Education Platform | AI Forecasting Prototype

Equity Education

LSTM Equity Forecasting Tkinter GUI Technical Indicators Educational Tooling

Equity Education is a Track 2 prototype designed to help beginner traders interpret market behavior using an educational interface that combines model predictions with supporting chart context.

The system pairs an LSTM-based forecasting pipeline with a Python GUI to display predictions, technical indicators, and training behavior in a format that is easier to explore than raw market data alone.

Open Full Report
Model LSTM sequence model for short/medium/long forecasting horizons
Data Yahoo Finance OHLCV data with engineered technical indicators
Target Users Beginner traders needing interpretable prediction context
Outcome Educational decision-support workflow with configurable training

Overview

The project addresses a practical problem for novice users: technical indicators are widely available, but often difficult to interpret without experience. Equity Education turns those signals into a guided workflow that connects predictions to readable visual context.

Instead of treating forecasting as an isolated model output, the system is built as an instructional interface where training progress, market context, and prediction behavior are visible together.

Approach & Takeaways

The workflow combines configurable LSTM training, feature engineering, and evaluation views in a Tkinter GUI. Users can choose training setups, run experiments, and compare predicted movement against historical price behavior and indicators.

Reported results show stable, interpretable predictions with directional performance in the low-to-mid 50% range, reinforcing the prototype’s strength as an educational tool rather than a standalone trading engine.

System & Evaluation

Prototype architecture, workflow context, and one-week directional evaluation snapshots.

System layout and workflow diagram for Equity Education UI and model pipeline
System view showing how data, forecasting, and educational interface components connect in the prototype.

One-Week Directional Results

  • SPY Evaluation Directionally correct on 4 of 5 days, with predicted directional accuracy values in the low-to-mid 50% range.
  • AAPL Evaluation Directionally correct on 3 of 5 days, highlighting more difficulty adapting to faster single-stock movement.
  • Interpretation The prototype was most effective as an interpretive, educational decision-support tool rather than a high-confidence forecasting engine.

SPY One-Week Evaluation

One-week SPY evaluation table showing predicted and actual movement with directional correctness
SPY results table from the report, showing day-by-day predicted move, actual move, directional accuracy, and direction correctness.

AAPL One-Week Evaluation

One-week AAPL evaluation table showing predicted and actual movement with directional correctness
AAPL results table from the report, used to compare model behavior on a more volatile single-stock scenario.