Stock Market Index Prediction: A Framework Based on Transfer Learning and Knowledge Graph Enrichment Through Uncertainty Using Natural Language and Fuzzy Logic
Bibliographic record
Abstract
The main characteristic of stock market dynamics is uncertainty. While most studies reduce this uncertainty to risk and use objective probabilities to represent it, natural language and fuzzy logic offer tools to go beyond objective probabilities and better represent the uncertainty between the main risk factors (SCRF or super causal risk factor) and the different stock market indices. In this article, an existing knowledge graph (KG) on a complex and uncertain causal process of the financial market dynamics is consider as the main input. We combine expert knowledge, natural language and fuzzy logic to go beyond objectives probabilities and propose a solution based on three pillars: i) The enrichment of each fact in a KG using fuzzy systems theory to introduce uncertainty, then formulation of an input sequence and an output sequence for training a Seq2Seq algorithm as an inference engine. ii) The generation of features that summarize the interaction between each SCRF and each stock index contained in a KG fact. The input sequence is the assumption at time t regarding the behavior of the stock index at time t+1, based on the current interaction between each SCRF and each stock index. The realisation of stock index at time t+1 is represented by the output sequence. (iii) Finally, transfer learning is used to share knowledge (parameter learning) between all stock indexes, increase the training sample, and preserve the stability of input distribution between training, validation, and test samples. Fine-tuning is then employed to improve the predictive power on the specific stock or target index. For instance, when targeting the most significant US index (S&P500), we achieved an accuracy rate of 76%, an F1-score of 74%, and a Matthew Correlation of 0.49, surpassing consistently industry benchmarks over a twelve-year test period. Furthermore, we maintained high and stable metrics compared to two other benchmarks during three sub-periods with high volatility and difficulty in prediction.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".