Video summary

AI Trading Indicator 2.0 đŸ”„ Smart Exits + Automation Ready

Main summary

Key takeaways

Finance

Finance-focused summary (markets, investing/trading framework, risk, performance)

What the video is about

  • Updates to a TradingView indicator (“V app indicator”) that uses Lorenz until classification machine-learning concepts plus additional filters to automate trade entries/exits.
  • A strong focus on “dynamic exit”—closing trades based on the model’s momentum/trend shift rather than only predefined take-profit/stop-loss levels.
  • The system is presented for bot/automation trading, with notes on how manual traders can mimic the bot’s behavior.

Instruments / tickers / assets mentioned

  • Nasdaq futures (5-minute chart) (described as “Nasdaq futures chart”; no explicit ticker)
  • Bitcoin
  • KuCoin (exchange name / execution venue)
  • NVIDIA (chart example)
  • Palantir (chart example)
  • MES (Micro E-mini S&P 500 Futures)
  • MNQ (Micro E-mini Nasdaq-100 Futures)
  • SPY (SPDR S&P 500 ETF)
  • IBKR / IBKR Gateway (execution platform mention)
  • KuCoin cryptos (context for order execution)

Core methodology / step-by-step framework

  1. Load / reload indicator

    • Reload the Lorenz until classification indicator and/or update to the new version (via the described notification/reload flow).

  2. Trade management structure

    • Traditional/manual flow:
      • Wait for the model’s buy/sell signal.
      • Close trades when:
        • Price hits predefined profit-taking or stop loss, or
        • The indicator shows a momentum shift (reverse signal).

  3. Dynamic exit (central feature)

    • Enable Dynamic Exit so the model decides when to exit/close based on predicted trend reversal timing, rather than relying only on the bottom momentum line.

  4. Kernel / model smoothing

    • Uses built-in kernel regression/trade setup referenced as:
      • “regression 25”
      • “trade with kernel” (already default)
    • Optional improvement: Kernel smoothing lag via “Enhance kernel smoothing” with an adjustable value (example target: ~20).
      • Discussion includes using smaller values for choppy conditions.
      • Mentions turning it off/adjusting during regular market hours.
    • Goal: reduce dead cat bounce / fake reversals in choppy action.

  5. VWAP filter (explicit risk/context filter)

    • Must enable VWAP filter and plot VWAP.
    • Trading logic uses whether price is below/above VWAP, with a tolerance band.
    • Example tolerance guidance:
      • Default example: tolerance = 7
      • Later guidance: tolerance ~0.7 (“0.7 is good”)
      • Warns that very tight tolerance (e.g., “two or one”) may be too strict and likely to miss trades.

  6. ADX filter (trend strength gating)

    • Initially described as unchecked by default, but recommended to enable with a threshold (example: ADX = 17).
    • Stricter use (e.g., around 20) can reduce signals.
    • Mentions “strict guidelines” such as being above a long moving average (e.g., “above 200-day moving average”).

  7. Bot alerts / marker logic

    • Enable options such as:
      • Show bot alert markers (to visualize bot-style actions like “exit long” / “exit short”).
      • Alert only on bar close (kept off for “early entry/early close” behavior; delaying alerts is described as part of bot behavior).
      • Use re-entry after stop loss to reduce impact of stop-loss whipsaws and restart positions when momentum returns.

  8. No fixed profit target for the bot

    • Bot behavior is described as lacking a predefined take-profit; it effectively “rides until” the indicator sends an exit signal.

Key numbers and performance metrics stated

  • Backtest win rate (manual-like, without Dynamic Exit):

    • 56% win rate
    • Last 62 trades: 35 won, 27 loss

  • Backtest win rate (with Dynamic Exit enabled):

    • 97% win rate
    • Reported as: “100% out of 55” on a 2-minute chart example (55 trades all detected correctly)

  • Important clarification

    • Win-rate numbers refer to how often the indicator correctly detected trend direction/reversal and matched its modeled trade-exit logic—not necessarily “profit vs loss” under a simplistic fixed TP/SL interpretation.

Explicit recommendations / cautions (as stated)

  • Recommended indicator settings (examples)

    • Enable Dynamic Exit
    • Enable ADX filter with ADX ≈ 17
    • Enable VWAP filter + plot VWAP
    • VWAP tolerance suggested around 7, and later “0.7 is good” (with experimentation)

  • Caution on VWAP tolerance tightness

    • If tolerance is too close to VWAP (e.g., 2 or 1), signals may occur only when price is extremely near VWAP—potentially causing missed major trades.

  • Timing guidance

    • After-hours: consider more smoothing / dead-cat-bounce reduction.
    • During market hours: smoothing may be less necessary; the first hour can be more volatile.

  • Bot-specific caution

    • Use re-entry after stop loss to reduce whipsaw effects (dead-cat-bounce / stop-loss then immediate reversal).

  • Disclosure note

    • No explicit “not financial advice” disclaimer was present in the provided subtitles.

Execution / automation architecture described

  • A three-system pipeline:

    1. TradingView runs the indicator (“VIP indicator”) and generates alerts.
    2. A bot runs on a virtual machine and listens for alert messages.
    3. Execution targets:
      • KuCoin for crypto
      • IBKR (via IB Key Gate Gateway) for stocks and futures

  • Bot alert actions mentioned:

    • Buy
    • Exit long
    • Sell
    • Exit short
    • plus an emergency stop loss referenced as part of alert payloading

  • Platform note

    • Claims configurability for other platforms (e.g., Thinkorswim, Charles Schwab, etc.), but that requires local machine setup.

Presenters / sources

  • The video uses “I/we” references only; no named individual or organization is explicitly stated in the provided subtitles.
  • No external research sources are cited in the provided subtitles.

Original video