Apogée
An open-source research initiative exploring crypto-market forecasting scaling laws
Why Apogée?
Challenging the conventional wisdom about market predictability through rigorous scientific research
Quantifying Predictability
Moving beyond the binary "predictable/unpredictable" debate to measure exactly how many bits of future price action can be inferred from historical data.
Crypto Market Scaling Laws
Testing whether market forecasting follows similar scaling laws to Large Language Models - where increased data, compute, and model size yield measurable improvements.
Information-Theoretic Approach
Using rigorous information theory to measure extractable bits of information from candlestick data, moving beyond traditional technical analysis.
Technical Approach
Leveraging state-of-the-art deep learning architectures and optimized infrastructure
Dataset & Pipeline
- Multiple assets (BTC, ETH, alts)
- Multi-frequency data (1min to daily)
- High-performance data pipelines
- Look-ahead bias prevention
Benchmarked Backbones
- Transformer variants (Llama, MoE, etc.)
- State-Space Models (SSMs)
- Attention-free architectures
- Domain-specific GPU kernels
Evaluation & Scaling
- Bits of extropy measurement
- Progressive compute scaling
- Dataset size experiments
- Reproducible benchmarks
Support Our Research
Help us establish the first open-source benchmark for crypto market forecasting
Research Grant
Support our research initiative with a grant to accelerate development and ensure high-impact results.
Compute Resources
Provide GPU infrastructure or cloud computing resources to power our large-scale experiments and benchmarks.
Open Collaboration
Join our open-source community to test alternative architectures and contribute to the research.