Ploy

A high-performance Polymarket trading bot focused on crypto and sports prediction markets. Ships with a terminal dashboard, multi-agent coordinator, AI-assisted analysis, and optional reinforcement learning.

Trading Platform Workspace

The workspace now also includes the platform-refactor spine:

• new-ployd: next-generation daemon entrypoint
• ployctl: operator client entrypoint
• ploytui: thin terminal operator console
• new-ploy-runner: next-generation strategy runner entrypoint
• crates/ploy-daemon-host: daemon host/bootstrap crate
• crates/ploy-runner-host: runner CLI host crate
• crates/ploy-control-client: shared operator client transport
• crates/ploy-market-data: collector/feed/scanner/discovery infrastructure
• crates/ploy-platform-runtime: control-plane runtime ownership boundary
• crates/ploy-strategy-runtime: strategy runtime ownership boundary
• crates/ploy-platform: control-plane core
• crates/ploy-trading: canonical trading lifecycle
• crates/ploy-deployments: worker protocol and supervisor
• crates/ploy-operator-contracts: shared DTO and event contracts
• crates/ploy-strategy-bundles: signal-to-intent runtime
• crates/ploy-research: replay and backtesting consumer of trading models

Current smoke path:

cargo run -p new-ployd
cargo run -p ployctl -- system status
cargo run -p ployctl -- system audit
cargo run -p ployctl -- trading status
cargo run -p ployctl -- deployments apply config/deployments/example.paper.json
cargo run -p ployctl -- deployments list
cargo run -p ployctl -- deployments inspect example.paper
cargo run -p ployctl -- trading cancel example.live <order-id>
cargo run -p ploytui
cargo run -p new-ploy-runner --features full -- run --config config/strategies/02-pm5d.unified.toml --dry-run
# realtime operator stream
curl -N http://127.0.0.1:8081/api/events/stream
rtk cargo test --test platform_smoke -- --nocapture

Deployment Workflow

The default operating model is now:

1. ployd is the only long-running platform daemon
2. strategy configs live in config/strategies/*.toml
3. deployment manifests live in config/deployments/*.json
4. operators use ployctl deployments apply|inspect|pause|resume|stop
5. ployd starts and supervises ploy-runner --config ... workers for each deployment

Do not create new per-strategy systemd units for normal operation. Use deployment manifests instead.

Bundle Resolution

Deployment manifests use bundle_id to select a strategy config:

• if bundle_id ends with .toml or contains a path separator, it is treated as a config path
• otherwise it resolves to config/strategies/<bundle_id>.toml

Example:

{
  "deployment_id": "pm5d.v3.dryrun",
  "bundle_id": "02-pm5d.v3-dryrun",
  "account_id": "acct-pm5d-dryrun",
  "max_gross_exposure": "5.00",
  "runtime_mode": "paper",
  "desired_state": "running"
}

This resolves to:

config/strategies/02-pm5d.v3-dryrun.toml

PM5D Dry-Run Deployments

The repo now includes three PM5D deployment manifests:

• config/deployments/pm5d.v2.dryrun.json
• config/deployments/pm5d.v3.dryrun.json
• config/deployments/pm5d.v4.dryrun.json

Each one targets:

• 02-pm5d.v2-dryrun.toml
• 02-pm5d.v3-dryrun.toml
• 02-pm5d.v4-dryrun.toml

Apply them with:

cargo run -p ployctl -- deployments apply config/deployments/pm5d.v2.dryrun.json
cargo run -p ployctl -- deployments apply config/deployments/pm5d.v3.dryrun.json
cargo run -p ployctl -- deployments apply config/deployments/pm5d.v4.dryrun.json

Inspect and control them with:

cargo run -p ployctl -- deployments list
cargo run -p ployctl -- deployments inspect pm5d.v3.dryrun
cargo run -p ployctl -- deployments pause pm5d.v3.dryrun
cargo run -p ployctl -- deployments resume pm5d.v3.dryrun
cargo run -p ployctl -- deployments stop pm5d.v3.dryrun

Remote host equivalent:

/opt/ploy/bin/ployctl deployments list
/opt/ploy/bin/ployctl deployments inspect pm5d.v3.dryrun
/opt/ploy/bin/ployctl deployments pause pm5d.v3.dryrun
/opt/ploy/bin/ployctl deployments resume pm5d.v3.dryrun
/opt/ploy/bin/ployctl deployments stop pm5d.v3.dryrun

Direct Runner Usage

Use new-ploy-runner -- run --config ... for:

• local backtests
• local dry-run debugging
• full live/dry-run mode requires the explicit Cargo feature: cargo run -p new-ploy-runner --features full -- run --config ...
• one-off manual strategy inspection

Example:

cargo run -p new-ploy-runner --features full -- run --config config/strategies/02-pm5d.v4-dryrun.toml --dry-run

Use deployment manifests when you want the platform daemon to own lifecycle and supervision.

Optional admin auth:

• Set PLOY_ADMIN_TOKEN or PLOY_API_ADMIN_TOKEN before booting ployd to require a bearer token on the control-plane API.
• Set PLOY_OPERATOR_TOKEN or PLOY_API_OPERATOR_TOKEN if you want a write-capable operator credential for deployment controls, intent ingress, and order cancel/replace without granting full admin access.
• Set PLOY_SIDECAR_AUTH_TOKEN if you want a read-only agent/sidecar credential for system status, deployment snapshots, trading snapshots, and the SSE event stream.
• Set PLOY_API_AUTH_COOKIE_SECRET as well if you want browser auth cookies to stay valid across daemon restarts or multiple instances.
• Set PLOY_REQUEST_RATE_LIMIT_PER_MINUTE if you need to tighten or relax the daemon HTTP request throttle. 0 disables the limiter.
• Set PLOY_LIVE_RECONCILE_BACKOFF_BASE_MS and PLOY_LIVE_RECONCILE_BACKOFF_MAX_MS if you want to tune venue outage retry backoff for live fill reconciliation.
• ployctl will automatically reuse PLOY_ADMIN_TOKEN, PLOY_API_ADMIN_TOKEN, PLOY_API_KEY, PLOY_OPERATOR_TOKEN, PLOY_API_OPERATOR_TOKEN, or PLOY_SIDECAR_AUTH_TOKEN for authenticated requests, in that order.
• Browser operator surfaces authenticate through /auth/login, which now sets an HttpOnly same-site signed session cookie so the frontend event stream can stay authenticated without storing the raw admin token in browser storage.
• ployd appends authenticated control-plane activity to run/platform/audit-log.jsonl, and ployctl system audit reads that stream back through the daemon API.
• ployctl system status now surfaces live reconcile failure count, next retry time, and the last reconcile error so venue outages are visible without digging through logs.
• Current access bands are public, read_only, operator, and admin. sidecar credentials stop at read_only; operator credentials can mutate deployments and trading controls; only admin can access audit logs and browser-login flows.

Runbooks:

• docs/runbooks/platform-startup.md
• docs/runbooks/platform-deploy.md
• docs/runbooks/live-deployment-checklist.md
• docs/runbooks/live-dry-run-drill.md

Default release workflow:

• .github/workflows/release-platform.yml
• Build-only verification: dispatch release-platform.yml with deploy=false to build, package, checksum, and upload the bundle artifact without restarting the remote service.

Remote live-host acceptance path:

• deploy with docs/runbooks/platform-deploy.md
• validate the host with docs/runbooks/live-deployment-checklist.md
• run the dry-run acceptance drill from docs/runbooks/live-dry-run-drill.md
• only then move to manual live go/no-go review

Compatibility note:

• new-ployd, new-ploy-runner, ployctl, and ploytui are the workspace entrypoints for the trading platform spine.
• The old root runtime tree has been retired from the compiled workspace.
• Remaining ploy ... examples below are historical reference only and are not runnable entrypoints in this branch.
• Current sports scope in this branch is discovery + live-state data capture + replay/backtest support.
• Sports live execution and market making remain out of scope until lifecycle safeguards are added for game-start orderbook resets.

Features

• Two runtime domains -- Crypto (BTC/ETH/SOL UP/DOWN), Sports (NBA/NFL live odds)
• Multiple strategies -- Momentum, Split-Arb, Event-Edge mispricing scanner, NBA Q3-Q4 comeback, market making
• Coordinator-managed runtime -- Canonical strategy runtime with central order queue, governance gate, risk gate, and position aggregation
• Event registry -- Automated DISCOVER -> RESEARCH -> MONITOR -> TRADE pipeline for new markets
• TUI dashboard -- Ratatui-based terminal UI with live positions, quotes, Binance price feed, and trade log
• Claude AI agent -- Advisory, autonomous, and chat modes for market analysis and trade execution
• Reinforcement learning -- PPO training, lead-lag strategies, ONNX inference (optional rl / onnx feature flags)
• Persistence -- PostgreSQL event store, checkpoints, dead-letter queue, and crash recovery
• Risk management -- Position limits, circuit breaker, daily loss limit, slippage protection, emergency stop

Architecture (Layered Live Runtime)

Production runtime now uses a 4-plane model:

• Strategy Plane: canonical Strategy implementations decide direction, timing, sizing, and state transitions.
• Capital Governance Plane: OpenClaw-style governance agents manage budget, pause/resume, throttle, and deployment-scoped policy.
• Execution Plane: the coordinator is the only live order ingress (StrategyIntent -> Governance/Risk Gate -> Queue -> Executor), plus audit trail and recovery.
• Control Plane: deployment/config projection, lifecycle control, health, observability, and rollout/shutdown wiring.

Key rule: OpenClaw does not sit in the synchronous per-order decision path for HFT. It governs boundaries; strategies decide entries/exits inside those boundaries.

Live strategies now start only through the canonical managed Strategy runtime.

Collector / backfill command routing is documented in docs/COLLECTOR_RUNBOOK.md.

New live strategies should implement the canonical Strategy contract. TradingAgent / DomainAgent are retired and only remain in historical design docs. The default workspace control plane is:

• GET /api/system/status
• GET /api/audit/logs
• GET /api/deployments
• GET /api/deployments/:id
• POST /api/deployments/:id/control
• GET /api/trading/state
• GET /reports/strategy
• POST /api/deployments/:id/intents
• POST /api/deployments/:id/orders/:order_id/cancel
• POST /api/deployments/:id/orders/:order_id/replace
• GET /api/events/stream

The older strategies/control, strategy-evaluations, /api/sidecar/*, and ploy rpc surfaces are historical reference only in this branch and are no longer part of the default operator path.

Governance agents live under crate::agents; canonical live strategy runtime ownership lives under crate::strategy, crate::coordinator, and crate::plugins.

Prerequisites

• Rust 1.75+ (2021 edition)
• PostgreSQL 15+ with an active database for event store, checkpoints, and strategy state
• Polymarket account with API credentials and a funded wallet on Polygon
• (Optional) ANTHROPIC_API_KEY for Claude AI agent commands
• (Optional) GROK_API_KEY for Grok-based sports analysis

Installation

# Clone and build
git clone https://github.com/proerror77/ploy.git
cd ploy
cargo build --release

# Build with optional feature flags
cargo build --release --features rl        # Reinforcement learning (burn + ndarray)
cargo build --release --features onnx      # ONNX model inference (tract)
cargo build --release --features analysis  # DuckDB parquet analysis

Run database migrations before first use:

export DATABASE_URL="postgres://localhost/ploy"
sqlx migrate run

Configuration

Environment Variables

Variable	Required	Description
POLYMARKET_PRIVATE_KEY	Yes	Ethereum private key for order signing
POLYMARKET_API_KEY	Yes	Polymarket CLOB API key
POLYMARKET_API_SECRET	Yes	Polymarket CLOB API secret
POLYMARKET_PASSPHRASE	Yes	Polymarket CLOB passphrase
POLYMARKET_FUNDER	No	Proxy/Magic wallet address
RELAYER_API_KEY	No	Polymarket Relayer API key for gasless account operations
RELAYER_API_KEY_ADDRESS	No	Address that owns the relayer key
DATABASE_URL	Yes	PostgreSQL connection string (overrides config)
ANTHROPIC_API_KEY	No	Required for agent and AI-powered commands
ANTHROPIC_BASE_URL	No	Optional Anthropic-compatible base URL (examples: MiniMax https://api.minimaxi.com/anthropic or https://api.minimax.io/anthropic)
ANTHROPIC_DEFAULT_OPUS_MODEL	No	Optional alias override (example: map opus → MiniMax-M2.5)
ANTHROPIC_CUSTOM_HEADERS	No	Optional custom headers in newline-separated Header: Value format (example: Authorization: Bearer <key>)
GROK_API_KEY	No	Required for Grok-based sports analysis
PLOY_RISK__CRYPTO_ALLOCATION_PCT	No	Domain capital split (0..1), used to derive crypto exposure cap
PLOY_RISK__SPORTS_ALLOCATION_PCT	No	Domain capital split (0..1), used to derive sports exposure cap
PLOY_RISK__CRYPTO_MAX_EXPOSURE_USD	No	Hard crypto domain exposure cap (overrides pct-derived cap)
PLOY_RISK__SPORTS_MAX_EXPOSURE_USD	No	Hard sports domain exposure cap (overrides pct-derived cap)
PLOY_RISK__CRYPTO_DAILY_LOSS_LIMIT_USD	No	Hard crypto domain daily loss stop
PLOY_RISK__SPORTS_DAILY_LOSS_LIMIT_USD	No	Hard sports domain daily loss stop
PLOY_RISK__MAX_DRAWDOWN_USD	No	Hard drawdown stop (runtime cumulative realized curve)
PLOY_ACCOUNT_ID	No	Runtime account scope identifier (default default)
PLOY_API_PORT	No	API listen port for new-ployd and operator clients (default 8081)
PLOY_API_ADMIN_TOKEN	No	Admin token for protected API routes, including operator terminal actions
PLOY_ADMIN_TOKEN	No	Dashboard-side fallback token name for operator polling/actions if PLOY_API_ADMIN_TOKEN is unset
PLOY_DRY_RUN__ENABLED	No	Force runtime dry-run mode (true/false)
PLOY_DEPLOYMENTS_REQUIRE_EVIDENCE	No	Require strategy evidence before enabling deployments (true/false)
PLOY_DEPLOYMENTS_REQUIRED_STAGES	No	Required evidence stages CSV (default backtest,paper)
PLOY_DEPLOYMENTS_MAX_EVIDENCE_AGE_HOURS	No	Max evidence staleness window in hours (default 168)
PLOY_ALLOW_DIRECT_LIVE	No	Allow direct (non-Coordinator) live order paths. Not recommended.
PLOY_ALLOW_DIRECT_STRATEGY_LIVE	No	Archived compatibility flag for the retired single-binary runtime; not part of the workspace default path.

Config File

The default configuration lives in config/default.toml. Override the path with --config / -c.

Section	Key examples
[market]	ws_url, rest_url, market_slug
[strategy]	shares, window_min, move_pct, sum_target, fee_buffer, slippage_buffer, profit_buffer
[execution]	order_timeout_ms, max_retries, max_spread_bps, poll_interval_ms
[risk]	max_single_exposure_usd, min_remaining_seconds, max_consecutive_failures, daily_loss_limit_usd, leg2_force_close_seconds
[database]	url, max_connections
[dry_run]	enabled (defaults to true)
[logging]	level, json
[event_edge_agent]	enabled, trade, interval_secs, min_edge, max_entry, shares, cooldown_secs, max_daily_spend_usd, titles
[nba_comeback]	enabled, min_edge, max_entry_price, shares, min_deficit, max_deficit, target_quarter, espn_poll_interval_secs

See the inline comments in config/default.toml for a full explanation of every field.

Strategy Runtime Backtests

new-ploy-runner --config config/strategies/*.toml uses the unified strategy-runtime config in crates/ploy-strategy-bundles.

• [reference_data]
  • pyth_symbols = [...] configures additive non-crypto reference capture.
  • capture_sports_state = true|false toggles live sports-state recording.
• [backtest_data]
  • include_reference_prices = true|false opts historical DB backtests into reference_price_ticks.
  • include_sports_state = true|false opts historical DB backtests into sports_state_events.
  • reference_symbols = [...] overrides the symbols loaded from reference_price_ticks; when omitted, the loader falls back to [reference_data].pyth_symbols.

Historical trust remains conservative in this phase:

• PM quotes stay pinned to the existing trusted source list in clob_quote_ticks.
• Official settlement truth remains pm_token_settlements.
• Reference prices and sports-state are additive observational inputs until a later trust-cutover phase blesses broader historical defaults.

Usage

Live Trading (Recommended)

Ploy uses a Coordinator-only live execution plane. For live orders, use the platform entry point:

Historical reference only: the old ploy platform start ... path has been retired from this branch. Use the ployd / ployctl workspace runbook instead.

Legacy commands that can place orders (example: ploy run, ploy momentum, ploy split-arb, ploy crypto split-arb, ploy sports split-arb, ploy event-edge --trade, ploy agent --enable-trading) are blocked for live execution by default.

If you need an explicit override (not recommended), set:

export PLOY_ALLOW_DIRECT_LIVE=true

Global Flags

--dry-run  / -d    Override dry-run mode (no real orders)
--market   / -m    Override market slug from config
--config   / -c    Config file path (default: config/default.toml)

Current Operator Commands

Use the workspace control-plane binaries as the active operator surface:

cargo run -p new-ployd
cargo run -p ployctl -- system status
cargo run -p ployctl -- deployments list
cargo run -p ployctl -- trading status
cargo run -p new-ploy-runner --features full -- run --config config/strategies/02-pm5d.unified.toml --dry-run

Older ploy run, ploy serve, ploy dashboard, ploy search, ploy book, ploy current, ploy watch, ploy account, ploy claim, ploy history, and ploy ev examples are archived compatibility references. Do not use them as the default runtime, research, or deployment path for new work.

Collector And Backfill

Current data work is split between CI-managed collectors and artifact-backed research workflows. Treat older ploy collect, ploy orderbook-history, and ploy strategy backfill-* references as archived compatibility paths.

• .github/workflows/deploy-tango-1-1.yml deploys the CI-built /opt/ploy/bin/ploy-runner and collector systemd units on tango-1-1.
• /opt/ploy/bin/ploy-runner collect-* commands are foreground operator diagnostics on the data host, not promotion evidence.
• .github/workflows/research-snapshot.yml creates retained sampled research artifacts.
• factor-review-v2-hosted-artifact.yml, factor-walk-forward-v2-hosted-artifact.yml, and runtime-candidate-replay.yml produce attribution, walk-forward, and runtime replay evidence from retained artifacts.

See docs/COLLECTOR_RUNBOOK.md for examples and workflow guidance.

Strategies

Strategy work should flow through reviewed strategy configs, new-ploy-runner, and the artifact-backed research workflows. Treat older direct strategy commands such as ploy trade, ploy momentum, ploy split-arb, ploy market-make, ploy scan, ploy analyze, and ploy paper as archived compatibility examples, not deployable evidence.

Post-settlement claiming is not handled by an in-process Ploy claimer daemon on the current workspace path. Use Polymarket account auto-claim for normal settlement, and keep only the official Relayer API credentials in the runtime secret store for gasless account operations:

export RELAYER_API_KEY=xxx
export RELAYER_API_KEY_ADDRESS=0x...

Do not reintroduce CLAIMER_* or POLY_BUILDER_* env vars unless a new account-ops capability is added and verified. The retired ploy-claimer crate and live-runner daemon path must remain out of the default runtime.

Example: split 100u capital into crypto/sports 50/50 and hard-stop each domain at 45u daily loss:

export PLOY_RISK__CRYPTO_ALLOCATION_PCT=0.5
export PLOY_RISK__SPORTS_ALLOCATION_PCT=0.5
export PLOY_RISK__CRYPTO_DAILY_LOSS_LIMIT_USD=45
export PLOY_RISK__SPORTS_DAILY_LOSS_LIMIT_USD=45

Archived Event-Edge And Agent CLI

The old ploy event-edge ... and ploy agent ... commands are archived compatibility examples. Do not use them as promotion evidence, dry-run handoff, or live execution paths. New strategy research should create artifact-backed evidence through the workflows in docs/runbooks/strategy-research-cicd.md.

Domain: Crypto

# Archived compatibility examples only:
# ploy crypto split-arb --coins SOL,ETH,BTC --dry-run
# ploy crypto monitor --coins SOL,ETH

Domain: Sports

# Archived compatibility examples only:
# ploy sports split-arb --leagues NBA --dry-run
# ploy sports monitor --leagues NBA
# ploy sports draftkings --sport nba --min-edge 5
# ploy sports analyze --team1 LAL --team2 BOS
# ploy sports polymarket --league nba --live
# ploy sports chain --team1 LAL --team2 BOS
# ploy sports live-scan --sport nba --min-edge 3

Strategy Management

cargo run -p ployctl -- deployments list
cargo run -p ployctl -- deployments inspect <deployment-id>
cargo run -p ployctl -- deployments pause <deployment-id>
cargo run -p ployctl -- deployments resume <deployment-id>
cargo run -p ployctl -- deployments stop <deployment-id>

Archived Single-Binary Platform CLI

The old ploy platform start ... commands are no longer runnable in this branch. Use docs/runbooks/platform-startup.md for the workspace daemon/client flow.

Operator Terminal

Use ployctl for scripted operator actions and ploytui for the terminal console. Both are clients of the new-ployd control plane and do not introduce a direct live order path.

Start the control plane and terminal console:

export PLOY_API_ADMIN_TOKEN=change-me
cargo run -p new-ployd
cargo run -p ploytui

Behavior:

• ployctl and ploytui connect to http://127.0.0.1:${PLOY_API_PORT:-8081}.
• operator requests use x-ploy-admin-token.
• all operator actions flow through the workspace control plane.

If the admin token is missing, protected operator requests fail closed. See docs/runbooks/operator-terminal.md for the minimal operator flow.

Deployment matrix entries support runtime scope controls:

{
  "id": "crypto-momentum-5m",
  "strategy": "momentum",
  "domain": "Crypto",
  "enabled": true,
  "account_ids": ["acct-main", "acct-paper"],
  "execution_mode": "any"
}

• account_ids: optional allow-list. Empty means all accounts.
• execution_mode: any | dry_run_only | live_only.

The older ploy rpc and sidecar evidence endpoints are archived compatibility references only. They are not part of the default workspace operator surface on this branch.

RL Commands (requires --features rl)

ploy rl train --episodes 1000 --series 10423        # Train RL model
ploy rl run --model ./models/best --series 10423     # Live trading with RL
ploy rl eval --model ./models/best --data test.csv   # Evaluate model
ploy rl info --model ./models/best                   # Inspect model stats
ploy rl export --model ./models/best -o model.onnx   # Export for deployment
ploy rl backtest --episodes 100                      # Backtest on sample data
ploy rl lead-lag --episodes 1000 --symbol BTCUSDT    # Train lead-lag RL
ploy rl lead-lag-live --symbol BTCUSDT --market btc-price-series-15m  # Live lead-lag
ploy rl agent --symbol BTCUSDT --market btc-price-series-15m \
    --up-token <id> --down-token <id>                # Full RL agent integration

Data Collection

gh workflow run research-snapshot.yml \
  -f git_ref=main \
  -f start_date=2026-05-16 \
  -f end_date=2026-05-18 \
  -f symbols=BTCUSDT,ETHUSDT,SOLUSDT

/opt/ploy/bin/ploy-runner check-db --db-url "$PLOY_DATABASE__URL"

Architecture

Ploy is organized around a multi-domain platform where each prediction market category (currently crypto and sports/NBA) has a dedicated trading agent. The agents submit orders through a central coordinator that applies risk checks, queues orders, and dispatches them to the Polymarket CLOB via authenticated API calls.

Strategies run independently and can be managed as daemons (start/stop/status). The event registry continuously discovers new markets, scores them for edge, and promotes them through a funnel from discovery to active trading. Persistence is handled by PostgreSQL with an event store for auditability, a checkpoint system for crash recovery, and a dead-letter queue for failed operations.

apps/
  new-ployd/     Next-generation daemon entrypoint
  new-ploy-runner/ Next-generation runner entrypoint
  ployctl/       Operator client entrypoint
  ploytui/       Thin terminal operator console
crates/
  ploy-daemon-host/ Daemon host/bootstrap crate
  ploy-runner-host/ Runner CLI host crate
  ploy-control-client/ Shared operator client transport
  ploy-market-data/ Collector/feed/scanner/discovery
  ploy-platform/ Control-plane core
  ploy-platform-runtime/ Runtime orchestration ownership
  ploy-trading/  Canonical trading lifecycle
  ploy-deployments/ Deployment worker protocol + supervisor
  ploy-operator-contracts/ Shared API and event contracts
  ploy-strategy-runtime/ Strategy runtime ownership
  ploy-strategy-bundles/ Signal-to-intent runtime
  ploy-research/ Replay and backtest consumers
ploy-frontend/   Web operator console
ploy-sidecar/    Agent-facing sidecar client
config/          Platform and deployment configuration
docs/            Design docs, runbooks, and migration notes

Development

cargo run -p new-ployd               # Boot the platform daemon
cargo run -p ployctl -- system status
cargo run -p ployctl -- trading status
cargo run -p ployctl -- deployments apply config/deployments/example.paper.json
cargo run -p ployctl -- deployments list
cargo run -p ployctl -- deployments inspect example.paper
cargo run -p ploytui
cargo run -p new-ploy-runner --features full -- run --config config/strategies/02-pm5d.unified.toml --dry-run
curl -N http://127.0.0.1:8081/api/events/stream
rtk cargo check -p new-ployd         # Fast daemon type-check loop
rtk cargo check -p new-ploy-runner   # Fast runner type-check loop
rtk cargo check -p ployctl           # Fast client type-check loop
rtk cargo check -p ploytui           # Fast terminal console type-check loop
rtk cargo test --test platform_smoke platform_smoke_registers_and_starts_one_deployment -- --nocapture
cargo fmt --check                    # Check formatting
cargo clippy -- -D warnings          # Lint
rtk cargo build -p new-ployd         # Build the daemon binary
rtk cargo build -p new-ploy-runner --features full   # Build the full runner binary
rtk cargo build -p ployctl           # Build the operator client binary
rtk cargo build -p ploytui           # Build the terminal console binary

See docs/CONTRIBUTING.md for the contributor guide.

License

MIT

Disclaimer

This software is for educational and research purposes only. Trading on prediction markets carries substantial risk of financial loss. Always start with dry_run.enabled = true and verify behavior before committing real funds. Use at your own risk.