Paper: arXiv 2311.10935
Authors: Leonard Mushunje, Maxwell Mashasha, Edina Chandiwana
Abstract
The fundamental theorem behind financial markets is that stock prices are intrinsically complex and stochastic. One of the complexities is the volatility associated with stock prices. Volatility is a tendency for prices to change unexpectedly [“1”]. Price volatility is often detrimental to the return economics, and thus, investors should factor it in whenever making investment decisions, choices, and temporal or permanent moves. It is, therefore, crucial to make necessary and regular short and long-term stock price volatility forecasts for the safety and economics of investors returns. These forecasts should be accurate and not misleading. Different models and methods, such as ARCH GARCH models, have been intuitively implemented to make such forecasts. However, such traditional means fail to capture the short-term volatility forecasts effectively. This paper, therefore, investigates and implements a combination of numeric and probabilistic models for short-term volatility and return forecasting for high-frequency trades. The essence is that one-day-ahead volatility forecasts were made with Gaussian Processes (GPs) applied to the outputs of a Numerical market prediction (NMP) model. Firstly, the stock price data from NMP was corrected by a GP. Since it is not easy to set price limits in a market due to its free nature and randomness, a Censored GP was used to model the relationship between the corrected stock prices and returns. Forecasting errors were evaluated using the implied and estimated data.
Complexity vs Empirical Score
- Math Complexity: 7.0/10
- Empirical Rigor: 5.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced mathematical concepts including Gaussian Processes, censored GP models, and probabilistic regression, but it lacks specific implementation details, backtest metrics, or public code, placing it in a high-math, moderate-rigor zone.
Research Flowchart
flowchart TD
A["Research Goal:<br>Short-term Volatility Estimation for<br>High Frequency Trades using GPs"] --> B{"Data Inputs"}
B --> B1["High-Frequency Stock Price Data"]
B --> B2["Numerical Market Prediction<br>NMP Model Outputs"]
B1 & B2 --> C["Methodology: Gaussian Process<br>Framework"]
C --> C1["Step 1: Price Correction<br>Applied GP to raw NMP outputs"]
C1 --> C2["Step 2: Censored GP Modeling<br>Relationship between corrected prices<br>and returns"]
C2 --> D["Computational Process"]
D --> D1["One-day-ahead volatility<br>forecasts generated"]
D1 --> D2["Error Evaluation<br>Implied vs. Estimated Data"]
D2 --> E{"Key Findings & Outcomes"}
E --> E1["GP effectively captures<br>short-term volatility"]
E --> E2["Outperforms traditional<br>ARCH/GARCH models"]
E --> E3["Improved accuracy for<br>high-frequency trading decisions"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes correcting NMP‑predicted price volatility with a Gaussian Process and using a censored GP to model the relationship between corrected prices and volumes for short‑term forecasts.
Source sentence
The predicted price volatility from an NMP model is corrected using a GP. This process helps to improve performance compared with earlier methods for combining statistical and physical models.
Assumes 6
- Stock prices have a defined upper limit and a lower limit of zero at the JSE. market
Source sentence
As a result of the stock price control strategies available in the market, there is always a defined upper limit 𝑺𝒖𝒑𝒑𝒆𝒓 and lower limit of 0 for the stock prices placed at JSE.
- Missing values are less than 30% and are imputed using K‑nearest neighbor. data
Source sentence
The missing values were less than 30%, and to cater to them, we used the K-nearest neighbor (KNN).
- Forecasting error for 1–3‑hour ahead forecasts must be less than 10% of actual values. statistical
Source sentence
we stress that the forecasting error of 1– 3 hours ahead should be less than 10% of the actual recorded figures.
- Stock price data are Markovian, so only current price and volume are used. model
Source sentence
Since stock data prices are Markovian, we excluded the historical data in this model at the pre‑processing stage
Show 2 more
- Latent values can be realized by a Gaussian process and censored GP accounts for unobserved price records. model
Source sentence
Our analysis assumes that the latent values 𝑦∗ = 𝑓(𝑥) can be realized by a Gaussian process. Thus, to predict the actual output(𝑦), the influence of censoring is considered inside the model.
- Error threshold for the methods is 10%; accuracy must be ≤10%. statistical
Source sentence
The forecasting error threshold for the above‑specified methods is 10%. The accuracy should be less or equal to 10%.
Data
- Johannesburg Stock Exchange (JSE) recent stock price databases market-prices public
Source sentence
The datasets used in this study were extracted from the Johannesburg Stock Exchange (JSE) recent stock price databases from January 2010 to January 2023.
Costs
Not applicable
Finds 3
- The proposed GP‑Stock Price model outperforms other models, especially over the 1‑3 hour forecast horizon. performance
Source sentence
From Tables 4 and 5, we note that the proposed GP-Stock Price model performs better than the other models, especially presenting an outstanding performance in the 1-3 hours forecast horizon.
- The GP‑based approach yields about an 11% improvement in forecasting accuracy compared with the MLP‑Stock Price model. predictive-accuracy
Source sentence
the proposed model has around 11% improvement in forecasting accuracy.
- Gaussian Process forecasting beats traditional stochastic time‑series models such as GARCH(1,1) and ARIMA. performance
Source sentence
We proved that the GP performs better than other time series stochastic models such as GARCH (1,1) and ARIMA models widely used in volatility forecasting.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 30 related papers). Coverage is partial, so this is neither replication nor refutation.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.