Objective Mispricing Detection for Shortlisting Undervalued Football Players via Market Dynamics and News Signals

Chinenye Omejieke; Shuyao Chen; Xia Cui

cs.LGMar 18, 2026

Objective Mispricing Detection for Shortlisting Undervalued Football Players via Market Dynamics and News Signals

Authors: Chinenye Omejieke, Shuyao Chen, Xia Cui

Organizations: School of Computing and Mathematics, Manchester Metropolitan University, Manchester, United Kingdom · Typewind Ltd., London, United Kingdom

Abstract

We present a practical, reproducible framework for identifying undervalued football players grounded in objective mispricing. Instead of relying on subjective expert labels, we estimate an expected market value from structured data (historical market dynamics, biographical and contract features, transfer history) and compare it to the observed valuation to define mispricing. We then assess whether news-derived Natural Language Processing (NLP) features (i.e., sentiment statistics and semantic embeddings from football articles) complement market signals for shortlisting undervalued players. Using a chronological (leakage-aware) evaluation, gradient-boosted regression explains a large share of the variance in log-transformed market value. For undervaluation shortlisting, ROC-AUC-based ablations show that market dynamics are the primary signal, while NLP features provide consistent, secondary gains that improve robustness and interpretability. SHAP analyses suggest the dominance of market trends and age, with news-derived volatility cues amplifying signals in high-uncertainty regimes. The proposed pipeline is designed for decision support in scouting workflows, emphasizing ranking/shortlisting over hard classification thresholds, and includes a concise reproducibility and ethics statement.

Figures & tables

Item	Value	Description
Unique Players	19,982	Structured player records from Transfermarkt
News Articles	70,627	Multi-source football news mentioning players
Market Value Records	280,334	Weekly time-stamped valuations
Avg. Articles/Player	44.20	Mean article count per player
Temporal Range	2015–2024	Shared coverage period

Table 1: Summary statistics of the integrated dataset (07/05/2015–23/01/2024).

Model	RMSE $\downarrow$	MAE $\downarrow$	R 2 $\uparrow$
XGBoost	0.453	0.337	0.935
TabNet	0.920	0.695	0.732
Linear Regression	1.109	0.863	0.611

Table 2: Regression model performance for expected log-market value. XGBoost substantially outperforms both the neural and linear baselines, capturing the complex and non-linear relationships in the data.

Figure 1: Observed versus expected market value (log scale). Players above the diagonal ( $y>x$ ) exhibit positive mispricing and are shortlisted as undervalued. The spread of points illustrates the model’s ability to identify valuation discrepancies.

	XGBoost		TabNet		Random Forest (RF)
Variant	ROC AUC	F1	ROC AUC	F1	ROC AUC	F1
Full Model	0.677	0.263	0.614	0.206	0.471	0.137
No Text	0.604	0.250	0.560	0.247	0.553	0.208
Text Only	0.514	0.264	0.508	0.295	0.488	0.050

Table 3: Ablation study for the shortlisting task (undervalued vs. non-undervalued). Performance is measured by ROC-AUC, which emphasizes ranking quality. The full model, incorporating both market and text features, consistently achieves the highest discriminative power.

Figure 2: SHAP Summary Plot for the XGBoost Regression Model, showing each feature’s contribution to predicted log - market value. Features are ranked by their global impact, with red indicating high feature values and blue indicating low values.

Panel of Figure 3 — Figure 3: Prototype deployment architecture for the multimodal undervaluation shortlisting system.

Sports Analytics

See all 27 papers

Strategic Multi-Agent Learning for Interpretable Action Valuation of All Players in Football
Oct 5, 2026Kenjiro Ide, Taiga Someya, Kohei Kawaguchi +1Sequential Decision Making Interpretability in RL
Graduate School of Informatics, Nagoya University, Nagoya, Japan. · Graduate School of Arts and Sciences, The University of Tokyo, Tokyo, Japan. · Department of Economics, The Hong Kong University of Science and Technology, Hong Kong, China. +1
StakeBench: Evaluating Language Understanding Grounded in Market Commitment
May 25, 2026Yunhua Pei, Jingyu Hu, Yiwei Shi +3LLM Evaluation Forecasting Benchmarks
University of Bristol, Bristol, UK
Modeling Matches as Language: A Generative Transformer Approach for Counterfactual Player Valuation in Football
Mar 16, 2026Miru Hong, Minho Lee, Geonhee Jo +4Autoregressive Transformer Counterfactual Prediction
University of Seoul, Seoul, Republic of Korea · Saarland University, Saarbrücken, Germany · KAIST, Daejeon, Republic of Korea +2

Oct 5, 2026J/K move · Enter open · S save

Strategic Multi-Agent Learning for Interpretable Action Valuation of All Players in Football

Kenjiro Ide, Taiga Someya, Kohei Kawaguchi, Keisuke Fujii

Graduate School of Informatics, Nagoya University, Nagoya, Japan. · Graduate School of Arts and Sciences, The University of Tokyo, Tokyo, Japan. · Department of Economics, The Hong Kong University of Science and Technology, Hong Kong, China.+1

Sequential Decision Making Interpretability in RL Action Quality Assessment Multi-Agent Reinforcement Learning

Open paper AskPDF arXiv

Valuing player actions in football requires accounting for strategic interactions among 22 players, including off-ball movements and defensive positioning. Existing reinforcement-learning-based methods commonly aggregate decisions at the team level or estimate player values independently, leaving strategic interdependence among players insufficiently represented. This study proposes an action valuation framework inspired by Markov perfect equilibrium (MPE) for all players. Each possession is modeled as a finite-horizon dynamic game, with each player represented as an autonomous agent whose policy depends on the current game state. MPE is used as a motivating solution concept rather than an exact equilibrium. To improve interpretability, we use Expandable Decision-Making States (EDMS) and decompose the Q-value into a successor-feature basis and a linear reward-weight vector. The value basis is estimated by linear TD initialization followed by nonlinear refinement. Using tracking and event data from 95 J1 League matches, we compare the proposed formulation with an independent reinforcement learning baseline. Because the two formulations define TD errors in different target spaces, TD MSE is used only for within-formulation consistency. With EDMS fixed, the independent baseline assigns the highest value to forward movement in 99.21% of evaluated off-ball states, whereas the most frequent direction under the proposed formulation accounts for 17.63%. Team-level average Q-values show a negative association with season-level expected goals for the baseline and a weakly positive association for the proposed formulation. Qualitative analyses illustrate context-dependent valuations of off-ball movements and defensive positioning. Overall, the proposed formulation produces more context-sensitive action rankings, although the comparison does not isolate the MPE-inspired component.

Abstract

Sports Analytics

Explore similar work

Strategic Multi-Agent Learning for Interpretable Action Valuation of All Players in Football

StakeBench: Evaluating Language Understanding Grounded in Market Commitment

Modeling Matches as Language: A Generative Transformer Approach for Counterfactual Player Valuation in Football