Vocabulary

The full pool of stocks eligible to be screened. [sage]SIGNALS starts with 1,000+ US technology stocks and narrows them using market cap, volume, and analyst consensus before any gate runs.

Market capitalization — the total dollar value of a company's outstanding shares. Calculated as share price × total shares outstanding. The screener requires mid-cap and above (≥ $1B).

A company with a market cap generally between $2B and $10B. The screener's $1B floor includes small-to-mid-cap stocks and above, filtering out micro-caps that are too illiquid or volatile.

The average number of shares (or dollar value) traded per day over a recent period. A liquidity floor — the screener requires ≥ $20M/day to ensure you can enter and exit a position without moving the price.

A rating issued by Wall Street analysts based on their research — typically Strong Buy, Buy, Hold, Sell, or Strong Sell. The screener requires Buy or better consensus before a stock enters the gate stage.

A binary pass/fail filter in the screener pipeline. A stock must pass all 8 gates to advance to ranking. Think of each gate as a question: does this stock qualify on this dimension?

Gate 1. Measures whether a stock's price trend is pointing up across multiple timeframes. The screener checks four conditions: price above SMA50, price above SMA200, positive 1-month return, and positive 3-month return.

Gate 2. Checks that the company is earning money (positive trailing P/E) and growing those earnings year-over-year (positive EPS growth). Weeds out money-losing growth stories.

Gate 3. Requires ≥ 20% 5-year compound annual revenue growth, and that growth is not decelerating quarter-over-quarter. Ensures the business is expanding, not just its stock price.

Gate 4. Checks that the company generates positive free cash flow (FCF) or operating cash flow (OCF). Currently a structural placeholder — Finviz does not expose this data, so the gate always passes as N/A.

Gate 5. Confirms the stock has Buy-or-better analyst consensus. This is actually enforced at the data fetch step (via a Finviz URL filter), so every stock in the universe already passes gate 5.

Gate 6. Would check whether the call/put ratio is ≥ 1.0, indicating bullish options activity. Currently a structural placeholder — Finviz doesn't provide this data, so the gate always passes as N/A.

The ratio of call options volume to put options volume. A ratio above 1.0 means more bullish bets than bearish bets are being placed in the options market — a signal that traders expect the stock to rise.

The average closing price of a stock over the past 50 trading days. Acts as a medium-term trend indicator. Price above SMA50 = the stock is in a short-to-medium uptrend.

The average closing price over the past 200 trading days. The standard long-term trend indicator. Price above SMA200 = the stock is in a long-term uptrend. Falling below it is often seen as a bearish regime change.

Relative Strength Index — a momentum oscillator that measures how fast a stock has been moving up or down, on a scale of 0–100. Above 70 = overbought (may be due for a pullback). Below 30 = oversold (may bounce).

Average True Range — a measure of how much a stock typically moves per day, in dollar terms. High ATR = volatile stock. Used only for the stop-loss suggestion shown on brand-new picks (stop = current price − 2.5 × ATR) — NOT used to check positions you already hold, which use a flat 10% stop instead (see "Stop-Loss Breach" under Sell Signals).

The highest price a stock has traded at over the past 52 weeks (one year). Used in the EXTENDED soft flag and in the stop-loss suggestion shown on brand-new picks (flat 12% below the 52-week high). Positions you already hold do NOT use the 52-week high at all — their stop-loss is a flat 10% off your entry price, and it never moves regardless of how high the stock trades afterward.

How far a stock's current price is above its 50-day SMA, expressed as a percentage. A stock more than 25% above its SMA50 is flagged as "EXTENDED" — meaning it has moved too fast and chasing it raises the risk of buying near a short-term peak.

A condition where a stock's price has risen so quickly that it may be due for a pause or pullback. The screener flags any stock with RSI above 70 as overbought — it passes the gate, but the email notes the elevated RSI.

Price-to-Earnings ratio — how much investors are paying per dollar of earnings. Calculated as share price ÷ earnings per share. A positive P/E means the company is profitable. The screener uses P/E as a profitability proxy since Finviz doesn't expose raw EPS.

Price/Earnings-to-Growth ratio — the P/E divided by the expected earnings growth rate. A PEG above 2.0 suggests a stock is overvalued relative to its growth. Stocks with PEG > 2.0 are hard-excluded from the rankings even if they pass all 8 gates.

Earnings Per Share — a company's net profit divided by the number of outstanding shares. A measure of how profitable a company is on a per-share basis. The screener checks EPS growth year-over-year.

The percentage of a stock's available shares that are currently sold short (borrowed and sold, betting the price falls). Gate 7 excludes stocks above 20% short float — that level signals heavy bear conviction and crowded-short volatility risk (the stock can spike hard on a short squeeze, or break down hard if the bears are right).

A company's current assets divided by current liabilities — a basic liquidity check (can it cover what it owes in the next 12 months?). Part of Gate 8 (Financial Health); the screener requires at least 1.0.

Total debt divided by shareholder equity — how leveraged a company is. Part of Gate 8 (Financial Health); the screener excludes anything above 2.0x, which only flags genuinely over-levered balance sheets, not normal use of debt.

Compound Annual Growth Rate — the steady annual rate at which something (like revenue) would have grown to reach a given result over a period. Used in the revenue gate: requires ≥ 20% 5-year revenue CAGR.

A single 0–1 number that summarizes how well a stock scored across 6 scored dimensions (momentum, profitability, revenue, cash flow, sentiment, options — each contributing an equal-weighted subscore). The 2 newer gates (ownership, financial health) are pass/fail only and don't feed the composite score. Stocks are ranked by composite score — the top 10 become the day's picks.

A 0–1 score for a single gate dimension that feeds into the composite score. Momentum subscore rewards more passing checks (e.g., above both SMAs and positive 1M and 3M). Revenue subscore rewards higher 5Y CAGR, capped at 100%.

A 1–10 number (10 = most dangerous) assigned to each ranked stock. Factors in ATR-based volatility, extension above SMA50, overbought RSI, and whether the P/E is negative. A baseline of 5 is adjusted up or down.

The price range the screener suggests as a reasonable point to buy a stock. If the stock is near its SMA50 (within 5%), that's flagged as a reasonable entry. If it's extended, the screener says wait for a pullback toward SMA50 first.

A pre-set price level suggested for a brand-new pick you don't own yet. The screener calculates two: an ATR-based stop (price − 2.5 × ATR) and a flat 12% trailing stop from the 52-week high. The tighter one (higher price) is marked "use this." This is a suggestion shown at pick time — not the same calculation used to check positions you already hold (see "Stop-Loss Breach (Sell Signal)" under Sell Signals, which uses a flat 10% off your entry price instead).

A stop-loss that follows a stock's price upward but never moves down. This concept only applies to the stop-loss SUGGESTED for brand-new picks (flat 12% below the 52-week high) — as of 2026-07-11, the stop-loss that actually checks positions you already hold is NOT trailing at all. It's fixed at 10% below your entry price forever, even if the stock runs way up afterward.

A warning note attached to a ranked stock — not a disqualifier, but a caution. Soft flags appear in the email for conditions like: extended above SMA50 by 25%+, RSI above 70, or missing QoQ revenue data.

A valuation filter that disqualifies a stock from the rankings even after it passes all 8 gates. One trigger: PEG ratio above 2.0.

The 2–3 gate dimensions a stock scored highest on. Pulled from the subscores and included in the daily email to explain why a stock ranked where it did.

A simulated trade — logged and tracked like a real trade, but no actual money changes hands. All [sage]SIGNALS picks are paper-traded while the model is being validated. This lets us measure performance without financial risk.

A day the NYSE is open — a weekday that isn't a market holiday. Weekends never count, and a handful of holidays each year (New Year's, MLK Day, Presidents' Day, Good Friday, Memorial Day, Juneteenth, July 4th, Labor Day, Thanksgiving, Christmas) don't either. When a holiday falls on a weekend, NYSE observes it on the nearest weekday instead — main.py's is_trading_day() encodes the exact list. Matters both for when the screener actually runs (it skips non-trading days) and for how forward-price checkpoints resolve on a closed market (see T+1/T+7/T+30/T+90 below).

Forward price checkpoints, measured in full calendar days after the pick date (weekends and holidays count). T = pick date. T+1 = next day. T+7 = one week later. T+30 = about a month later. T+90 = about a quarter later. If a checkpoint lands on a day the market's closed, it uses the most recent close before it. Used to evaluate how each pick performed over time.

The percentage of picks that were profitable at a given time horizon. A hit rate above 50% means the model is picking more winners than losers.

The return of the screener's picks above and beyond what QQQ (the tech benchmark) returned over the same period. Positive alpha = the model is outperforming passive tech investing.

The Invesco QQQ ETF — tracks the Nasdaq-100, a basket of the 100 largest non-financial companies on the Nasdaq. Used as the benchmark for [sage]SIGNALS because it represents the tech-heavy market the screener competes against.

A reference index used to measure the screener's relative performance. [sage]SIGNALS uses QQQ. If the picks don't consistently beat QQQ, you'd be better off just buying QQQ and doing nothing.

The paper-trading phase before any real capital is deployed. The model needs 12–16 weeks of consistent data to establish statistical confidence that its hit rate and alpha are real, not noise.

Where a stock's score or return falls relative to all other stocks in the same run. A stock in the 90th percentile scored higher than 90% of all screened stocks that day.

How much you stand to gain divided by how much you stand to lose on a trade — the distance from your entry to your target, over the distance from your entry to your stop. Saying an entry and a stop are "too close together" is really saying the reward-to-risk ratio is too low to justify the trade.

Getting stopped out by ordinary price noise, then watching the stock immediately reverse back in your original direction. The failure mode of a stop set tighter than a stock’s normal daily movement.

Replaying a rule against history to see what it would have done, instead of guessing from a few memorable trades. Its answer is only as good as the period it covers.

The prevailing market environment — rising, falling, or choppy. Stop-loss rules behave oppositely across regimes, so a rule tested in only one of them has not really been tested.

Drawing conclusions from the cases that survived and are therefore memorable, while the ones that quietly failed never come to mind. It makes a rule look better or worse than it was.

Letting a backtest use information that would not have been available at the time, which makes results look better than anything achievable in reality.

An automated flag generated each morning for positions currently held (read from your Alpaca paper account). Three triggers: deteriorating EPS or revenue growth, analyst sentiment breakdown or softening, or a stop-loss breach.

A held position's sell-signal stop-loss trigger: current price below entry price × 0.90 (a flat 10% below your entry, set 2026-07-11). Static — it never moves as the stock rises, and it's the same 10% for every position, no exceptions. Not to be confused with the "Stop-Loss" shown for brand-new picks (see Ranking & Scoring) — that's a different, unrelated calculation for stocks you don't own yet.

A fake-money brokerage account (via the Alpaca API) where Sage and Ashwad place trades manually to track this strategy in real time. sageSignals never places orders automatically — it only reads the account (positions, balance) each morning to check sell signals and show them on the /positions dashboard page.

When you buy the same stock more than once at different prices, your "entry price" becomes the blended average across all purchases — not your original buy price. sageSignals uses this average (pulled directly from Alpaca) as the entry price for stop-loss calculations.

The paper gain or loss on a position you're still holding — "unrealized" because you haven't sold yet, so it's not locked in. Shown in dollars and as a percentage on the /positions dashboard page.

The primary data source for [sage]SIGNALS. The screener fetches 6 different Finviz "views" per run (overview, valuation, ownership, performance, financial, technical) and merges them by ticker to build a complete data picture of each stock.

The data provider used to fetch historical closing prices for forward price tracking. A separate backfill job runs at 4:30pm ET each weekday to pull T+1 through T+90 closing prices for all picked stocks.

The automation platform that runs the screener on a schedule. A cron job triggers the full pipeline at 9:00am ET every weekday (Monday–Friday). No manual intervention required.

A time-based job scheduler — the mechanism that tells GitHub Actions "run this pipeline at exactly this time on these days." The screener cron is set to `0 13 * * 1-5` (UTC), which maps to 9:00am ET.

The PostgreSQL database that stores all paper trade data, forward prices, QQQ benchmarks, and run metadata. The dashboard reads from Supabase to display performance analytics.

One day of trading summarised as four prices: open, high, low, and close. The low is what tells you whether a stop-loss level was actually touched — a closing price alone cannot.