What it does

This repo is a personal experiment in applying deep learning to algorithmic trading. So far, that’s meant “part one”: simple time-series forecasting. The author plans to layer in more sophisticated models, ensemble them, backtest rigorously, and eventually trade live. Plans are not code, but they’re clearly labeled as such.

The interesting bit

The honesty is almost refreshing. Most finance-meets-ML repos pretend they’ve cracked the market; this one admits it’s still sketching on napkins. The OpenEdge ABL language choice is a curiosity—more common in enterprise business apps than quant shops—though the README doesn’t explain why.

Key highlights

• Explicitly framed as experimental work, not a production system
• Modular roadmap: forecasting → ensembles → strategy → live trading
• 1,457 stars suggest appetite for the topic outpaces current deliverables
• Single released component: basic time-series forecasting
• No backtests, no Sharpe ratios, no tearsheets shown yet

Caveats

• README is extremely thin; no code examples, no architecture diagrams, no dataset details
• “Go live” is aspirational—no brokerage integration or risk controls are visible
• OpenEdge ABL is an unusual choice for this domain; portability concerns go unaddressed

Verdict

Worth a bookmark if you’re researching the intersection of deep learning and trading and want to watch someone else’s homework in progress. Skip it if you need working code, reproducible results, or anything resembling a strategy you could evaluate today.