J-Reit Investment Utilizing Mobile Spatial Statistics

Proceedings of The 4th International Conference on Advanced Research in Management, Business and Finance

Year: 2022



J-Reit Investment Utilizing Mobile Spatial Statistics

Takuya Kaneko, Yutaro Mishima, Shinya Wada, Rui Kimura



In this paper, we propose an idea of investing strategy on J-REIT (Japan REIT) by utilizing Mobile Spatial Statistics (MSS). More specifically, we counted the number of mobile phone users, who permitted us to analyze, mesh by mesh (MSS) and utilized these statistics to estimate operating rate of specific real estate which are included in the target REIT portfolio. Firstly, we checked its correctness by comparing with actual monthly occupancy rate officially disclosed on IR (answer data). And after our confirming its accuracy, we utilized daily operating rates for daily investing judgement to improve its efficiency. We supposed that we take long position when the standardized occupancy rate (SOR) is high (SOR is greater than upper threshold) and we take short position when the SOR is low (SOR is smaller than lower threshold). Otherwise, we supposed to take sideline strategy (neutral position). Our numerical experiments indicated that our strategy’s performance resulted better than 24% annual return and its sharp ratio was better than 14 for almost two years’ verification. These numbers are enough excellent to be included in the list of top performance J-REIT fund reported/aggregated by Morningstar [1]. We obtained above results by utilizing clearly classified training and verification data. We standardized daily data only with training / historical data (N-days historical data for each trading day) also decided/set daily trading threshold only with historical data (M-days historical SOR). We introduce detail settings and further experimental results in our presentation.

keywords: Big Data analytics, Alternative Data, Machine Learning, AI in Finance.