DEEP NEURAL NETWORK MODELS FOR REAL-TIME FINANCIAL FORECASTING AND MARKET INTELLIGENCE
Open Access
- 1 January 2026
- journal article
- Published by American Scholarly Publishing Group Inc
- Vol. 06 (01) , 51-89
Abstract
This study addresses the problem that organizations deploy deep neural network (DNN) forecasting services in cloud and enterprise environments, yet decision teams lack quantitative evidence on which operational capabilities drive real-time forecasting effectiveness and whether forecasting gains convert into decision-ready market intelligence. The purpose was to evaluate a case-based DNN forecasting service and test how perceived capability dimensions influence Forecasting Effectiveness (FE) and Market Intelligence Effectiveness (MIE). Using a quantitative cross-sectional, case-study design, a five-point Likert survey was administered to N = 210 active users in the selected enterprise case (58.1% analysts, 21.9% traders, 20.0% risk or portfolio staff). Key capability variables were Data Quality (DQ), Feature Richness (FR), Update Responsiveness (UR), Robustness (ROB), and Explanation Quality (EQ); outcomes were FE and MIE. The analysis plan used descriptive statistics, reliability testing (Cronbach’s alpha), Pearson correlations, and two multiple regression models with diagnostic checks. Reliability was strong (α = .84 to .90). Descriptive results indicated high perceived maturity (DQ M = 4.12, SD = 0.54; FR M = 3.98, SD = 0.61; UR M = 3.85, SD = 0.66; ROB M = 3.90, SD = 0.63; EQ M = 3.76, SD = 0.70; FE M = 3.94, SD = 0.58; MIE M = 4.01, SD = 0.55). Associations supported the proposed pathway: capability correlated with FE (r = .68, p < .001) and MIE (r = .62, p < .001), and FE correlated with MIE (r = .71, p < .001). Regression Model 1 explained 56% of variance in FE (R² = .56), with significant effects for DQ (β = .32), ROB (β = .28), FR (β = .21), and UR (β = .14). Regression Model 2 explained 61% of variance in MIE (R² = .61), driven by FE (β = .52), EQ (β = .29), and UR (β = .12). The findings imply that cloud and enterprise programs should prioritize data integrity and robust delivery to improve forecast usefulness, and invest in explainability and low-latency refresh to maximize intelligence value and decision confidence. These results provide actionable levers for governance, service-level monitoring, and user-centered adoption of secure DNN forecasting platforms.Keywords
This publication has 116 references indexed in Scilit:
- Financial time series forecasting with deep learning : A systematic literature review: 2005–2019Applied Soft Computing, 2020
- Genetic algorithm-optimized multi-channel convolutional neural network for stock market predictionNeural Computing and Applications, 2019
- Deep learning with long short-term memory networks for financial market predictionsEuropean Journal of Operational Research, 2018
- Deep neural networks, gradient-boosted trees, random forests: Statistical arbitrage on the S&P 500European Journal of Operational Research, 2017
- Getting value from Business Intelligence systems: A review and research agendaDecision Support Systems, 2017
- The effect of task–individual–technology fit on user attitude and performance: An experimental investigationDecision Support Systems, 2013
- Towards business intelligence systems success: Effects of maturity and culture on analytical decision makingDecision Support Systems, 2012
- Evaluating the effects of task–individual–technology fit in multi-DSS models context: A two-phase viewDecision Support Systems, 2011
- Measuring the effects of business intelligence systems: The relationship between business process and organizational performanceInternational Journal of Accounting Information Systems, 2008
- A forecast comparison of volatility models: does anything beat a GARCH(1,1)?Journal of Applied Econometrics, 2005