Video summary

CUAN RIBUAN PERSEN TRADING PAKAI AI! FT. Kevin Hendrawan

Main summary

Key takeaways

Finance

Finance-focused subtitle summary (markets, investing, AI trading)

Core message / framing

  • AI is presented as a tool to save time and systemize repetitive trading/investing decisions, not as “magic” that guarantees returns.
  • The host/guest emphasize process, discipline, and risk management, including:
    • avoiding over-bias/over-greed
    • recognizing limitations such as liquidity constraints and model error

Assets / tickers mentioned

Indonesian stocks (6 predefined candidates for the AI product)

  • BBRI
  • BMRI
  • Telkom (TLKM)
  • ANTAM (ANTM)
  • PGAS
  • (A “ROW” mention also appears in the subtitles)

Other stocks mentioned

  • Indosat (rumored topic; no ticker given in subtitles)
  • Petro / “Petro” referenced (no explicit ticker)

Index

  • IHSG (Indonesia Stock Exchange / composite index)

Note: “MSCI” is mentioned as an event context, but not as an investable instrument.

Methodology / frameworks described

1) AI-driven trading loop (“buy or sell” decision automation)

  • The system aims to imitate human decision-making by combining:
    • quantitative market data
    • qualitative / sentiment-like subjective signals
  • An LLM + machine learning pipeline processes signals into a final “buy or sell” decision.
  • Repeated simulation / trial-and-error:
    • The AI simulates trading ~170,000 times per day
    • It discards “wrong” strategies and iteratively improves accuracy.

2) “Long-term swing trade” concept (not scalping/daily)

  • The system targets a market-cycle aware swing, such as:
    • Buy → sell when high → buy again when low
  • It is positioned as less constrained by liquidity than scalping, though liquidity is still a critical practical constraint.
  • Dynamic holding period:
    • No fixed take-profit/stop-loss is pre-set.
    • The AI re-evaluates periodically.
    • Longest hold cited: BMRI ~18 months
  • Some trading may still be frequent (even daily) when momentum supports it, but the product is framed more for non-scalping users.

3) Correlation / macro-to-company linkage (index impact)

  • For each stock, AI estimates its relationship with IHSG using long history (e.g., ~15 years):
    • whether correlation is positive or negative
    • the magnitude (how much it moves opposite, if negative)
  • Inputs can include waiting period / timing changes, meaning signals depend on time lag.

4) Liquidity-aware risk constraint (anti-“self-fulfilling profit” issue)

  • Liquidity is framed as a major real-world risk:
    • Even if AI “predicts” an exit at a target price, thin liquidity may prevent selling the full position at that price.
    • If you can’t exit as assumed, historical backtests may become misleading.
  • This is used to justify:
    • why the AI prefers large-liquidity stocks
    • why the product may cap users / scale constraints exist (collective market impact can invalidate strategy assumptions)

5) Bias management / trading discipline (human framework)

  • Avoid bias such as confirmation bias (only seeking evidence that supports your thesis).
  • Emphasize:
    • having a clear plan (TP/SL and execution rules)
    • self-control to prevent greed after TP and hope after SL
    • knowing when not to trade (staying quiet) as part of discipline

Key numbers / explicit performance expectations & constraints

  • Product scope / eligibility: more suitable for capital “IDR 100 million and above” (a “R00 million” phrasing appears garbled but indicates a threshold).
  • AI simulation scale: ~170,000 simulated trades per day
  • Stock selection data requirement: prefer stocks with long history, “10 years and above”
  • Development timeline:
    • BMRI AI took ~9 months to develop / patent-release (per subtitles)
    • project started late 2022 / early 2023
    • released recently “within the last year”
  • Hold duration: example longest hold BMRI ~18 months
  • Return target expectations (non-guaranteed):
    • target described as 15–20% per year (user hopes)
    • rejects “multibagger chasing”; prioritizes time freedom
  • Liquidity threshold (qualitative):
    • to move to faster timeframes (e.g., 1-year or 6-month aggressive strategies), IHSG liquidity must be at least ~4× higher than current
  • Example sellability price levels:
    • mentions selling BMRI around 4,300 and the 4,290 / 4,280 range to argue execution realism vs model targets

Recommendations / cautions explicitly stated

  • AI is not magic: it mainly helps save time and reduce repetitive work.
  • Do not use AI “stock picking” as a screener replacement:
    • the product is not positioned as “instantly says this stock is good”
    • emphasis is on cycle identification and exit timing
  • Liquidity matters:
    • thin liquidity can invalidate signals because you can’t exit fully at predicted prices.
  • Not all stocks should be analyzed:
    • some may be “useless to analyze” because their movement/structure doesn’t suit the approach.
  • Avoid scalping stress:
    • scalping requires too many correct decisions; it’s described as stressful and difficult to do consistently.
  • Be skeptical of one-source AI tools:
    • if the AI lacks rich local market data, it may fail (contrast is made vs generic tools)

Disclosures / disclaimers

  • No explicit “financial advice” disclaimer appears in the provided subtitles.
  • The conversation stresses uncertainty and iterative improvement (i.e., results are not guaranteed; AI is still learning).

Presenters / sources mentioned

  • Kevin Hendrawan (guest; developer of “SARA AI / Sara Invest / people’s shares” referenced in subtitles)
  • Podcast host: referred to as “Cakin” (speaker introducing the guest; full identity not clearly spelled out beyond “Cakin”).

Original video