dojozero 0.2.0

A platform for running AI agents on realtime sport data and make predictions about game outcomes.

DojoZero

A platform for running AI agents on realtime sport data and make predictions about game outcomes.

Installation

pip install dojozero

Optional extras

pip install dojozero[alicloud]   # Alibaba Cloud integration (OSS, credentials, SLS)
pip install dojozero[redis]      # Redis-backed data stores
pip install dojozero[ray]        # Distributed execution via Ray

Environment Setup

cp .env.example .env

Quick Start

Run a single trial locally

dojo0 run --params trial_params/nba-moneyline.yaml --trial-id nba-local-001

Resume an interrupted trial

dojo0 run --trial-id nba-local-001 --resume-latest

Run through dashboard server

# Start server with tracing
dojo0 serve --trace-backend jaeger

# Submit a trial from another terminal
dojo0 run \
  --params trial_params/nba-moneyline.yaml \
  --trial-id nba-server-001 \
  --server http://localhost:8000

Automatic scheduling with trial sources

dojo0 serve --trace-backend jaeger --trial-source "trial_sources/daily/*.yaml"

Backtest from captured events

dojo0 backtest \
  --events outputs/2026-01-12/401772976.jsonl \
  --params outputs/2026-01-12/401772976.yaml \
  --speed 100 --max-sleep 1

Arena (live visualization)

dojo0 arena --trace-backend jaeger

Command Index

Command	Purpose
dojo0 run	Start a local or server-submitted trial
dojo0 serve	Dashboard / orchestration server
dojo0 arena	Trace-backed Arena visualization server
dojo0 backtest	Replay persisted event streams
dojo0 list-sources	List trial sources
dojo0 list-trials	List scheduled trials
dojo0 clear-schedules	Clear scheduled runs

Documentation

• Documentation

License

MIT