QuantDinger
Open-source AI Trading OS: research, backtest, paper/live execution, Agent Gateway + MCP; optional TypeSafe Jev pre-trade filter on entries.
Overview
QuantDinger (github.com/OpenByteInc/QuantDinger, Apache-2.0 backend, ai.quantdinger.com and quantdinger.com) is an open-source AI Trading OS: research to Python strategies to backtest to paper or live execution to monitoring, with Agent Gateway and MCP. The stack is multi-tenant SaaS-capable across crypto and stock or forex brokers. GitHub topics include jev, typesafe-ai, mcp-server, trading, and quant. Public star count was about 12,268 when this listing was drafted.
Problem: Quant teams want AI-assisted entries without letting a chat model narrate trades into the exchange, and without trapping open positions when the inference provider blips.
Built for: Operators self-hosting OpenByteInc QuantDinger for research, backtest, paper, or live crypto and equities workflows who may enable the optional TypeSafe pre-trade gate.
First indexed on Jev Directory: 2026-09-28
Creator and team
- Name
- Open Byte Inc.
- Handle
- @OpenByteInc
How Jev is used
- Role in the product flow
- Optional pre-trade entry filter: typed Choice with confidence before regular live strategies and Quick Trade orders reach the exchange
- Primitives
- ChoiceScore
- State in
- Order intent plus strategy context, exposure, open positions, and budget state sent to TypeSafe Jev via POST /v1/systemone per README JEV-powered pre-trade decisions section.
- Decision out
- Typed Choice results with probabilities and confidence (not prose); independent checks on evidence quality, signal consistency, market regime, account risk, execution quality, and entry pass or reject; auditable decision timeline with provider, checks, result, confidence, latency, and reason in the app.
- Strategy or Quick Trade proposes an entry order
- When AI Decision Filter is enabled, QuantDinger assembles context and calls TypeSafe Jev
- Jev returns structured pass or reject with confidence and check breakdown
- Rejected entries never reach the exchange; exits, stop-loss, take-profit, and emergency paths bypass the filter
- On provider failure, policy fails open with audit log; if Jev unset, configured LLM fallback; if no AI, allow and log fail-open
QuantDinger puts Jev on the money path as an optional gate, not as the strategy brain. jarrodwatts-jev-trader is a compact Monad demo that Chooses buy or sell each block; QuantDinger is a full trading OS with research, execution, and monitoring. The README section JEV-powered pre-trade decisions is explicit: entries can be blocked before they hit the broker; risk exits and emergencies skip the filter so you are not stuck. Provider outage fails open with auditing so AI downtime cannot trap a position. Configure JEV_API_KEY, JEV_BASE_URL, JEV_MODEL, and JEV_TIMEOUT_SECONDS under System Settings AI or LLM. Live trading carries real loss risk; this listing is not investment advice; follow local law. Cite OpenByteInc/QuantDinger README for behavior, not rumor posts.
Sourced performance claims
- README documents grid, DCA, and martingale strategies excluded from the first AI Decision Filter version.Source: github.com/OpenByteInc/QuantDinger README JEV-powered pre-trade decisions
- Public GitHub repo OpenByteInc/QuantDinger had about 12268 stars and 2510 forks when this listing was drafted.Source: GitHub API September 2026
Features and stack
Features
- End-to-end quant stack: research, Python strategies, backtest, execution, monitoring
- Agent Gateway plus MCP server topics in repo metadata
- Optional TypeSafe Jev AI Decision Filter on entries with auditable timeline UI
- Fail-open provider failure handling documented in README
- Multi-broker crypto and stocks or forex support in product positioning
Stack
- Python Apache-2.0 backend
- TypeSafe System One POST /v1/systemone
- Agent Gateway and MCP
- Self-hosted or SaaS-capable deployment
Pricing: Open-source core; live trading spends real capital and TypeSafe usage when the Jev filter is enabled. Jev filter is optional.
Links
FAQ
- Is the Jev filter required?
- No. README describes it as optional on regular live strategies and Quick Trade entries. Exits and emergency actions bypass it.
- What happens if TypeSafe is down?
- README policy: provider failure is audited and fails open; if Jev is unset the app tries a configured LLM; if no AI is available it allows the order and logs fail-open so outages cannot trap positions.
- How is this different from jev-trader?
- jev-trader is Jarrod Watts Monad block demo on Kuru MON-USDC. QuantDinger is a full trading operating system with research, backtest, multi-broker execution, and a production-shaped pre-trade gate.
- Which environment variables configure Jev?
- README lists JEV_API_KEY, JEV_BASE_URL, JEV_MODEL, and JEV_TIMEOUT_SECONDS in System Settings under AI or LLM.
Related learn guides
Original Jev guidance that pairs with this product pattern.
- Jev use cases
The patterns builders actually search for: moderation, routing, triage, RAG verify, and agent gates.
- System One model
The model family behind Jev: parallel typed questions, one forward pass, probabilities you can threshold.
Related products
Hand-picked neighbors with rich profiles or overlapping tags.