Methodology

The bot is fully autonomous: nothing is hand-picked per run. Every decision — which stocks, which direction, which strike and expiration, and whether to fire at all — comes out of the fixed rules below. It runs once each trading morning, about 15 minutes after the open, and posts up to three signals to Discord. On most days no stock passes every check, and the bot posts a short “nothing qualified” digest instead of forcing a trade.

The screening rules

1. 1

   Universe

   A seed list of ~180 liquid U.S. stocks, filtered each morning by price range, average share volume, dollar volume, and enough daily range (ATR) that a move can plausibly pay for the premium.

2. 2

   Trend

   For a call: close above the 21-day EMA, which is above the 50-day EMA, which is above the 200-day average, with ADX showing a real trend. Puts are the mirror image.

3. 3

   Momentum

   RSI inside a non-extreme band (50 to 70 for calls) and a positive MACD histogram, so the bot isn't chasing something already stretched or rolling over.

4. 4

   Trigger

   A new 20-day high on relative volume of at least 1.5x its normal level, without price already running too far above the 21-day EMA. An EMA-pullback bounce trigger exists but is off by default because it lost money in testing.

5. 5

   Contract checks

   The live option chain has to pass: open interest of at least 500, bid/ask spread of 10% or less, IV not in the top of the stock's own range (true historical IV isn't available on free data, so current ATM IV is compared to trailing realized volatility), a premium under the per-trade cap, and no earnings report inside the contract's life.

From score to contract

Stocks that pass every rule are ranked by a 0–100 setup score built from trend strength, relative volume, relative strength versus SPY, and setup quality. The score ranks setups against each other; it is not a win probability. For each survivor the bot picks the expiration closest to its preferred days-to-expiry and the strike nearest a target delta (slightly out-of-the-money, for leverage). Target and stop are projected from an ATR multiple of the underlying. Until 7 October 2026 the bot used a five-layer blend (technicals, options flow, news sentiment, an XGBoost model, and market regime) instead; signals published before that date came from it, and the full record of them stays on this site.

How outcomes are graded

A tracking job checks every open signal daily against the underlying’s actual traded high and low (dividend-unadjusted, so a real touch of a level isn’t hidden by a back-adjustment). Any touch of the target at any point during the signal’s life — not just at expiration — counts. The first level touched wins: target → WIN, stop → LOSS. When a single day’s bar spans both levels, the grader drops to hourly bars for that day to see which came first; if hourly data isn’t available (older than ~60 days) it falls back to scoring the day as a loss. The signal’s first day is graded from hourly bars after the 09:45 ET publish time, so price action before the signal existed never counts. A signal that reaches expiration having touched neither level is EXPIRED and reported separately — never folded into the loss column. Win rate = wins ÷ (wins + losses).

The backtest below uses the same intraday first-touch rule, but free historical data has no hourly bars going back years, so its same-day-both-touched case always scores as a loss.

The bot’s source lives in a private repository. This site reads its database and published backtest directly and adds nothing.