基于注意力机制的LSTM股价趋势预测研究

针对中国股票市场,提出了一种基于注意力机制的LSTM股价趋势预测模型。选取42只中国上证50从2009年到2017年的股票数据为实验对象,根据股票市场普遍认可的经验规则,分别对每个技术指标进行量化处理得到股票涨跌的趋势数据,并和交易数据混合作为预测模型的输入,然后使用基于注意力机制的LSTM模型提取股价趋势特征进行预测。实验结果表明:引入股票离散型趋势数据到预测模型中,能够在已有交易数据和技术指标的基础上提升预测精确度,与传统的机器学习模型SVM和单一的LSTM模型相比,基于注意力机制的LSTM模型具有更好的预测能力。     

Holt’s exponential smoothing and neural network models for forecasting interval-valued time series

Interval-valued time series are interval-valued data that are collected in a chronological sequence over time. This paper introduces three approaches to forecasting interval-valued time series. The first two approaches are based on multilayer perceptron (MLP) neural networks and Holt’s exponential smoothing methods, respectively. In Holt’s method for interval-valued time series, the smoothing parameters are estimated by using techniques for non-linear optimization problems with bound constraints. The third approach is based on a hybrid methodology that combines the MLP and Holt models. The practicality of the methods is demonstrated through simulation studies and applications using real interval-valued stock market time series.     

A mixed index approach to identifying hedonic price models

Recent literature suggests identifying house price hedonic regressions by using instrumental variables, spatial statistics, the borders approach, panel data, and other techniques. We present an empirical application of a mixed index model, first proposed by Bowden [Bowden, R.J., 1992. Competitive selection and market data: the mixed-index problem. The Review of Economic Studies 59(3):625–633.] to identify hedonic price regressions. We compare the performance of the mixed index model to a traditional hedonic model and to a hedonic model that includes characteristics of the buyer of each house. We find the mixed index model outperforms the other models based on bootstrap distributions of predicted housing values, prediction variance, and predicted policy effects. The mixed index model distributions are less skewed and kurtotic than the other models, suggesting it more closely satisfies the classical linear regression assumption of normally distributed errors. Compared to the mixed index model, the traditional hedonic overstates the importance of lot size and school quality to house price and understates the importance of environmental quality.     

货币和税收政策对我国住房价格的影响——基于VAR模型的实证研究

基于我国1999-2010年房地产市场季度数据,本文建立向量自回归模型(VAR)和向量误差修正模型(VEC),将住房供给和需求同时纳入模型,分析我国利率政策、信贷政策和税收政策对住房价格的影响。结果显示,贷款利率在短期内对住房供给有负效应,但长期效应不明显;贷款规模对住房价格的短期冲击明显,二者之间存在正相关关系;针对保有环节征税可以通过影响住房供给,从而有效抑制住房价格上升。根据我国住房供求特点,应优先选择税收工具,通过促进住房供给实现控制房价的政策目标。     

Dissecting the myth of the house price in Chinese metropolises: allowing for behavioral heterogeneity among investors

This paper aims to demystify the housing boom in Chinese metropolises by allowing for behavioral heterogeneity among investors. We construct an agent-based model where investors are categorized into two groups: fundamentalists and chartists. In addition, the investment strategy switching is allowed between these two groups contingent on the historical performance. Using the data of five Chinese metropolises over the period 2008–2014, the results suggest that chartists dominate the housing market and make the house price maintain an upward trend, while fundamentalists play a stabilizing role. Specifically, fundamentalists can serve as a “price anchor” in the market, because the proportion of the fundamentalists is negatively associated with both the growth rate of the house price and the deviation relative to the fundamental value. Overall, the impact of the chartists on the house price is much greater than that of the fundamentalists, which contributes to the ever-increasing house price in Chinese metropolises.

     

Hedonic prices and composite commodities

In this paper we conjecture and, to an extent, prove that recently noted restrictions required for the logical coherence and empirical relevance of hedonic price models make these models no more general than traditional housing services models. In particular, intra-urban variation in hedonic prices may not be substantively related to market equilibration at all, and therefore is not evidence for the existence of housing sub-markets. Moreover, in the case of jointly produced housing characteristics, the hedonic price models are found to be less general than the traditional homogeneous housing services models.     

Incorporating a tracking signal into a state space model

It is a common practice to complement a forecasting method such as simple exponential smoothing with a monitoring scheme to detect those situations where forecasts have failed to adapt to structural change. It will be suggested in this paper that the equations for simple exponential smoothing can be augmented by a common monitoring statistic to provide a method that automatically adapts to structural change without human intervention. The resulting method, which turns out to be a restricted form of damped trend corrected exponential smoothing, is compared with related methods on the annual data from the M3 competition. It is shown to be better than simple exponential smoothing and more consistent than traditional damped trend exponential smoothing.     

A simple combination of univariate models

This paper describes the approach that we implemented for producing the point forecasts and prediction intervals for our M4-competition submission. The proposed simple combination of univariate models (SCUM) is a median combination of the point forecasts and prediction intervals of four models, namely exponential smoothing, complex exponential smoothing, automatic autoregressive integrated moving average and dynamic optimised theta. Our submission performed very well in the M4-competition, being ranked 6^th for the point forecasts (with a small difference compared to the 2^nd submission) and prediction intervals and 2^nd and 3^rd for the point forecasts of the weekly and quarterly data respectively.     

二手房价格评估的影响因素及方法选择

随着我国房地产市场的飞速发展,二手房市场日益兴盛,二手房估价也显得越来越重要。但是,二手房市场情况比较复杂,如何能使所出售的房屋符合市场价格体系定位,同时又能使交易双方的利益达到最大化,关系到群众的切身利益和二手房市场的长期健康发展。基于此,本文对二手房估价的影响因素和方法进行了探讨,以期推动二手房价格评估方法的完善。     

城市住房价格指数编制方法的拓展设计及实证检验

科学监测城市房价走势,在当前环境下尤为重要。为拓展国际通行方法编制国内单一城市房价指数的适用性,引入样本匹配重复交易法构建房价指数,以提高样本容量与可比性。基于上海数据的实证结果表明,相较于传统重复交易法和特征价格法,样本匹配重复交易法能更准确地反映住房价格变动,结果异常波动性更小,噪声影响程度更低,在克服样本代表性误差和变量缺失误差方面效果更显著,对编制国内城市房价指数具有较好应用价值。     

上海住房价格指数完善研究

在中国房地产的快速发展过程中,房地产行业的风险问题日益突出。用科学的方法反映房地产价格的变化,从而提供给市场主体正确的引导信息已变得十分迫切。本文在分析了发达国家关于住房指数理论的研究成果,结合我国目前住房指数的现状及其特点的基础上,提出了一种改进的住房价格指数方案,作为对于现有上海住房指数系统的补充和完善。     

A dymimic model of housing price determination

This paper presents and estimates a model of the resale housing market. The data are a cross-section of monthly time series obtained from the multiple-listing service for a suburb of San Diego. The model is specified and estimated as a dynamic multiple indicator multiple cause system of equations where the capitalization rate is taken to be an unobservable time series to be estimated jointly with the unknown parameters. These are estimated by maximum likelihood using an EM algorithm based upon Kalman filtering and smoothing.The specification of the model features hedonic equations for each house sale and a dynamic equation for the capitalization rate which is constrained to make the expectation of prices equal the present value of the net returns to home ownership whenever the economic variables stabilize at steady state values. Out of steady state, the capitalization rate slowly adapts to new information.The model attributes a large portion of housing price increases of the 1970's to a fall in the capitalization rate which in turn was driven by rental inflation, tax rates and mortgage rates. Post-sample simulations indicate an initial flattening of housing inflation rates and later a fall brought on by the increase in steady state capitalization rates. In-sample simulations show that although both Proposition 13 and the inflation induced rise in the marginal income tax rates provided partial explanations for the fall in capitalization rates, the single most important factor was the acceleration in price of housing services which interacted with the tax treatment of home ownership to produce an amazing 18% average annual rate of price increase over the last seven years of the 1970's.     

List price and sales prices of residential properties during booms and busts

This paper investigates the relationship between the list and sale price of residential properties over the housing cycle. In down or normal markets the list price generally exceeds the sales price; however, when the housing market is strong, homes sell for more than their list price. This observation is not consistent with the assumptions made in the standard model of home sellers’ search behavior. We consider alternative models. In one, sellers set list prices based on their expectations of future changes in sales prices and the arrival rate of buyers; however, demand shocks occur. This model partially explains our data from the Belfast, U.K. housing market, but it fails to predict the list to sales price ratio during a sustained housing boom. We next describe a model where sellers’ endogenously select their search mechanism depending on the strength of the housing market. We find support for the conjecture that sellers switch to an auction-like model during housing booms. There also is evidence that during a downturn in the market, sellers’ list prices are sticky.     

Forecasting stock index price using the CEEMDAN-LSTM model

This paper uses a mixture model that Long Short-Term Memory (LSTM) combines with Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) to forecast stock index price of Standard & Poor's 500 index (S&P500) and China Securities 300 Index (CSI300). CEEMDAN decomposes original data to obtain several IMFs and one residue. The LSTM forecasting model utilizes the decomposed data to obtain the prediction sequences. The prediction sequences are reconstructed to gain final prediction. The paper introduces contrast models such as Support Vector Machine (SVM), Backward Propagation (BP), Elman network, Wavelet Neural Networks (WAV) and their mixture models combined with the CEEMDAN. The MCS test is used as evaluation criterion and empirical results present that forecasting effects of CEEMDAN-LSTM is optimal in developed and emerging stock market.     

Treating and Pruning: New approaches to forecasting model selection and combination using prediction intervals

We propose a new way of selecting among model forms in automated exponential smoothing routines, consequently enhancing their predictive power. The procedure, here addressed as treating, operates by selectively subsetting the ensemble of competing models based on information from their prediction intervals. By the same token, we set forth a pruning strategy to improve the accuracy of both point forecasts and prediction intervals in forecast combination methods. The proposed approaches are respectively applied to automated exponential smoothing routines and Bagging algorithms, to demonstrate their potential. An empirical experiment is conducted on a wide range of series from the M-Competitions. The results attest that the proposed approaches are simple, without requiring much additional computational cost, but capable of substantially improving forecasting accuracy for both point forecasts and prediction intervals, outperforming important benchmarks and recently developed forecast combination methods.     

fETSmcs: Feature-based ETS model component selection

The well-developed ETS (ExponenTial Smoothing, or Error, Trend, Seasonality) method incorporates a family of exponential smoothing models in state space representation and is widely used for automatic forecasting. The existing ETS method uses information criteria for model selection by choosing an optimal model with the smallest information criterion among all models fitted to a given time series. The ETS method under such a model selection scheme suffers from computational complexity when applied to large-scale time series data. To tackle this issue, we propose an efficient approach to ETS model selection by training classifiers on simulated data to predict appropriate model component forms for a given time series. We provide a simulation study to show the model selection ability of the proposed approach on simulated data. We evaluate our approach on the widely used M4 forecasting competition dataset in terms of both point forecasts and prediction intervals. To demonstrate the practical value of our method, we showcase the performance improvements from our approach on a monthly hospital dataset.     

我国新建住宅长期供给价格弹性估算——基于35个大中城市的实证研究

大多数住宅模型和政策分析,都直接或间接依赖于住宅供给价格弹性的估计值：为了应对市场需求冲击,是多供给住房还是提高住宅价格？基于Mayo（1981）构建的模型,估算了我国35个主要大中型城市的新建住宅供给价格弹性。根据流量模型,2000-2007年我国的新建住宅价格弹性系数在4-11之间,2008到2013年的价格弹性在5-13之间。而存量调整模型得到了截然不同的估算结果：2008-2013年我国的新建住宅供给价格弹性在1-6之间,更精确的估算出了我国新建住宅供给市场的价格弹性。     

Bubble-like housing boom–bust cycles: Evidence from the predictive power of households’ expectations

This study examines the role of households’ expectations in predicting the housing boom–bust cycles in the United States. It incorporates two nonlinear features of housing price dynamics: a threshold co-movement between households’ expectations and housing price growth and a structural break in their interrelation. It uses the monthly good-time-to-buy (GTTB) index as a proxy for households’ expectations about the U.S. housing market, and employs the structural break threshold vector autoregression (SBTVAR) to specify breakpoints in housing market dynamics during the recent decades. The findings indicate that shifts in interactions between households’ expectations and housing price growth are synchronous with the recent housing boom–bust cycles. The SBTVAR framework outperforms other models as it captures more of the housing market's unique dynamic characteristics. The GTTB index, which governs expectation regime-switching patterns, is able to signal the recent housing bust three periods in advance.     

Automatic selection of unobserved components models for supply chain forecasting

For many companies, automatic forecasting has come to be an essential part of business analytics applications. The large amounts of data available, the short life-cycle of the analysis and the acceleration of business operations make traditional manual data analysis unfeasible in such environments. In this paper, an automatic forecasting support system that comprises several methods and models is developed in a general state space framework built in the SSpace toolbox written for Matlab. Some of the models included are well-known, such as exponential smoothing and ARIMA, but we also propose a new model family that has been used only very rarely in this context, namely unobserved components models. Additional novelties include the use of unobserved components models in an automatic identification environment and the comparison of their forecasting performances with those of exponential smoothing and ARIMA models estimated using different software packages. The new system is tested empirically on a daily dataset of all of the products sold by a franchise chain in Spain (166 products over a period of 517 days). The system works well in practice and the proposed automatic unobserved components models compare very favorably with other methods and other well-known software packages in forecasting terms.     

住房流动性对房价指数的影响及其定量分析

目前中国城市住房市场不稳定程度较高,这不仅体现为房价的迅速变化,而且反映在交易量的大幅波动上,后者就是住房流动性的变化。有学者研究指出,以实际交易价格为基础的房价指数可能会低估住房市场的波动程度。为了更准确地把握住房市场的运行状态,本文借鉴美国MIT的相关技术,分析了住房流动性（交易活跃程度）对房价指数的影响,并尝试将流动性信息引入房价指数当中。我们发现,住房流动性对于房价指数有较大影响,且符合人们对于市场走势的直观判断,能够较好地反映市场转折点。