are sparse, outcomes are uncertain, and even top investors perform modestly. At inception, the market index achieves a precision of 1.9%. Y Combinator outperforms the index by a factor of 1.7x, while tier-1 firms are 2.9x better. VCBench provides 9,000 anonymized founder profiles, standardized to preserve predictive features while resisting identity leakage, with adversarial tests showing more than 90% reduction in re-identification risk. We evaluate nine state-of-the-art large language models (LLMs). DeepSeek-V3 delivers over six times the baseline precision, GPT-4o achieves the highest F0.5, and most models surpass human benchmarks. Designed as a public and evolving resource available at vcbench.com, VCBench establishes a community-driven standard for reproducible and privacy-preserving evaluation of AGI in early-stage venture forecasting.

cs.AI↗

arXiv2025

From Limited Data to Rare-event Prediction: LLM-powered Feature Engineering and Multi-model Learning in Venture Capital

This paper presents a framework for predicting rare, high-impact outcomes by integrating large language models (LLMs) with a multi-model machine learning (ML) architecture. The approach combines the predictive strength of black-box models with the interpretability required for reliable decision-making. We use LLM-powered feature engineering to extract and synthesize complex signals from unstructured data, which are then processed within a layered ensemble of models including XGBoost, Random Forest, and Linear Regression. The ensemble first produces a continuous estimate of success likelihood, which is then thresholded to produce a binary rare-event prediction. We apply this framework to the domain of Venture Capital (VC), where investors must evaluate startups with limited and noisy early-stage data. The empirical results show strong performance: the model achieves precision between 9.8X and 11.1X the random classifier baseline in three independent test subsets. Feature sensitivity analysis further reveals interpretable success drivers: the startup's category list accounts for 15.6% of predictive influence, followed by the number of founders, while education level and domain expertise contribute smaller yet consistent effects.

cs.LG↗

arXiv2025

Random Rule Forest (RRF): Interpretable and Manageable Ensembles of LLM-Generated Questions for Predicting Success from Unstructured Data

Many high-stakes screening tasks require predicting rare outcomes from unstructured text, where errors are costly and decisions must be auditable. We introduce Random Rule Forest (RRF), an interpretable ensemble that uses a large language model (LLM) not as an end-to-end predictor but as a generator of simple YES/NO questions. Each question acts as a weak learner, and their responses are combined by a plain unit-weight vote into an auditable ``green-flags'' scorecard: enough independent positive signals indicate a higher chance of success. We argue this deliberate simplicity is a robust default when positives are scarce and learned weights are hard to estimate. We evaluate RRF in two low-base-rate domains. On early-stage startup screening from founder profiles, RRF produces a transparent scorecard whose precision is several times the base rate (with light expert input raising it further) and, unlike direct prompting, its operating point can be controlled directly. On an established Phase~I clinical-trial benchmark, RRF outperforms published baselines on the threshold-independent metrics PR-AUC and ROC-AUC. Together these show that LLMs can serve as auditable feature generators for high-stakes text-based decisions, combining transparency with competitive predictive performance.

cs.AI↗

arXiv2025

Policy Induction: Predicting Startup Success via Explainable Memory-Augmented In-Context Learning

Early-stage startup investment is a high-risk endeavor characterized by scarce data and uncertain outcomes. Traditional machine learning approaches often require large, labeled datasets and extensive fine-tuning, yet remain opaque and difficult for domain experts to interpret or improve. In this paper, we propose a transparent and data-efficient investment decision framework powered by memory-augmented large language models (LLMs) using in-context learning (ICL). Central to our method is a natural language policy embedded directly into the LLM prompt, enabling the model to apply explicit reasoning patterns and allowing human experts to easily interpret, audit, and iteratively refine the logic. We introduce a lightweight training process that combines few-shot learning with an in-context learning loop, enabling the LLM to update its decision policy iteratively based on structured feedback. With only minimal supervision and no gradient-based optimization, our system predicts startup success far more accurately than existing benchmarks. It is over 20x more precise than random chance, which succeeds 1.9% of the time. It is also 7.1x more precise than the typical 5.6% success rate of top-tier venture capital (VC) firms.

cs.AI↗

arXiv2025

Reasoning-Based AI for Startup Evaluation (R.A.I.S.E.): A Memory-Augmented, Multi-Step Decision Framework

We present a novel framework that bridges the gap between the interpretability of decision trees and the advanced reasoning capabilities of large language models (LLMs) to predict startup success. Our approach leverages chain-of-thought prompting to generate detailed reasoning logs, which are subsequently distilled into structured, human-understandable logical rules. The pipeline integrates multiple enhancements - efficient data ingestion, a two-step refinement process, ensemble candidate sampling, simulated reinforcement learning scoring, and persistent memory - to ensure both stable decision-making and transparent output. Experimental evaluations on curated startup datasets demonstrate that our combined pipeline improves precision by 54% from 0.225 to 0.346 and accuracy by 50% from 0.46 to 0.70 compared to a standalone OpenAI o3 model. Notably, our model achieves over 2x the precision of a random classifier (16%). By combining state-of-the-art AI reasoning with explicit rule-based explanations, our method not only augments traditional decision-making processes but also facilitates expert intervention and continuous policy refinement. This work lays the foundation for the implementation of interpretable LLM-powered decision frameworks in high-stakes investment environments and other domains that require transparent and data-driven insights.

cs.AI↗

arXiv2025

GPT-HTree: A Decision Tree Framework Integrating Hierarchical Clustering and Large Language Models for Explainable Classification

This paper introduces GPT-HTree, a framework combining hierarchical clustering, decision trees, and large language models (LLMs) to address this challenge. By leveraging hierarchical clustering to segment individuals based on salient features, resampling techniques to balance class distributions, and decision trees to tailor classification paths within each cluster, GPT-HTree ensures both accuracy and interpretability. LLMs enhance the framework by generating human-readable cluster descriptions, bridging quantitative analysis with actionable insights.

cs.LG↗

arXiv2024

GPTree: Towards Explainable Decision-Making via LLM-powered Decision Trees

Traditional decision tree algorithms are explainable but struggle with non-linear, high-dimensional data, limiting its applicability in complex decision-making. Neural networks excel at capturing complex patterns but sacrifice explainability in the process. In this work, we present GPTree, a novel framework combining explainability of decision trees with the advanced reasoning capabilities of LLMs. GPTree eliminates the need for feature engineering and prompt chaining, requiring only a task-specific prompt and leveraging a tree-based structure to dynamically split samples. We also introduce an expert-in-the-loop feedback mechanism to further enhance performance by enabling human intervention to refine and rebuild decision paths, emphasizing the harmony between human expertise and machine intelligence. Our decision tree achieved a 7.8% precision rate for identifying "unicorn" startups at the inception stage of a startup, surpassing gpt-4o with few-shot learning as well as the best human decision-makers (3.1% to 5.6%).

cs.LG↗

arXiv2024

SSFF: Investigating LLM Predictive Capabilities for Startup Success through a Multi-Agent Framework with Enhanced Explainability and Performance

LLM based agents have recently demonstrated strong potential in automating complex tasks, yet accurately predicting startup success remains an open challenge with few benchmarks and tailored frameworks. To address these limitations, we propose the Startup Success Forecasting Framework, an autonomous system that emulates the reasoning of venture capital analysts through a multi agent collaboration model. Our framework integrates traditional machine learning methods such as random forests and neural networks within a retrieval augmented generation framework composed of three interconnected modules: a prediction block, an analysis block, and an external knowledge block. We evaluate our framework and identify three main findings. First, by leveraging founder segmentation, startups led by L5 founders are 3.79 times more likely to succeed than those led by L1 founders. Second, baseline large language models consistently overpredict startup success and struggle under realistic class imbalances largely due to overreliance on founder claims. Third, our framework significantly enhances prediction accuracy, yielding a 108.3 percent relative improvement over GPT 4o mini and a 30.8 percent relative improvement over GPT 4o. These results demonstrate the value of a multi agent approach combined with discriminative machine learning in mitigating the limitations of standard large language model based prediction methods.

cs.AI↗