algo-strategy

Create a complete dual-mode strategy file from a template.

Arguments

Parse $ARGUMENTS as: template symbol exchange interval

• $0 = template (e.g. ema-crossover, rsi, supertrend, donchian, macd, opening-range, atr-breakout, bb-squeeze, ml-logistic, ml-xgb, pairs-trading, regime-trend, event-driven)
• $1 = symbol (e.g. SBIN, RELIANCE, NIFTY). Default: template's catalog default
• $2 = exchange (e.g. NSE, NSE_INDEX, NFO). Default: template's catalog default
• $3 = interval (e.g. 1m, 5m, 15m, D). Default: template's default

If no arguments, ask the user which template they want. See algo-expert/rules/strategy-catalog.md.

Required prompts before generation

After parsing arguments, ASK the user (do not assume defaults silently):

1. Indicator library: (1) openalgo [default] (2) talib

   • Default openalgo. talib's user must confirm explicitly.
   • Specialty indicators (Supertrend, Donchian, Ichimoku, HMA, KAMA) always use openalgo regardless.

2. Execution type - default depends on template (see strategy-catalog.md):

   • (1) end-of-candle (eoc) - signal at bar close, MARKET on next bar (default for most)
   • (2) real-time limit (limit) - tick-driven LIMIT placement (default for atr-breakout)
   • (3) stop-trigger (stop) - broker-side SL/SL-M (default for opening-range)

   Tell the user the recommended default for their template and let them override.

Instructions

1. Read algo-expert/rules/unified-strategy-pattern.md, mode-toggle.md, and execution-types.md.
2. Read the matching template at algo-expert/rules/assets/<template>/strategy.py as the starting point.
3. Create strategies/<template>_<symbol>/ directory if not present.
4. Copy the template into strategies/<template>_<symbol>/strategy.py and:
   • Replace SYMBOL, EXCHANGE, INTERVAL constants with user values
   • Set INDICATOR_LIB to user's choice ("openalgo" or "talib")
   • Set EXECUTION_TYPE to user's choice ("eoc", "limit", "stop")
   • Update STRATEGY_NAME to <template>_<symbol> (default; can be overridden by env)
   • Adjust PRODUCT to match the asset class (NSE equity → MIS or CNC; NFO → NRML)
   • Adjust QUANTITY to a reasonable starting value (1 for equity, lot size for futures)
5. For ML templates (ml-logistic, ml-xgb), also copy train.py and remind the user to run it BEFORE running the strategy:

   python strategies/ml_logistic_RELIANCE/train.py
   

6. The file must be host-compatible per self-hosted-strategies.md:
   • Reads HOST_SERVER first, then OPENALGO_HOST fallback
   • Reads OPENALGO_STRATEGY_EXCHANGE for exchange
   • SIGTERM/SIGINT handlers installed
   • stdout-only logging
   • dispatcher reads env MODE if no --mode CLI arg
7. Print:
   • File location
   • Next steps:
     • Backtest: python strategies/<name>/strategy.py --mode backtest
     • Live: python strategies/<name>/strategy.py --mode live
     • Upload to /python: /algo-host <name> to validate and generate upload guide

Available templates

Template	Default symbol	Default exchange	Default interval	Default execution
ema-crossover	SBIN	NSE	5m	eoc
rsi	RELIANCE	NSE	15m	eoc
supertrend	NIFTY	NSE_INDEX	5m	eoc
donchian	NIFTY	NSE_INDEX	D	eoc
macd	INFY	NSE	D	eoc
opening-range	SBIN	NSE	5m	stop
atr-breakout	RELIANCE	NSE	5m	limit
bb-squeeze	TCS	NSE	15m	eoc
ml-logistic	RELIANCE	NSE	15m	eoc
ml-xgb	RELIANCE	NSE	15m	eoc
pairs-trading	SBIN/PNB	NSE	D	eoc (two-leg)
regime-trend	RELIANCE	NSE	D	eoc (ADX+VIX+volume gates)
event-driven	RELIANCE	NSE	D	scheduled-time

For options templates (short-straddle, iron-condor), use /algo-options instead.

Costs

The file's COSTS = cost_lookup(PRODUCT, EXCHANGE) auto-resolves:

• MIS + NSE → 0.0225% + Rs 20 + 5 bps slippage
• CNC + NSE → 0.111% + Rs 20 + 3 bps slippage
• NRML + NFO → 0.018% + Rs 20 + 2 bps slippage (futures)
• NRML + NFO with options → 0.098% + Rs 20 + 10 bps slippage

If the user wants different broker rates, point them to algo-expert/rules/transaction-costs.md to override the constants.

Risk defaults

The template comes with reasonable RISK values (see strategy-catalog.md). Tell the user these are starting points, not optimal - tune via backtest before live.

Avoid

• Do not use icons/emojis in code, logger output, or skill text
• Do not generate files in arbitrary paths - always under strategies/<name>/
• Do not auto-flip MODE=live without the user explicitly running it