Searcharxiv⌕ Search

arXiv subjects

Chutian Ma

Publications and source records attributed to Chutian Ma.

9 recordsLinked to original sources
arXiv2026

Beyond Accuracy: A Stability-Aware Metric for Multi-Horizon Forecasting

Traditional time series forecasting methods optimize for accuracy alone. This objective neglects temporal consistency, in other words, how consistently a model predicts the same future event as the forecast origin changes. We introduce the forecast accuracy and coherence score (forecast AC score for short) for measuring the quality of probabilistic multi-horizon forecasts in a way that accounts for both multi-horizon accuracy and stability. Our score additionally allows user-specified weights to balance accuracy and consistency requirements. As an example application, we implement the score as a differentiable objective function for training seasonal auto-regressive integrated models and evaluate it on the M4 Hourly benchmark dataset. Results demonstrate consistent improvements over traditional maximum likelihood estimation. Regarding stability, the AC-optimized model generated out-of-sample forecasts with 15.8\% reduced variance over forecasts targeting the same timestamp. In terms of accuracy, the AC-optimized model achieved considerable improvements for medium-to-long-horizon forecasts. While one-step-ahead forecasts exhibited a 3.9\% increase in MSE, forecasts from horizon three onward experienced improved accuracy, with a peak improvement of approximately 6\% in MSE at horizons 9-12. These results indicate that our metric successfully trains models to produce more stable and accurate multi-step forecasts in exchange for a relatively small degradation in one-step-ahead performance.

cs.LG↗
arXiv2025

Causal Inference in Energy Demand Prediction

Energy demand prediction is critical for grid operators, industrial energy consumers, and service providers. Energy demand is influenced by multiple factors, including weather conditions (e.g. temperature, humidity, wind speed, solar radiation), and calendar information (e.g. hour of day and month of year), which further affect daily work and life schedules. These factors are causally interdependent, making the problem more complex than simple correlation-based learning techniques satisfactorily allow for. We propose a structural causal model that explains the causal relationship between these variables. A full analysis is performed to validate our causal beliefs, also revealing important insights consistent with prior studies. For example, our causal model reveals that energy demand responds to temperature fluctuations with season-dependent sensitivity. Additionally, we find that energy demand exhibits lower variance in winter due to the decoupling effect between temperature changes and daily activity patterns. We then build a Bayesian model, which takes advantage of the causal insights we learned as prior knowledge. The model is trained and tested on unseen data and yields state-of-the-art performance in the form of a 3.84 percent MAPE on the test set. The model also demonstrates strong robustness, as the cross-validation across two years of data yields an average MAPE of 3.88 percent.

cs.AI↗
arXiv2025

A Benchmark of Causal vs. Correlation AI for Predictive Maintenance

Predictive maintenance in manufacturing environments presents a challenging optimization problem characterized by extreme cost asymmetry, where missed failures incur costs roughly fifty times higher than false alarms. Predictive maintenance in manufacturing environments presents a challenging optimization problem characterized by extreme cost asymmetry, where missed failures incur costs roughly fifty times higher than false alarms. Conventional machine learning approaches typically optimize statistical accuracy metrics that do not reflect this operational reality and cannot reliably distinguish causal relationships from spurious correlations. This study benchmarks eight predictive models, ranging from baseline statistical approaches to Bayesian structural causal methods, on a dataset of 10,000 CNC machines with a 3.3 percent failure prevalence. While ensemble correlation-based models such as Random Forest (L4) achieve the highest raw cost savings (70.8 percent reduction), the Bayesian Structural Causal Model (L7) delivers competitive financial performance (66.4 percent cost reduction) with an inherent ability of failure attribution, which correlation-based models do not readily provide. The model achieves perfect attribution for HDF, PWF, and OSF failure types. These results suggest that causal methods, when combined with domain knowledge and Bayesian inference, offer a potentially favorable trade-off between predictive performance and operational interpretability in predictive maintenance applications.

cs.AI↗
arXiv2025

The effect of latency on optimal order execution policy

Market participants regularly send bid and ask quotes to exchange-operated limit order books. This creates an optimization challenge where their potential profit is determined by their quoted price and how often their orders are successfully executed. The expected profit from successful execution at a favorable limit price needs to be balanced against two key risks: (1) the possibility that orders will remain unfilled, which hinders the trading agenda and leads to greater price uncertainty, and (2) the danger that limit orders will be executed as market orders, particularly in the presence of order submission latency, which in turn results in higher transaction costs. In this paper, we consider a stochastic optimal control problem where a risk-averse trader attempts to maximize profit while balancing risk. The market is modeled using Brownian motion to represent the price uncertainty. We analyze the relationship between fill probability, limit price, and order submission latency. We derive closed-form approximations of these quantities that perform well in the practical regime of interest. Then, we utilize a mean-variance method where our total reward function features a risk-tolerance parameter to quantify the combined risk and profit.

q-fin.MF↗
arXiv2025

On the Effect of Alpha Decay and Transaction Costs on the Multi-period Optimal Trading Strategy

We consider the multi-period portfolio optimization problem with a single asset that can be held long or short. Due to the presence of transaction costs, maximizing the immediate reward at each period may prove detrimental, as frequent trading results in consistent negative cash outflows. To simulate alpha decay, we consider a case where not only the present value of a signal, but also past values, have predictive power. We formulate the problem as an infinite horizon Markov Decision Process and seek to characterize the optimal policy that realizes the maximum average expected reward. We propose a variant of the standard value iteration algorithm for computing the optimal policy. Establishing convergence in our setting is nontrivial, and we provide a rigorous proof. Addtionally, we compute a first-order approximation and asymptotics of the optimal policy with small transaction costs.

math.OC↗
arXiv2024

Optimal two-parameter portfolio management strategy with transaction costs

We consider a simplified model for optimizing a single-asset portfolio in the presence of transaction costs given a signal with a certain autocorrelation and cross-correlation structure. In our setup, the portfolio manager is given two one-parameter controls to influence the construction of the portfolio. The first is a linear filtering parameter that may increase or decrease the level of autocorrelation in the signal. The second is a numerical threshold that determines a symmetric "no-trade" zone. Portfolio positions are constrained to a single unit long or a single unit short. These constraints allow us to focus on the interplay between the signal filtering mechanism and the hysteresis introduced by the "no-trade" zone. We then formulate an optimization problem where we aim to minimize the frequency of trades subject to a fixed return level of the portfolio. We show that maintaining a no-trade zone while removing autocorrelation entirely from the signal yields a locally optimal solution. For any given "no-trade" zone threshold, this locally optimal solution also achieves the maximum attainable return level, and we derive a quantitative lower bound for the amount of improvement in terms of the given threshold and the amount of autocorrelation removed.

math.OC↗
arXiv2023

Almost sharp global wellposedness and scattering for the defocusing conformal wave equation on the hyperbolic space

In this paper we prove a global well-posedness and scattering result for the defocusing conformal nonlinear wave equation in the hyperbolic space $\mathbb{H}^d, d \geq 3$. We take advantage of the hyperbolic geometry which yields stronger Morawetz and Strichartz estimates. We show that the solution is globally wellposed and scatters if the initial data is radially symmetric and lies in $H^{\frac{1}{2}+\epsilon}(\mathbb{H}^d)\times H^{-\frac{1}{2}+\epsilon}(\mathbb{H}^d)$, $\epsilon>0$.

math.AP↗
arXiv2022

A Scattering Result of the Radial Cubic Defocusing Schrödinger Equation on the 3d Hyperbolic Space

In this paper, we study the defocusing cubic Schrödinger equation on three dimensional hyperbolic space $\mathbb{H}^3$ with radial initial data in the Sobolev Space $H^s(0<s<1)$. Our main result is that the initial value problem is globally wellposed and scatters for $\frac{15}{16}<s<1$. This is an extension of the work of Staffilani and Yu to the three dimensional hyperbolic space.

math.AP↗