StratDeck CLI (v0.3 – Live Scout)

StratDeck is now a pure Python CLI that can scan candidates, run compliance, log paper fills, and summarize your journal. Mock mode remains the default; flip STRATDECK_DATA_MODE=live to pull tastytrade chains, quotes, balances, and positions.

Quickstart

python3 -m venv .venv
source .venv/bin/activate
pip install -e .  # or pip install -r requirements.txt if you prefer

# copy env template and edit credentials / mode
cp .env.example .env
# set STRATDECK_DATA_MODE=mock to stay fully local

# From this point onward we assume `python` points to your Python 3 interpreter.
# If your shell lacks a `python` binary, keep using `python3` (e.g. `python3 -m stratdeck.cli scan --top 5`).
python -m stratdeck.cli scan --top 5

Environment

.env is auto-loaded by stratdeck/__init__.py. Relevant keys:

Key	Description
STRATDECK_DATA_MODE	mock (default) or live
TASTY_USER / TASTY_PASS	tastytrade username/password (live mode)
TT_USERNAME / TT_PASSWORD	optional aliases
TASTY_ACCOUNT_ID	optional; autodetected when omitted

Mock mode uses the offline heuristics + CSV ledgers in stratdeck/data/. Live mode logs into tastytrade (session token stored in-memory only) and fetches nested chains, option quotes (with greeks), balances, and positions.

CLI Cheat Sheet

Command	Purpose
python -m stratdeck.cli scan --top N	Rank candidates (now backed by live chains when enabled)
python -m stratdeck.cli enter --pick i --qty q [--confirm] [--live-order]	Run compliance on the i‑th scan result; --confirm logs a paper fill, --live-order calls tastytrade preview/place (still optional)
python -m stratdeck.cli positions	List paper ledger entries from data/positions.csv
python -m stratdeck.cli close --position-id ID --exit-credit X	Close a paper trade, compute realized P/L, journal it
python -m stratdeck.cli report --daily	Summarize opens/closes, win rate, realized P/L, and live balances
python -m stratdeck.cli doctor	Run diagnostics (env, provider reachability, config files)
python -m stratdeck.cli chartist -s SYMBOL [--json-output]	Run ChartistAgent technical analysis for one or more symbols and emit either a fallback summary or JSON TA_RESULT maps
python -m stratdeck.cli scan-ta [--json-output]	Run Scout → Chartist to score candidates, print TA-enriched data, or dump the JSON for automation
python -m stratdeck.cli trade-ideas [--json-output]	Scout → Chartist → TradePlanner pipeline that outputs structured trade ideas with legs, rationale, and vol/trend context

enter --confirm prints both the broker ticket id (from the simulated fill) and the ledger position id recorded in positions.csv. Use that ID with the close command.

Chartist TA & trade idea pipeline

ChartistAgent now sits alongside the existing Scout/Trader stack. It wraps ChartistEngine (stratdeck/tools/ta.py), a deterministic technical-analysis core that classifies trend/volatility/regime, spots structure and simple patterns, computes TA scores, and returns a TAResult rich with trend_regime, vol_regime, momentum, structure, scores, and options_guidance. ChartistAgent can run with an optional LLM client and uses the prompts in stratdeck/conf/prompts/chartist_system.md and chartist_report.md to drive human summaries; if no LLM is provided it falls back to the built-in plain-text summary you see on the chartist command.

ChartistEngine is data-agnostic: it consumes data_client objects when supplied, defaults to synthetic mock candles (STRATDECK_DATA_MODE=mock), and, when running in live mode, will try to pull OHLCV from yfinance (install yfinance if you want live-chart candles without wiring a custom client). Missing data or yfinance failures gracefully revert to synthetic candles with a warning so the CLI remains runnable.

The new CLI helpers bridge Scout → Chartist → TradePlanner:

• python -m stratdeck.cli chartist -s SPX -s XSP runs ChartistAgent for each symbol, optionally accepts --strategy-hint, --timeframe, --lookback-bars, and emits either the fallback summary or the raw TA_RESULT JSON when --json-output is set.
• python -m stratdeck.cli scan-ta runs ScoutAgent and feeds the candidate list through ChartistAgent to produce TA-enriched rows (ta_score, ta_directional_bias, ta_vol_bias, support/resistance, scoring metadata). By default it prints a simple table; add --json-output to capture the entire enriched payload for other tooling.
• python -m stratdeck.cli trade-ideas pipes Scout → Chartist → TradePlanner (stratdeck/agents/trade_planner.py). TradePlanner converts the TA context into TradeIdea structs with legs, rationale, notes, and underlying hints. The command prints human-readable trade ideas but you can rerun it with --json-output (optionally appending a path to persist the JSON, e.g. python -m stratdeck.cli trade-ideas --json-output ./ideas.json) to consume the structured output elsewhere (e.g., TraderAgent, journaling scripts, or a RiskAgent).

The Chartist prompts in stratdeck/conf/prompts/chartist_system.md and chartist_report.md are the place to tweak how an LLM interprets the deterministic TA data, while the TAResult/TradeIdea JSONs remain machine-friendly for automation.

Live Mode Notes

• Credentials are posted to https://api.tastyworks.com/sessions; the bearer token lives only in memory. Nothing is written to disk except the existing paper-ledger CSVs.
• Chains come from /option-chains/<symbol>/nested and option quotes from /market-data/by-type?equity-option=…. We currently limit each scan to 75 strikes near the underlying.
• The scan output now includes the actual short/long strikes, live mid credit, delta, and tastytrade IVR. Compliance checks reuse those strikes when you call enter.
• Order preview/place endpoints are still stubbed; --confirm remains a paper-only simulation.
• macOS + Python 3.9 ships with LibreSSL, so urllib3 emits a warning (“supports OpenSSL 1.1.1+”). It’s harmless and will disappear once the project moves to a newer Python build.

IV Rank (IVR)

StratDeck uses Tastytrade's /market-metrics endpoint to populate an IV Rank snapshot for the live trading universe.

• Source field: implied-volatility-index-rank from /market-metrics (this is the same value that backs the Tasty watchlist “IV Rank” column).
• StratDeck normalises this to a 0–1 float called ivr.
• The latest snapshot is stored in stratdeck/data/iv_snapshot.json.
• ivr is then used by the scan cache, trade planner and agents.

To refresh the snapshot with live data:

export STRATDECK_DATA_MODE=live

python -m stratdeck.cli refresh-ivr-snapshot

Recommended cadence: run once per trading day before the market open (and again if you want to pick up major volatility regime changes intraday).

Paper Ledger + Reporting

• stratdeck/data/journal.csv logs OPEN/CLOSE events with JSON metrics (credit, qty, preview stats, realized P/L).
• stratdeck/data/positions.csv tracks OPEN/CLOSED status, exit credit, and P/L per position ID.
• report --daily (or --days N) aggregates those files and augments the summary with live balances when available.

Diagnostics

python -m stratdeck.cli doctor now checks:

• Required folders/files in stratdeck/
• YAML configs load without errors
• When STRATDECK_DATA_MODE=live: env vars are present, tastytrade login succeeds, default account resolves, and a short chain request returns data

If any step fails you’ll see a bullet under “Doctor found issues”. Keep mock mode around for travel days or when the tastytrade API is unavailable.