Automated daily and weekly options strategies running on AI Velocity Trading
A same-day, market-neutral SPY iron condor that collects premium by betting the market stays within a range. Each Monday through Thursday session, the strategy sells an out-of-the-money put and call and buys further out-of-the-money wings for protection, so the maximum loss is capped on both sides. The short-strike offset and wing width are sized as a percentage of SPY's price and rounded to tradable strikes, so the structure scales with the market. Entry is priced off SPY at 9:40 AM ET. There is no minimum-credit filter: the position opens whenever the live net credit is positive, retrying each minute for up to 20 minutes if quotes are unusable. A $250 minimum tested in backtesting filled too rarely to be practical, so it was removed. All legs expire on the next Friday weekly, but the position is always closed the same day at 3:40 PM ET, regardless of profit or loss, to eliminate overnight and assignment risk. Entry and exit are each submitted as a single four-leg order.
A same-day, market-neutral SPY iron butterfly that collects the richest premium in the four-leg options family. Each Monday through Thursday session, the strategy sells an at-the-money straddle, with the short put and short call at the same strike, and buys equidistant out-of-the-money wings to cap risk on both sides. Maximum profit comes when SPY stays pinned near the entry price. Wing width is sized as a percentage of SPY's price and rounded to tradable strikes. Entry is priced off SPY at 9:40 AM ET, and every trade sells the same Friday weekly expiry, so the required net credit steps down as the week progresses: $650 on Monday, $600 on Tuesday, $550 on Wednesday and $500 on Thursday. If the live credit falls short, entry is retried each minute for up to 20 minutes before the day is skipped. The position is always closed the same day at 3:40 PM ET, regardless of profit or loss, to eliminate overnight and assignment risk. Entry and exit are each submitted as a single four-leg order.
A weekly, market-neutral SPY iron condor that holds a range-bound position for most of the trading week. The strategy enters on Monday (or Tuesday after a Monday holiday), selling an out-of-the-money put and call and buying further out-of-the-money wings so the maximum loss is capped on both sides. The short-strike offset and wing width are sized as a percentage of SPY's price and rounded to tradable strikes. Entry is priced off SPY at 9:40 AM ET, and the condor, expiring that Friday, opens only if the live net credit is at least $200 per contract. If the credit falls short, entry is retried each minute for up to 20 minutes before the week is skipped. If Monday's run fails for a technical reason, Tuesday retries the entry, but a deliberate low-credit skip is not retried. The position is closed on Thursday at 3:40 PM ET, regardless of profit or loss, to avoid Friday expiration and assignment risk. Entry and exit are each submitted as a single four-leg order.
A weekly, market-neutral SPY iron butterfly that collects rich at-the-money premium and holds it for most of the trading week. The strategy enters on Monday (or Tuesday after a Monday holiday), selling an at-the-money straddle, with the short put and short call at the same strike, and buying equidistant out-of-the-money wings to cap risk on both sides. Maximum profit comes when SPY finishes near the entry price. Wing width is sized as a percentage of SPY's price and rounded to tradable strikes. Entry is priced off SPY at 9:40 AM ET, and the butterfly, expiring that Friday, opens only if the live net credit is at least $700 per contract. If the credit falls short, entry is retried each minute for up to 20 minutes before the week is skipped. If Monday's run fails for a technical reason, Tuesday retries the entry, but a deliberate low-credit skip is not retried. While the position is open, it is checked every minute and closed as soon as it is up $150 per contract; otherwise it is closed on Thursday at 3:40 PM ET, to avoid Friday expiration and assignment risk. There is no stop loss. Entry and exit are each submitted as a single four-leg order.
An autonomous options agent from the Alpaca AI Trading Agents Hackathon. It sells defined-risk iron condors on liquid ETFs (SPY, QQQ, DIA, GLD, SLV, TLT, XLE) when implied volatility is rich relative to realized volatility and the trend is weak, then takes profits at 50% of max profit or closes the day before expiry. Every trade must pass a deterministic risk engine before an order is placed.
Weekly Bull Call Spread on XLK is a net-debit options strategy driven by weekly_bull_call_spread_xlk.py, holding one position per week with no directional entry filter. On Monday between 9:40 and 1:00 PM ET, a narrower window than its sibling weekly scripts, it unconditionally buys a call roughly 2% above spot and sells a further call roughly 5% above spot (rounded to $5, next Friday expiry) for a net debit, betting on XLK strength. The single weekly trade is held until Thursday at 3:45 PM ET, when it is unconditionally closed, followed by a cleanup pass that force-closes any lingering short call leg. It runs on Alpaca paper accounts resolved from Supabase, with a --dry-run flag for inspection.
Weekly Bull Put Spread on AAPL is a net-credit options strategy driven by weekly_bull_put_spread_aapl.py, holding one position per week with no entry filter beyond the schedule. On Monday between 9:40 AM and 2:30 PM ET, it unconditionally sells a put roughly 2% below spot and buys a further put roughly 5% below spot (rounded to $5, next Friday expiry) for a net credit, betting AAPL holds above the short strike. The single weekly trade is held until Thursday at 3:45 PM ET, when it is unconditionally closed, followed by a cleanup pass that force-closes any lingering short put leg. It runs on Alpaca paper accounts resolved from Supabase, with a --dry-run flag for inspection.
Daily Bull Call Spread on XLK is a net-debit, defined-risk, single-day options strategy driven by daily_bull_call_spread_xlk.py. Unlike its MU/AMZN daily siblings, it applies no directional entry filter, every trading day at 9:40 AM ET it simply buys a call roughly 2% above spot and sells a further call roughly 5% above spot (both rounded to $5, next Friday expiry) for a net debit, betting XLK will rise. The spread is opened and closed within the same session, unwound at 3:45 PM ET regardless of P&L, with a post-close cleanup routine that force-closes any lingering short call leg. Trading defaults to Alpaca paper accounts resolved from Supabase per-account credentials, and supports a --dry-run CLI flag plus --test-trade/--test-csv diagnostics modes.
Daily Bull Put Spread on AAPL is a net-credit, defined-risk, single-day options strategy driven by daily_bull_put_spread_aapl.py. Each trading day around 9:40 AM ET it computes AAPL's 14-day Wilder RSI from daily bars and skips the trade if RSI is below 45 (bearish momentum), since this is a bullish-biased position; otherwise it sells a put roughly 2% below spot and buys a further put roughly 5% below spot (both rounded to $5, next Friday expiry) for a net credit. The spread is opened and closed the same day, unwound at 3:45 PM ET regardless of outcome, followed by a cleanup routine that force-closes any residual short put leg. Trading defaults to Alpaca paper accounts loaded from Supabase, with a --dry-run CLI flag for safe testing.
Weekly Bear Call Spread on MU is a net-credit, defined-risk options strategy driven by weekly_bear_call_spread_mu.py, holding a single position across the week rather than intraday. On Monday between 9:40 and 2:30 PM ET, it checks two volatility filters, 20-day annualized historical volatility under 150% and the prior 5-day high-low range under 12% of the previous close, and if both pass, sells a call at roughly 3% above spot while buying a further call at roughly 6% above spot (rounded to $5, next Friday expiry) for a net credit. The one trade for the week is then held until Thursday at 3:45 PM ET, when it is unconditionally closed. The script's own backtest notes cite 36 trades, an 86.1% win rate and a 2.00 Sharpe over a 2024-2026 sample. It runs on Alpaca paper accounts loaded from Supabase, with a --dry-run flag for inspection.
Daily Bear Put Spread on AMZN is a net-debit, defined-risk, single-day options strategy driven by daily_bear_put_spread_amzn.py. Each trading day around 9:40 AM ET it applies the same bearish-signal stack as the MU bear call variant, prior day down, no gap up, price below its 10-day SMA, and intraday rise under 0.3% since the 9:30 open, and only then buys a put roughly 2% below spot while selling a further put roughly 5% below spot (both rounded to $5, next Friday expiry) for a net debit. The spread is opened and closed the same day, unwound at 3:45 PM ET regardless of outcome, with a post-close routine that force-closes any lingering short put leg via limit-then-market buy orders. Trading defaults to Alpaca paper accounts loaded from Supabase, with a --dry-run flag for safe testing.
Daily Bear Call Spread on MU is a net-credit, defined-risk, single-day options strategy driven by daily_bear_call_spread_mu.py. Every trading day around 9:40 AM ET it checks a stack of bearish filters, prior day closed down, no gap up at the open, price below its 10-day SMA, and less than 0.3% intraday rise since 9:30, and only then sells a call roughly 2% above spot while buying a further call roughly 5% above spot (both rounded to $5, next Friday's expiry) for a net credit, skipping the trade if the estimated credit is under $0.50. The position is opened and closed within the same session, unwound by 3:45 PM ET regardless of P&L. Trading is paper-only by default, with a Supabase-driven multi-account Alpaca OAuth setup and a --dry-run flag for safe testing.
Weekly Bear Put Spread on AMZN is a net-debit options strategy driven by weekly_bear_put_spread_amzn.py, holding one position per week instead of closing intraday. On Monday between 9:40 and 2:30 PM ET, it requires a bearish-signal stack, prior day closed down, no gap up at entry, and price below its 10-day SMA, before buying a put roughly 2% below spot and selling a further put roughly 5% below spot (rounded to $5, next Friday expiry) for a net debit. The single weekly trade is held until Thursday at 3:45 PM ET, when it is unconditionally closed, followed by a cleanup pass that force-closes any lingering short put leg. It runs on Alpaca paper accounts resolved from Supabase, with a --dry-run flag for inspection.
Long-only intraday mean-reversion bot trading TSLA equity shares via Alpaca, polling every 60 seconds between 4:00 AM and 8:00 PM ET on trading days, auto-resuming after off-hours sleep. It computes RSI(14), MACD(12/26/9), 20-period/2-sigma Bollinger Bands, and a 20-bar volume MA (with a relaxed 0.5x volume threshold reflecting TSLA's higher volatility) on 1-minute bars. Entries require RSI < 35 plus a confirmation score >= 1.5 (MACD rising, price at/below the lower or mid Bollinger Band, sufficient volume). Exits trigger on any of: a 0.5% stop-loss below entry, RSI > 55, price at/above the upper Bollinger Band, or MACD histogram fading while RSI remains above 55. Multiple Alpaca accounts are supported via Supabase's user_strategies table, each with its own configured share quantity, synced periodically and traded off a shared market-data snapshot each cycle.
Long-only intraday mean-reversion bot trading QQQ equity shares via Alpaca (paper/live), polling every 60 seconds between 4:00 AM and 8:00 PM ET on trading days (auto-detected via the Alpaca market calendar, sleeping until next open). It computes RSI(14), MACD(12/26/9), 20-period/2-sigma Bollinger Bands, and a 20-bar volume MA on 1-minute bars. Entries require RSI < 25 plus a confirmation score >= 1.5 (MACD histogram rising = 1, price at/below the lower Bollinger Band = 1 or at/below the mid band = 0.5, volume >= 0.8x its average = 1). Exits trigger on any of: a 0.2% stop-loss below entry, RSI > 60, price at/above the upper Bollinger Band, or MACD histogram fading while RSI > 55. Share quantity per account is pulled from Supabase's user_strategies table rather than a fixed dollar amount, and multiple accounts are synced and traded each cycle using shared market data.
Long-only intraday mean-reversion bot trading SPY equity shares via Alpaca, polling every 60 seconds between 4:00 AM and 8:00 PM ET on trading days, exiting at session end and resuming via cron/next-day restart. It computes RSI(14), MACD(12/26/9), 20-period/2-sigma Bollinger Bands, and a 20-bar volume MA on 1-minute bars. Entries require deep oversold RSI < 25 plus at least one confirming signal (MACD histogram rising, price at/below the lower Bollinger Band, or volume >= 0.8x its average). Exits fire on any of: RSI > 60, price at/above the upper Bollinger Band, MACD histogram simply turning down (no RSI condition required, unlike the QQQ/TSLA variants), or a tight 0.2% stop-loss below entry. Multiple Alpaca accounts are supported, each trading its own fixed share quantity sourced from Supabase's user_strategies table, with a legacy single-account .env-file credential loader also present but superseded by the Supabase-driven multi-account flow.
An evidence-gated ETF strategy from the Alpaca AI Trading Agents Hackathon. It holds SPY overnight only when the 12-month trend is up and credit is healthy (HYG above its 100-day average), buys capitulation selloffs in a basket of ETFs for a 3-session hold, and parks idle cash in T-bills. Every rule was validated on 33 years of data before it was allowed to trade.
Top Gainers Morning Short bets that the day's biggest gainers give back part of their move. At 09:00 ET it takes the top gainers from the MachineTrader top-stocks list, keeps those that Alpaca reports as shortable and easy to borrow, and shorts $1,000 of each at the SIP bid, skipping any with a spread over 5%. It covers every position at 11:00 ET, with no stops. Stocks under $1, moves over 300%, warrants, units, rights and frozen quotes are excluded. In a backtest of 8 sessions (Oct 1 - Oct 12, 2026) it made $951 on 36 trades with a 75% win rate, but it was the best of 31 entry-time and holding-period combinations tested on the same days, so it now trades forward, in paper account PA36VG1H5VKQ since Oct 23, 2026, to see whether the result holds.
SPY Premarket Mean Reversion fades sharp SPY moves in the premarket session (04:00-09:30 ET), when liquidity is thin and prices overshoot. On 1-minute bars it buys $10,000 of SPY when the price falls more than 3 standard deviations away from its 20-bar average, exits as soon as the price is back at the average or 0.5% against the entry, and is always flat by 09:30. Parameters were chosen on 2025 data only. In the backtest (Jan 1, 2025 - Oct 21, 2026, spread costs from real quotes included) it made $1,446 on 343 trades (65% winners); $816 in 2025 and $630 in the 2026 out-of-sample period. The same rules lost money in the regular, after-hours and overnight sessions. Paper trading in PA39MXU6BZPA.
QQQ Premarket Mean Reversion fades sharp QQQ moves in the premarket session (04:00-09:30 ET), when liquidity is thin and prices overshoot. On 1-minute bars it buys or sells short $10,000 of QQQ when the price moves more than 3 standard deviations away from its 20-bar average, exits as soon as the price is back at the average or 0.5% against the entry, and is always flat by 09:30. Parameters were chosen on 2025 data only. In the backtest (Jan 1, 2025 - Oct 21, 2026, spread costs from real quotes included) it made $2,439 on 676 trades (61% winners); $406 in 2025 and $2,033 in the 2026 out-of-sample period. The same rules lost money in the regular, after-hours and overnight sessions. Paper trading in PA39MXU6BZPA.
TSLA Premarket Mean Reversion fades sharp TSLA moves in the premarket session (04:00-09:30 ET), when liquidity is thin and prices overshoot. On 1-minute bars it buys or sells short $10,000 of TSLA when the price moves more than 3 standard deviations away from its 20-bar average, exits as soon as the price is back at the average or 2.0% against the entry, and is always flat by 09:30. Parameters were chosen on 2025 data only. In the backtest (Jan 1, 2025 - Oct 21, 2026, spread costs from real quotes included) it made $2,743 on 991 trades (54% winners); $954 in 2025 and $1,789 in the 2026 out-of-sample period. The same rules lost money in the regular, after-hours and overnight sessions. Paper trading in PA39MXU6BZPA.
QQQ Premarket Oversold Bounce buys QQQ after sharp pre-market sell-offs and sells into the rebound, usually after the regular session opens. On 15-minute bars it waits for RSI(14) below 25 with confirming MACD, Bollinger Band and volume signals, then buys 100 shares on the first bar that closes higher, as long as that bar closes between 04:00 and 09:30 ET. It sells when RSI rises above 60, the price reaches the upper Bollinger Band, or MACD momentum fades, holding overnight if needed, with no stop. The rules came out of a study of strategy 1026 and were chosen on 2016-2022 data only. In a backtest from Jan 2016 to Oct 2026, with fill costs calibrated to real quotes, it made $27,078 on 299 trades (77% winners), including $18,099 in the 2023-2026 out-of-sample period. It trades about 28 times a year. Paper trading in PA3PJANDI40H since Oct 29, 2026.
An inverse-signal equity trader, and the shared implementation module for the CNBC-signals pair, that reads today's ticker mentions from a local SQLite database (joined to an episodes table by publication date, with a fallback to created_at) and fades each signal: a BUY/BULLISH mention is shorted, a SELL/BEARISH mention is bought long, via Alpaca limit orders. It defines the reusable signal-lookup, quote-fetching, and data-key helpers also imported by cnbc_signals_follow_cramer.py. For every active account in Supabase's user_strategies table (strategy_id 1027), it submits a fixed-share-quantity extended-hours DAY limit order at the SIP ask (for buys) or bid (for shorts), confirms the order status, and logs it. Like its follow counterpart, it is a one-shot script with no stop-loss, exit, or position-monitoring logic; risk is bounded only by the fixed per-account share size.
A direct-signal equity trader that reads a shared local SQLite database of ticker mentions (joined to an episodes table by publication date, falling back to created_at if no episode matches today) and follows each signal literally: a BUY/BULLISH action opens a long position, a SELL/BEARISH action opens a short, via Alpaca limit orders. Signals are deduplicated per ticker (highest conviction kept) and any other action value is skipped. It reuses helper functions imported from the companion inverse-signal script. For every active Alpaca account configured in Supabase's user_strategies table (strategy_id 1028), it submits a fixed-share-quantity (not dollar-sized) extended-hours DAY limit order, buy at the current ask, short-sell at the current bid, using SIP quotes, then logs an order confirmation. It runs as a one-shot script (not a polling loop) and has no built-in stop-loss or exit logic.
A once-daily equity strategy (intended to run around 3:45 PM ET, before the market-on-close cutoff) that trades around earnings announcements by detecting unusual pre-earnings price divergence. Phase 1 exits: any held position whose tracked earnings date has arrived is sold market-on-close, and any position with an unrealized loss >=30% is sold immediately at market. Phase 2 entries: it pulls tomorrow's NASDAQ earnings-calendar tickers, compares each stock's intraday return today against SPY's, and, long-only, price >= $5, skipping tickers already held, buys (market-on-close) the tickers whose divergence from SPY is >=5%, ranked by divergence magnitude, until a per-account $60,000 gross-exposure cap is reached. Position state and trade history are persisted per Alpaca account in local JSON/CSV files, with quantities sized from Supabase's user_strategies.shares per account.
The winning entry of the Alpaca AI Trading Agents Hackathon. An AI agent buys short-dated calls and puts (2-14 days, 0.35-0.55 delta) on SPY, QQQ, AAPL, MSFT and NVDA when it has a clear directional thesis, while hard-coded exits do the rest: a 40% stop-loss, a 60% take-profit and a forced close on expiry day. Positions are capped at $5,000 and four at a time, and a 2% daily loss halts trading.
Third place in the Alpaca AI Trading Agents Hackathon. Every 15 minutes Claude reviews 15 ETFs and stocks and proposes a direction; strict code gates decide what trades. It buys 30-45-day calls or puts near 0.65 delta, sized at 4% of equity, and exits at a 25% loss, a 50% gain, a 2x ATR move against it, or 7 days before expiry.
A finalist in the Alpaca AI Trading Agents Hackathon. A multi-agent options desk reads the volatility regime on SPY, QQQ, IWM and five large-cap stocks: when options are expensive it sells credit spreads or iron condors, when they are cheap it buys debit spreads, and otherwise it stands aside. Claude picks the trade, a Monte Carlo audit and twelve risk breakers must approve it, and every position is a defined-risk spread capped at 2% of equity.
A finalist in the Alpaca AI Trading Agents Hackathon. For seven large-cap stocks, Claude agents weigh news, industry, macro and market reaction, a decision agent proposes a long call, long put or debit spread, and a risk agent argues against it before deterministic rules decide. Each trade risks at most 1% of equity, and positions exit on a 50% stop, a trailing profit stop, or within four trading days.
The runner-up in the Alpaca AI Trading Agents Hackathon. It scans 29 cryptocurrencies around the clock and 10 large-cap stocks during market hours, scores each on momentum, trend, volume, volatility, liquidity, catalysts, reward-to-risk and regime, and buys only the strongest setups. A Claude review can veto any buy. Positions are capped at $5,000 and five at a time, with a 5% stop-loss and a 4-5% profit target.
BOATS Premarket Momentum trades MU and NVDA in the direction of their overnight move. At 04:00 ET it compares each stock's last trade on the Blue Ocean ATS overnight session (BOATS, 20:00-04:00 ET) with the prior regular-session close: if the stock traded up overnight it buys 1 share, and if it traded down it shorts 1 share. It closes the position at 09:29 ET, one minute before the open, so it never holds through the regular session. The rule comes from research showing the overnight move continues into the premarket (correlation 0.52 for MU, p < 0.001) but has no predictive value after the open. In a replay of Sep 1 - Oct 19, 2026, it averaged +33 bp per trade before costs with a 63% win rate. The results are in-sample, and this strategy runs as a paper forward test to see whether the edge survives premarket bid/ask spreads.
Top Gainers Premarket Long buys the day's biggest gainers when the premarket opens and sells them before the regular session. At 04:00 ET it takes the top 50 US gainers by % change from the previous close, from the MachineTrader top-stocks list (previous close at least $1.00, at least 100,000 shares of daily volume, no warrants). It buys $1,000 of each stock that has a fresh quote with a spread of 5% or less, using a limit order at the bid/ask midpoint, and exits at 09:30 ET. The bet is that overnight momentum keeps going through the premarket. In 8 dry-run sessions (Sep 24 - Oct 13, 2026) it lost $2,276.66 on 152 trades, with a 39% win rate, so it runs as a test, not a proven edge.