arXiv subjects

Zheli Xiong

Publications and source records attributed to Zheli Xiong.

6 recordsLinked to original sources

arXiv2026

Relief-Gated Relative Rotation for QQQ-DIA Allocation: Globally Screened Relative States, Fixed Position Mapping, Incremental Interaction Admission, and Walk-Forward Validation

This paper studies Relief-Gated Relative Rotation (RGRR), a two-ETF rule that allocates between QQQ and DIA by mapping screened relative and macro states into a continuous QQQ weight. RGRR is economic rather than mechanical: it rotates between a growth-heavy sleeve and a Dow/value-heavy sleeve only when QQQ-DIA relative states are confirmed by rate, volatility, credit, or broad-market relief conditions. Candidate main effects and interactions are globally screened with horizon-specific HAC regressions and correlation de-duplication, then held fixed during walk-forward validation. Rolling out-of-sample validation re-selects only signal-family lambdas, not the signal universe or the position mapping. The final stack contains one main effect, nine second-order interactions, and two third-order interactions. Third-order terms must also improve rolling out-of-sample Sharpe versus the main plus second-order base and survive economic-family de-duplication. The final mapping uses a fixed bounded weight transformation and includes a 10 bps one-way turnover cost. Across the 2018, 2020, and 2022 out-of-sample starts, RGRR improves Sharpe versus 100% QQQ and 50/50 QQQ-DIA in every tested interval. It improves CAGR versus 50/50 in every interval, but beats 100% QQQ on CAGR only in the 2022 window. In 2018, RGRR earns an 18.33% CAGR and 0.94 Sharpe, versus 20.50% and 0.89 for QQQ and 16.69% and 0.86 for 50/50. In 2022, it earns a 15.19% CAGR and 0.87 Sharpe, versus 14.65% and 0.70 for QQQ. The evidence supports RGRR as a risk-adjusted relative allocation rule, not a pure return-dominance rule. Its main practical weakness is high turnover, ranging from 354% to 506% annualized.

q-fin.PM↗

arXiv2026

A DeepLearning Framework for Dynamic Estimation of Origin-Destination Sequence

OD matrix estimation is a critical problem in the transportation domain. The principle method uses the traffic sensor measured information such as traffic counts to estimate the traffic demand represented by the OD matrix. The problem is divided into two categories: static OD matrix estimation and dynamic OD matrices sequence(OD sequence for short) estimation. The above two face the underdetermination problem caused by abundant estimated parameters and insufficient constraint information. In addition, OD sequence estimation also faces the lag challenge: due to different traffic conditions such as congestion, identical vehicle will appear on different road sections during the same observation period, resulting in identical OD demands correspond to different trips. To this end, this paper proposes an integrated method, which uses deep learning methods to infer the structure of OD sequence and uses structural constraints to guide traditional numerical optimization. Our experiments show that the neural network(NN) can effectively infer the structure of the OD sequence and provide practical constraints for numerical optimization to obtain better results. Moreover, the experiments show that provided structural information contains not only constraints on the spatial structure of OD matrices but also provides constraints on the temporal structure of OD sequence, which solve the effect of the lagging problem well.

cs.LG↗

arXiv2026

Large-Scale OD Matrix Estimation with A Deep Learning Method

The estimation of origin-destination (OD) matrices is a crucial aspect of Intelligent Transport Systems (ITS). It involves adjusting an initial OD matrix by regressing the current observations like traffic counts of road sections (e.g., using least squares). However, the OD estimation problem lacks sufficient constraints and is mathematically underdetermined. To alleviate this problem, some researchers incorporate a prior OD matrix as a target in the regression to provide more structural constraints. However, this approach is highly dependent on the existing prior matrix, which may be outdated. Others add structural constraints through sensor data, such as vehicle trajectory and speed, which can reflect more current structural constraints in real-time. Our proposed method integrates deep learning and numerical optimization algorithms to infer matrix structure and guide numerical optimization. This approach combines the advantages of both deep learning and numerical optimization algorithms. The neural network(NN) learns to infer structural constraints from probe traffic flows, eliminating dependence on prior information and providing real-time performance. Additionally, due to the generalization capability of NN, this method is economical in engineering. We conducted tests to demonstrate the good generalization performance of our method on a large-scale synthetic dataset. Subsequently, we verified the stability of our method on real traffic data. Our experiments provided confirmation of the benefits of combining NN and numerical optimization.

cs.AI↗

arXiv2026

Ensemble RL through Classifier Models: Enhancing Risk-Return Trade-offs in Trading Strategies

This paper presents a comprehensive study on the use of ensemble Reinforcement Learning (RL) models in financial trading strategies, leveraging classifier models to enhance performance. By combining RL algorithms such as A2C, PPO, and SAC with traditional classifiers like Support Vector Machines (SVM), Decision Trees, and Logistic Regression, we investigate how different classifier groups can be integrated to improve risk-return trade-offs. The study evaluates the effectiveness of various ensemble methods, comparing them with individual RL models across key financial metrics, including Cumulative Returns, Sharpe Ratios (SR), Calmar Ratios, and Maximum Drawdown (MDD). Our original experimental results demonstrate that ensemble methods often outperform base models in terms of risk-adjusted returns, providing better management of drawdowns and overall stability. However, both the original analysis and the additional reproduction reported in this version show that ensemble performance is sensitive to the choice of variance threshold $τ$, classifier group, RL-agent pair, and market universe. The reproduction evidence strengthens the conclusion that classifier-assisted ensemble selection can improve robustness, while also clarifying that the advantage is conditional rather than automatic across all datasets. This study emphasizes the value of combining RL with classifiers for adaptive decision-making, with implications for financial trading, robotics, and other dynamic environments.

cs.LG↗

arXiv2026

Continuous Cash-Overlay Filters for a Static Growth--Defensive Risk Sleeve: Slow-Tail Compensation, V-Shape Crash Brakes, Walk-Forward Validation, and Max-Cash Combination

This paper studies a modular cash-overlay rule for allocating between a fixed growth-defensive risky sleeve R and interest-bearing cash C. The risky sleeve is a static 50/50 combination of equal-weight growth/technology and defensive income/value ETF baskets; the target is future R-C return, with the cash leg earning the contemporaneous cash rate. Two independent filters are tested. The slow-tail filter maps continuous compensation, rate-headwind, risk-premium-compression, and rate-path-stress states into a cash weight with a 30% material-trade gate. The V-shape filter is a fast crash brake based on continuous VIX, rate, credit, drawdown, and re-entry states. A fixed max-cash layer then uses the larger cash weight requested by either filter each day. On the 2017-2026 common window, the selected max-cash combination earns an 18.83% CAGR versus 16.62% for 100% R and reduces maximum drawdown from -33.59% to -18.05%. In the main walk-forward OOS window, the expanding combination earns 19.35% versus 17.59% for 100% R, with maximum drawdown of -22.05% versus -33.59%; the rolling version earns 18.50% with the same -22.05% drawdown. Post-2022 tests show lower drawdown but lower CAGR during a strong risky-sleeve rebound. The results support modular cash overlays as drawdown-control tools rather than standalone return-enhancement claims; fully real-time variable re-screening and multiple-testing-adjusted inference remain future work.

q-fin.PM↗

arXiv2026

Continuous Timing Signals for Growth-Defensive Style Allocation: Factor Attribution, Risk Matching, and Out-of-Sample Evidence

This paper studies conditional allocation between a growth/technology ETF basket, denoted by $G$, and a defensive income/value-oriented ETF basket, denoted by $D$. The objective is not to discover a new standalone alpha factor, but to examine whether known style exposures can be dynamically allocated using macro-market timing signals. Fama-French five-factor plus momentum attribution shows that the relative portfolio $G-D$ is a recognizable style portfolio: its market beta is 0.273, its HML beta is -0.552, its momentum beta is 0.117, and its annualized alpha is 1.95\% with a Newey-West t-statistic of only 0.81. The empirical object is therefore interpreted as a growth-versus-defensive style allocation problem rather than a new return anomaly. The allocation framework replaces discrete regime labels and if-then trading rules with a continuous smooth score. The score combines rate relief, SPY drawdown depth, high-VIX stress relief, and a growth-crowding penalty. Interaction terms are smoothed with softplus functions, the total score is mapped to G/D weights through a hyperbolic tangent function, and realized weights are smoothed with EWMA. In the main aligned comparison window from June 28, 2017 to May 15, 2026, with 10bp transaction costs, the selected smooth-score policy uses a 50\% maximum active tilt and obtains a 19.24\% CAGR, a Sharpe ratio of 1.01, and a maximum drawdown of -31.63\%. It improves over 50/50 G/D, matched TNX-only, matched core-only, SPY, and volatility-matched 100\% G benchmarks. It does not, however, exceed 100\% G or the best high-G static portfolios in raw CAGR. Walk-forward and post-2022 validations provide additional evidence of drawdown reduction and risk-adjusted allocation value. Overall, the evidence supports continuous, interpretable style timing, while also showing that high static growth exposure remains a strong benchmark.

q-fin.PM↗