Best prompts for VEO3 using stock market analysis

You are VEO3 analyzing daily OHLCV data. For each ticker in the provided CSV (columns: date, ticker, open, high, low, close, volume) compute: SMA_short (default 20), SMA_long (default 50), and RSI (14). Generate a signal per ticker: BUY if SMA_short > SMA_long AND RSI < 60, SELL if SMA_short < SMA_long AND RSI > 40, otherwise HOLD. Include current price, SMA_short, SMA_long, RSI, signal, and a one-sentence rationale. Output a JSON array of objects: {ticker, date, price, SMA_short, SMA_long, RSI, signal, rationale}. Use UTC dates and round numeric outputs to 4 decimals. If a ticker has less than 60 days of data, return "insufficient data" for that ticker.

Backtest the provided trade rule on daily OHLC data (CSV: date,ticker,open,high,low,close,volume) for a single ticker over the given date range. Strategy: enter LONG at next open when signal column equals 'BUY', exit LONG at next open when signal = 'SELL'; ignore additional signals while in position. Parameters: initial capital = 1,000,000 USD, position sizing = entire capital per trade, transaction cost = 0.02% per side, slippage = 0.05% of execution price. Output JSON with: ticker, start_date, end_date, total_trades, win_rate, CAGR, annual_volatility, max_drawdown (peak-to-trough %), Sharpe_ratio (risk-free = 0%), total_return, and a time-series array equity_curve [{date, equity}]. Round numbers to 4 decimals. If no trades, return metrics with zeros and an empty equity_curve.

Given historical daily price CSV for a single ticker and a strategy parameter grid, perform walk-forward optimization. Inputs: train_window_months (e.g., 24), test_window_months (e.g., 6), parameter_grid (JSON array, e.g., [{"SMA_short":10,"SMA_long":50},{"SMA_short":20,"SMA_long":100}]), transaction_cost_per_side (bps). For each rolling fold: (1) find best parameter set on train by CAGR; (2) apply that set to test period to compute test CAGR, volatility, and max_drawdown. Output JSON: folds array with {train_start,train_end,test_start,test_end,best_params,train_CAGR,test_CAGR,test_volatility,test_max_drawdown}, and aggregated metrics: average_test_CAGR, median_test_drawdown, percent_folds_positive. Use monthly alignment and round numbers to 4 decimals.

Optimize portfolio weights given input JSON: {tickers:[], expected_returns:{ticker:annual_return}, cov_matrix:{ticker:{ticker:cov}} , constraints:{max_weight:0.3,min_weight:0.0,turnover_limit:0.2, current_weights:{...}}}. Objective: maximize Sharpe ratio (risk-free = 0%) subject to constraints and weights sum to 1. If turnover_limit provided, minimize turnover while keeping Sharpe within 98% of unconstrained optimum. Output JSON: {optimized_weights:{ticker:weight}, expected_portfolio_return, portfolio_volatility, expected_sharpe, turnover}. Round weights to 4 decimals and metrics to 4 decimals. If optimization fails, return an error field with explanation.

Given daily portfolio returns CSV or price series for each asset with current weights, compute: 1-day and 10-day historical VaR at 95% and 99% (historical simulation), corresponding CVaR (expected shortfall), maximum drawdown and top 5 drawdown events with dates and duration, and daily return volatility. Also compute P&L impact of two stress scenarios: -10% global equity shock and +200 bps rise in rates (apply given interest-rate sensitivity per asset if provided). Output JSON with sections: var_cvar, drawdowns, volatility, stress_results. Round to 4 decimals and express VaR/CVaR as portfolio % loss and absolute USD loss given portfolio_value input.

Given an events table (columns: date,event_type,ticker,actual,consensus,prev) and macro calendar (date,indicator,actual,consensus,prev), produce alert rules that map: earnings surprise > +5% -> 'consider long' for ticker with confidence High; earnings surprise < -5% -> 'consider reduce' with confidence High; CPI MOM > consensus by >=0.2% -> 'reduce cyclical equities' with confidence Medium. For each observed event row, produce an alert object: {date,event_type,ticker_or_indicator,delta_pct,action,confidence,reason, suggested_size_pct_of_portfolio}. Use default suggested sizes: High=5-10%, Medium=2-5%, Low=1-2%. Output JSON array of alerts sorted by confidence then date.