About EvalRun
A research-focused platform for exploring market data, evaluating machine learning models, and observing how time-series systems behave in live conditions.
Real-time market feeds
Repeatable evaluations
Transparent methodology
Our Story
EvalRun started as a late-night conversation between two friends who shared a passion for both coding and financial markets. We were fascinated by the academic papers and research studies describing sophisticated machine learning models that could predict market movements, but we noticed something:
Most tools in this space either present results without methodology, or require jumping between academic papers, notebooks, and datasets. EvalRun brings the workflow into one place: live data, repeatable evaluation runs, and a clear record of what was tested and why.
Again, the platform is built for experimentation and learning. It is not an automated trading product and does not provide financial advice.
Models & Research Notes
EvalRun currently provides one baseline model. Additional model families are under active evaluation and will be added once validated.
LSTM (Long Short-Term Memory) — Baseline Model
LSTMs are a standard recurrent architecture for sequential data. In EvalRun, the baseline LSTM is used for time-series regression over OHLCV windows with configurable window size and prediction horizon. Results are reported per run (e.g., RMSE/MAE/MAPE) and vary by asset, timeframe, and market regime.
Planned Additions
Model families such as GRU, tree-based methods (e.g., gradient boosting), classical baselines (e.g., ARIMA), and attention-based architectures are being assessed. New models are introduced only after repeatable evaluation across multiple assets and timeframes.
Model Pipeline
Train
Choose a model, window size, and timeframe. Run training on historical candles.
Evaluate
Store metrics, sample charts, and run details for comparison.
Predict
Reuse saved artifacts to forecast the next 10-20 candles.
Predictions are probabilistic and depend on the market regime. They are meant for research, not trading automation.
Terms of Use & Disclaimer
EvalRun is provided for research and educational purposes only. The platform offers analytical tools and model outputs derived from historical and live market data. Nothing on this platform constitutes financial advice, investment recommendations, or an offer to buy or sell any financial instrument.
Model outputs are experimental and may be incomplete, inaccurate, or subject to change. Past performance, historical evaluations, and simulated results do not guarantee future outcomes. Users are solely responsible for how they interpret and use any information provided.
EvalRun does not execute trades, manage funds, or provide automated trading services. By using the platform, you acknowledge that all analysis is performed at your own risk.