🇭🇰 HK Mean Reversion: The Stamp Duty Constraint

Why daily scalping is impossible, but weekly/monthly works

⚠️ The Problem

Hong Kong has a 0.1% stamp duty on stock transactions. This sounds small, but it compounds devastatingly for active traders.

Round Trip Cost = 0.1% × 2 = 0.2% per trade

Why This Destroys Mean Reversion

Mean reversion strategies require frequent trading. You buy oversold, sell overbought, repeat. Every trade costs 0.2%.

250
Trades/Year (daily)
50%
Capital Consumed
10%
At Weekly (50 trades)
2.5%
At Monthly (12 trades)
⚠️ Daily mean reversion is mathematically impossible in HK.
Even if you have a perfect strategy, the stamp duty consumes 50% of your capital per year.

📊 Annual Cost by Timeframe

Timeframe Trades/Year Annual Cost Monthly Cost Daily Cost Verdict
Daily 250 50% 4.2% 0.20% ❌ Impossible
2-day 125 25% 2.1% 0.10% ❌ Very Hard
5-day (weekly) 50 10% 0.83% 0.04% ⚠️ Challenging
10-day 25 5% 0.42% 0.02% ✅ Viable
20-day 12 2.4% 0.20% 0.01% ✅ Good
30-day 8 1.6% 0.13% 0.007% ✅ Great

📈 Backtest: Tencent Mean Reversion

We tested a simple mean reversion strategy on Tencent (0700) from 2015-2026: buy when price is >X% below 20-day high, sell when >X% above 20-day low.

Mean reversion backtest
How stamp duty destroys mean reversion alpha

Results: 2% Threshold

Timeframe Trades Gross P&L Stamp Duty Net P&L Annual (Net)
2-day 48 +14.2% +9.6% +4.6% +0.3%
5-day 372 +2.1% +74.4% -72.3% -6.3%
10-day 586 +339.7% +117.2% +222.5% +3.6%
20-day 709 +457.6% +141.8% +315.8% +3.8%
30-day 721 +478.9% +144.2% +334.7% +4.8%
💡 Key Finding: The 5-day timeframe looks profitable gross (+2.1%) but is DESTROYED by costs (+74.4% in stamp duty!). This is the danger zone.

🧮 The Mathematics of Survival

For mean reversion to work after stamp duty:

Required Annual Return to Break Even = Trades/Year × 0.2%
Timeframe Trades/Year Required Return Achievable?
Daily 250 50% ❌ Impossible
5-day 50 10% ⚠️ Very Hard
10-day 25 5% ✅ Challenging but possible
20-day 12 2.4% ✅ Very achievable
30-day 8 1.6% ✅ Easy

🎯 Conclusions

✅ Mean Reversion CAN Work in HK

But only with weekly or monthly timeframes (10-30 day bars). At this scale, stamp duty becomes a manageable 1-5% annual cost.

⚠️ The Danger Zone: 5-day Trading

This timeframe looks attractive (weekly reversion) but requires 10% annual return just to break even on costs. Most strategies will fail here.

❌ Daily Scalping is Mathematically Impossible

250 trades/year × 0.2% = 50% annual cost. Even Warren Buffett can't generate 50% returns after this cost.

Recommendations for HK Mean Reversion

  1. Use 20-30 day bars minimum — reduces costs to 1.6-2.4%
  2. Combine with trend filter — reduce trade frequency
  3. Use wide thresholds — only trade extreme deviations
  4. Consider stamp duty refund stocks — some ETFs have lower costs
  5. Track costs separately — know your breakeven before trading

← Back to Timeframe Analysis | Main Page

Analysis using EODHD data | QuestDB | Python