S7

Overview

The Agent is your personalized AI trading assistant. Unlike a one-size-fits-all approach, you control:

Market Context

Choose which breadth indicators and market data feed into the market diagnosis.

40+ Strategies

Large pattern library (breakouts, pullbacks, mean reversion, sector momentum, VIX, and more) with named trader attribution where applicable.

Playbooks

One-click profiles that bundle trading intent, risk, strategy mix, and filters—activate from the header without rebuilding settings each time.

Stock Screening

Real stocks found via our live market screener and database screening. Strategy-specific criteria. Zero hallucinated tickers.

Custom Strategies

Create your own strategies by describing them in plain English. The AI interprets and applies them.

Stock universe

All stocks, Twitter/X picks, your watchlists, or a union of both—with prioritize (boost your names) or strict (only your list) modes.

First time on The Agent?

Open How it works or Page tour in the header on The Agent for a quick explanation or spotlight walkthrough (tour works best on a wide screen).

Where to configure things

On The Agent, controls are in the header:

  • New Signal — opens the setup wizard (intent, risk, universe, optional playbook)
  • Playbooks — browse and activate a saved profile
  • Watchlists — create lists, add tickers, set one as the active filter
  • Advanced (gear) — Twitter/X handles, custom strategies, stock filter mode (all / watchlist / Twitter / union), filter strength, and related options. Save when done.
  • Re-run — regenerates using your current preferences and filters (use after you change Advanced or playbooks)
  • Quick Refresh — refreshes market context for chat without a full signal regeneration
Caching

Signals are cached per user for today (US/Eastern trading date) with the same intent and compatible settings—reloads can be instant until something material changes. Use Re-run for a deliberate full pass.

How The Agent works (six-step pipeline)

When you New Signal or Re-run, the thinking panel streams these steps. Market Data and Stock Screening are grounded in live feeds and code; AI steps interpret, choose strategies, rank screened names, and build the plan—they do not invent tickers outside the screen results.

1Market Data DATA

VIX, breadth (e.g. MCO/MCSI), trend strength, sector heat, and other context you’ve enabled feed the next step.

2Market Diagnosis LLM

The model classifies regime (bull, range, compression, defensive, etc.) so later steps know which strategies are in play.

3Strategy Selection LLM

From your enabled library (playbook or saved preferences), the AI chooses one or more strategies that fit the regime. When nothing fits, you may get a deliberate no trade / cash outcome. Attribution (e.g. who inspired the pattern) appears when available.

4Stock Screening CODE

Deterministic screeners (broad market pre-filters, price-data verification, and/or database-backed patterns) produce real candidates per strategy—not LLM-guessed symbols.

5Ticker Selection LLM

The model ranks and selects from the screened set (with your watchlist/Twitter filter rules applied in the pipeline).

6Trade Plan LLM

Concrete entry, stop, target, and conviction for the top idea(s). Optional Backtest Validation may appear afterward when historical KPIs exist for that strategy and symbol.

Transparency

Each step shows reasoning in the thinking panel. For a shorter conceptual overview, use How it works on The Agent page.

Market context configuration

In the Market Data step, the agent pulls context you’ve enabled (often adjusted via New Signal wizard or saved preferences). Typical inputs include:

Available Market Context Inputs

  • VIX Level & Interpretation - Volatility index level and what it means (elevated, normal, low)
  • ADX (Trend Strength) - Measures how strong the current market trend is
  • MCO/MCSI (McClellan Oscillator) - Market breadth momentum indicators
  • % Above 5/20/50/200 SMA - Percentage of stocks above various moving averages
  • Advancing/Declining % - Market breadth: how many stocks are going up vs down
  • Twitter Sentiment - Aggregated sentiment from financial Twitter
  • Hot Sectors - Currently outperforming market sectors
Pro Tip

Day traders: Enable % Above 5 SMA and Twitter sentiment for intraday edge. Swing traders: Focus on MCO/MCSI and % Above 50 SMA for multi-day moves.

Strategy library

The platform includes 40+ screenable patterns and strategies. In Strategy Selection, the AI chooses from the subset you’ve enabled (via playbook or preferences). The list below samples well-known named patterns with trader attribution; many additional screens (e.g. Jeff Sun momentum, VIX pivots, inside-day variants) live in the product without changing the core idea: code finds candidates, AI chooses and plans.

Breakout & Momentum (7 strategies)
  • VCP (Volatility Contraction Pattern) by Mark Minervini — Tight base near 52-week high after progressive contractions. Bull market, Stage 2 uptrend.
  • Breakout (Momentum Consolidation) by Kristjan Kullamagi (Qullamaggie) — Top 1-2% momentum stocks consolidating along 10/20 EMA then breaking out. Bull/trending market.
  • Chart Pattern Breakout by Dan Zanger — Cup & handle, bull flag, flat base breakouts with volume surge. Strong bull market.
  • Gap and Go by Ross Cameron / Andrew Aziz — Stocks gapping up 4%+ on catalyst with heavy volume. Any market (catalyst-driven).
  • Turtle Breakout by Richard Dennis — 20/55-day channel breakouts in emerging trends.
  • Holy Grail (Pullback) by Linda Raschke — Pullbacks to 20-period EMA in strong uptrends (ADX > 30).
  • Weinstein Stage 2 Breakout by Stan Weinstein — Break above Stage 1 base with rising 30-week MA and relative strength vs SPY.
Earnings & Catalyst (1 strategy)
  • Episodic Pivot (Power Earnings Gap) by Kristjan Kullamagi (Qullamaggie) — Stock gaps up 10%+ on earnings with 3-5x volume. Bull market / earnings season.
Trend Pullback (1 strategy)
  • Prime Trading Pullback (21 EMA) by Alex (TradersLab / PrimeTrading) — Liquid leaders pulling back to 21-day EMA structure zone. Confirmed uptrend pullback.
Compression & Squeeze (2 strategies)
  • Inside Bar Breakout by Al Brooks — Compression patterns (inside days) ready to break out on volume.
  • TTM Squeeze by John Carter — Bollinger Bands inside Keltner Channels, about to expand into a directional move.
Reversal (5 strategies)
  • San-Ku / R3 (Three Gaps Down) by Larry Connors — Japanese exhaustion pattern after three consecutive gaps down.
  • "2B" Reversal (Bear Trap) by Victor Sperandeo — Failed breakout below support that reverses sharply higher.
  • Selling Climax (Volume Spike) by Richard Wyckoff — Panic selling with massive volume spike, signaling capitulation and reversal.
  • Class A Bullish Divergence by Dr. Alexander Elder — Price makes new low but momentum indicator doesn't (bullish divergence).
  • Bullish Engulfing at Support by Steve Nison — Large bullish candle engulfs prior bearish candle at key support level.
Mean Reversion (1 strategy)
  • Mean Reversion (Fade) by Jim Simons (RenTech) — Oversold bounces in range-bound markets.
Momentum & Sector Rotation (3 strategies)
  • Relative Momentum Leaders by Gary Antonacci / Mebane Faber — Stocks outperforming SPY over trailing quarter, above 50/200-day SMA, near 52-week highs.
  • Momentum Breakout by William O'Neil / Mark Minervini — New 52-week highs with monthly uptrend and volume confirmation.
  • Sector Leaders (Rotation) by Sam Stovall / Mebane Faber — Large-cap names trending above all three SMAs with institutional-grade liquidity.
Bearish / Short (4 strategies)
  • Parabolic Short by Kristjan Kullamagi (Qullamaggie) — Shorting stocks that have gone parabolic (up 50-1000%+ in days/weeks). Any market.
  • Bear Flag / Pennant by Thomas Bulkowski — Weak bounce after sharp decline, then breakdown below flag low.
  • Bear Flag / Descending Triangle by Thomas Bulkowski — Bearish continuation with lower highs and flat support.
  • Descending Triangle Breakdown by John Murphy — Repeated tests of horizontal support with lower highs, then breakdown.
Backtested Strategy Performance

The following strategies have been backtested with historical data (SPY, 2-year lookback):

StrategyWin RatePFSharpeMax DD
Absolute Momentum 37.9% 1.24 1.56 -39.0%
Bb Squeeze 78.2% 8.18 10.25 -8.8%
Bearish 37.8% inf -0.11 -8.2%
Bottom Picker 65.5% 1.41 1.69 -36.1%
Cci Momentum 70.9% 4.50 10.50 -16.3%
Deep Trend Pullback 80.7% inf 13.93 -0.4%
Donchian Breakout 81.7% 16.83 16.27 -16.5%
Earnings Catalyst 41.1% 1.42 -0.38 -15.8%
Ema Cross Fast 69.4% 6.31 9.28 -18.7%
Ema Crossover 34.5% 1.70 2.09 -22.6%
Gap And Go 42.1% 5.47 0.24 -16.4%
Golden Cross 66.4% 4.29 6.22 -11.5%
Holy Grail 41.4% inf -0.51 -4.8%
Ichimoku Cloud 74.4% 8.51 12.00 -11.2%
Inside Day 57.6% 1.61 2.63 -22.5%
Jeff Sun Momentum 51.6% inf 5.84 -13.4%
Keltner Momentum 38.1% 1.35 1.21 -26.3%
Macd Bb Combo 70.8% 5.09 7.99 -24.8%
Macd Trend 38.7% 1.44 0.90 -19.3%
Mean Reversion 63.5% 1.26 1.35 -22.4%
Momentum Macd 43.4% 3.93 1.79 -25.3%
Obv Trend 69.9% 4.73 11.22 -9.8%
Parabolic Short 23.4% 0.75 -2.19 -24.4%
Power Earnings Gap 41.1% 2.19 1.04 -13.6%
Prime Pullback 36.1% 1.22 1.47 -33.9%
Qullamaggie Breakout 37.9% 1.34 -0.07 -19.4%
Reversal 2b 51.7% inf 5.55 -7.8%
Reversal Support 53.0% inf 1.41 -17.6%
San Ku 63.9% inf 4.12 -7.5%
Sma Trend 30.6% 1.25 1.12 -32.5%
Strength List 41.8% 1.02 0.20 -39.6%
Supertrend 41.1% 1.59 2.87 -22.4%
Supertrend Pullback 81.6% 87.96 10.55 -5.8%
Triple Ema Crossover 37.2% 1.54 1.55 -23.0%
Ttm Squeeze 45.3% inf 2.18 -22.7%
Turtle Breakout 37.1% 1.18 0.72 -34.9%
Vcp Patterns 41.9% inf 0.64 -7.2%
Vix Pivot Bands 52.8% 1.13 1.57 -31.7%
Vix Pivot Daily 49.8% 1.18 0.92 -35.5%
Vix Pivot Weekly 48.2% 1.55 2.57 -32.1%
Vix Reversal 59.6% 1.70 2.91 -21.0%
Waterfall Decline 33.8% 1.39 1.81 -34.1%
Weinstein Stage2 43.7% inf 6.82 -23.4%
Williams R Reversal 80.1% 21.43 16.08 -21.1%
Strategy Attribution

When the agent selects a strategy, you'll see who created it — for example, "VCP by Mark Minervini (SEPA)". This appears in both the thinking display and the final strategy card.

How stock screening works

After Strategy Selection, Stock Screening runs as code: fast market-wide pre-filters, per-strategy rules, live price checks, and/or precomputed database patterns—so every candidate symbol is grounded in data before Ticker Selection.

Strategy-Specific Screening

Each strategy category uses different screening criteria:

  • VCP / Breakout — Screener: Price above SMA20/50/200, within 10% of 52-week high, $2B+ market cap
  • Episodic Pivot — Screener: Top Gainers signal, Week Up performance, 500K+ volume
  • Prime Trading — Screener: Month Up, above SMA20/50, $2B+ market cap, 1M+ volume
  • Reversal / VIX — Database: VIX signal data with institutional-grade setups
  • Compression — Database: Inside day pattern data with pattern scoring
  • Momentum — Database: Sector strength rankings with top-performing stocks
  • Parabolic Short — Screener: Most Volatile signal, Month Up, 500K+ volume
Zero Hallucination Guarantee

Every stock ticker the agent recommends was found by real market screening (our live screener or our database). The AI never invents stock names — it only selects from stocks that actually match the strategy criteria today.

When No Stocks Match

If the screening finds zero candidates for the chosen strategy, the agent tells you clearly: "No stocks matched screening criteria for [strategy] today." Sometimes the best trade is no trade, and the agent won't force a pick when there are no real setups.

Creating Custom Strategies

You can create your own strategies by describing them in plain English. The AI will interpret your description and apply the logic during synthesis.

How to create a custom strategy

  1. Open Advanced on The Agent page
  2. Scroll to the Custom Strategies section
  3. Click "Add Custom Strategy"
  4. Enter a name (e.g., "My Gap & Go")
  5. Describe the strategy in detail
  6. (Optional) Add market condition tags
  7. Click "Create Strategy"

Example: Gap & Go Strategy

"Buy stocks gapping up 3% or more at the open with above-average volume. Look for continuation in the first 30 minutes. Target 1:2 risk-reward. Best in trending markets when sector rotation is strong."

Best Practices

Be specific in your descriptions. Include: entry criteria, volume requirements, market conditions, and risk management guidelines. The more detail you provide, the better the AI can apply your strategy.

Stock universe & watchlists

Configure the stock filter in Advanced (Stock Filter section). Watchlists themselves are managed from the header Watchlists button; then set Use as Filter on a list or wire the filter mode to watchlist/Twitter/union as needed.

Filter modes

All stocks

Full market universe (default).

Twitter / X picks

Stocks surfaced from handles you enable (lookback and confidence thresholds in Advanced).

My watchlist

Focus on symbols from your active watchlist.

Union

Combine watchlist and Twitter picks.

Prioritize vs strict (watchlist)

Prioritize (common default): screening still runs broadly, but names on your watchlist get a score boost so they rise to the top. Strict: only symbols in your filter set are eligible; you may still see reference context if nothing on your list triggers.

Creating a watchlist

  1. On The Agent, click Watchlists
  2. Create a list, add tickers, then Use as Filter if you want it to drive the pipeline
  3. In Advanced, set Stock Filter to watchlist (or union) and save

Understanding the thinking panel

The Agent streams six core steps while it runs (labels match the UI). A seventh row—Backtest Validation—appears when historical KPIs are available for the chosen strategy and symbol.

  • Market Data DATA — Live context fetch
  • Market Diagnosis LLM — Regime and narrative
  • Strategy Selection LLM — Which pattern(s) fire for this run
  • Stock Screening CODE — Candidate counts and sources
  • Ticker Selection LLM — Ranking picks from the screened set
  • Trade Plan LLM — Levels, conviction, and rationale
Chat (Pro+)

After a signal, Pro+ can use the floating Ask The Agent chat for follow-ups (platform AI key or your own BYOK). The assistant can call curated tools for live data, screening, and pipeline context. Pro and below do not include this chat.

Chat assistant (Pro+)

On Pro+, the floating chat is an agentic assistant: it can call tools to fetch live market data, inspect your latest signal, run strategy screens, and more. It is not included on Free, Starter, or Pro.

What the chat can do

  • Fetch live data — VIX levels, market breadth, sector rankings, stock prices
  • Inspect your signal — See what each step of the pipeline decided
  • Compare morning vs now — Check if market conditions changed since your signal
  • Screen stocks — Find candidates across the same broad strategy library as The Agent
  • Social sentiment — See what tracked Twitter accounts are recommending
  • Manage settings — Switch playbooks or update preferences via natural language

When the agent calls a tool, you'll see real-time progress indicators showing what data it's fetching.

Slash Commands

Power users can type slash commands for quick access to specific tools:

/refresh Compare morning signal vs current market
/vix Current VIX levels and interpretation
/breadth Market breadth indicators
/sectors Hot sectors ranked by performance
/screen <strategy> Screen stocks (e.g., /screen inside_day)
/price <symbol> Stock price (e.g., /price NVDA)
/steps Signal pipeline steps
/playbook Active playbook details
/twitter Social sentiment picks
/help Show all available commands

Example Questions

Market Check

"Has anything changed since my signal was generated?"

Stock Screening

"Screen stocks for inside day patterns"

Signal Deep-Dive

"Why was this strategy chosen over others?"

Settings

"Switch to my aggressive playbook"

Models & keys

Pro+ can use the platform default model or bring your own provider keys where configured. Suggestion chips appear when the chat is empty to help you get started.