เทรดแบบ Quant ...

Main ideas / concepts covered

1. Trading isn’t only “directional”

   • Common misconception: trading = predicting whether price will go up/down.
   • Reality: there are many strategies grounded in math/statistics.
   • The discussion frames trading as directional vs non-directional (market-neutral / pairs / relative value).

2. The role of statistics (“statical” / “stat”)

   • Statistics help identify repeating patterns and assume they may continue—but not with certainty.
   • The emphasis is on models, distributions, and process validity, not guaranteed outcomes.

3. Core statistical building blocks

   • Correlation: how two assets move together (positive/negative relationship).
   • Cointegration: assets whose spread tends to revert (highlighted as central to pairs trading).
   • Half-life: the time it takes for the spread to revert toward equilibrium (used conceptually for mean-reversion timing).
   • Asymptotic arbitrage / mean-reversion frameworks: strategies aiming to profit from convergence over time.

4. Trading frameworks: what each implies

   • Directional trading
     • “Follow the trend/direction” (value/predictive investing style).
     • Potential profit, but risk is higher because markets can behave unpredictably.
   • Non-directional trading
     • Focuses on relationships between two assets rather than broader market moves.
     • Seeks to reduce exposure to the overall market by trading relative spread behavior.

5. Pairs trading and hedging

   • Uses two assets (or a small set) to form a spread.
   • Hedge/mix the assets using a hedging ratio so the combined series is more predictable.
   • Expect the spread to move away from its mean and then return toward equilibrium (mean “reversal” / convergence).

6. Market factor and “market neutral”

   • Market factor: the common movement driven by overall market-wide risk.
   • Market-neutral strategies try to cancel that market exposure by trading assets with similar factor exposure in opposite directions (e.g., long one, short the other).

7. OU process (Ornstein–Uhlenbeck) for mean reversion

   • OU is introduced as a simple stochastic model for mean-reverting behavior of the spread.
   • Practical output: estimating parameters like reversion speed / half-life.
   • Limitation: the OU model can fail if assumptions break (spread dynamics shift; reversion becomes unstable).

8. Cash-and-carry / futures basis and exponential payoff intuition

   • Introduces cash-and-carry arbitrage (described as a “future spot”-type relationship).
   • Explains that futures price vs. spot is influenced by:
     • Financing/interest
     • Storage/holding effects
   • Uses exponential terms (e.g., (e^r) / (e^{-r})) as intuition.
   • Practical constraints may include liquidity/executeability and the need for correct assumptions.

9. Risk Management (central theme)

   • Even with good models, the talk stresses that you must actively manage risk.
   • Includes:
     • Position sizing based on volatility/standard deviation and assumptions such as stationarity.
     • Stop loss informed by distribution-derived thresholds (e.g., a “Value at Risk / VaL” style low-probability tail cutoff like ~1%).
     • Breaks / out-of-distribution events: if the spread distribution “breaks” (no longer reverts), you should exit rather than assume convergence will resume.

10. Portfolio construction using multiple pairs

    • Extends beyond one pair:
      • Use many pairs so that if some fail, others may still work.
      • “Asymptotic arbitrage” is described as scaling via diversification and tracking which pairs remain valid.
    • Pair “lifetimes” vary (some last months; others longer), and validity can change over time.

11. AI and machine learning—help, not autonomous trading

    • Clear stance: AI is not the decision-maker for trading.
    • Humans must still:
      • define strategy/assumptions,
      • validate models and outputs,
      • decide whether to trade.
    • AI’s role: assist with data cleaning, analysis, scanning, backtesting, and workload reduction.
    • AI output can vary, so it still requires human oversight.

---

Methodology / instruction-like steps (as presented)

A) Identifying a trade idea (relative-value / pairs style)

1. Select candidate assets

   • Often within equity/ETFs (a “beginner pick from textbooks” approach is mentioned).

2. Screen the candidates

   • Use correlation for initial relationship strength/sign screening.
   • Look for pairs with meaningful joint behavior; avoid selecting purely on high correlation, and prioritize robustness.

3. Build the hedged spread

   • Compute a hedging ratio so the combined series reduces market exposure.
   • Combine assets using appropriate proportions (“mixing/cooking” analogy).

4. Verify spread distribution stability

   • Check whether spread behavior is stable and mean-reverting.
   • If the distribution isn’t stable (or an OU fit fails), the model may not work.

5. Model mean reversion

   • Use OU process concepts to estimate:
     • reversion speed / half-life,
     • a typical reversion horizon.

6. Determine entry/exit logic

   • Enter when the spread deviates enough (below/above thresholds).
   • Expect return toward the mean/equilibrium.

---

B) Risk controls while trading a mean-reversion spread

1. Use stop loss on every entry

   • Stop loss levels are tied to distribution/tail probabilities (e.g., a low-probability boundary such as a 1% tail).

2. Size positions by risk

   • If the process appears stationary/stable, sizing can be simpler.
   • If stability is uncertain, reduce size and be more conservative.
   • Sizing is linked to standard deviation / expected variability.

3. Monitor for “distribution breaks”

   • If the spread goes outside the assumed distribution range and shows no reversion,
   • exit earlier rather than waiting.

4. Diversify across pairs

   • Don’t depend on a single pair.
   • Remove pairs that break and replace them with newly validated pairs.

---

C) Market-neutral implementation idea

1. Identify market-factor exposure

   • Look for assets influenced by the same market/sector forces.

2. Trade long/short to cancel the market factor

   • If both are affected similarly by the market, hedging them against each other reduces market risk.

3. Evaluate performance

   • Market-neutral should reduce sensitivity to broad market up/down moves.

---

D) Cash-and-carry / futures arbitrage (conceptual execution)

1. Compare spot vs futures
2. Account for “carry” effects
   • Financing/interest and costs are embedded in the futures-vs-spot relationship.
3. Check for mispricing vs the carry model
   • If futures are “expensive” or “cheap” relative to modeled carry, execute the arbitrage structure (long/short legs across spot and futures).
4. Watch practical constraints
   • Liquidity and execution limits real-world profit.

---

Speakers / sources featured (as named or directly shown)

• Host / interviewer: “The Trading Lab” host (addressed as “Op”)
• Guest: Vice President of an association and a teacher at Maejo University
  • Name appears as “Mr. Apor” / “P’Op” / “P’Gup” in subtitle variants (exact full name unclear due to caption errors).
• Association / University cited: Maejo University
• Books/authors mentioned: Argo and Newton (referenced as a textbook-style starting point for pairs/strategies)
• Sponsor / ad source: Ledger
• Hardware wallet mentioned: Ledger Nano Gen 5 (in the crypto segment)
• Additional references (not primary speakers):
  • Morgan (mentioned as a place offering positions; speaker not identified)
  • References to US equity / NASDAQ / New York and Thai-market context (no distinct speaker named)