A few weeks ago I published a finding that surprised me: the rule "don't trade in market corrections" turned out to be wrong for my US system. Correction-entered trades averaged +0.49R. The real dead zone was "Uptrend Under Pressure" — the chop nobody names.
A natural follow-up question: does that hold in Thailand? I already knew Thai signals behave differently in down regimes — I'd found evidence of this before. So I ran the same test on my Thai system's walk-forward history. The answer: opposite in almost every way.
The test — identical method, two markets
For every trade in the system's walk-forward backtest, I tagged the market regime on the trade's entry date — using the same IBD market-condition state machine (Confirmed Uptrend / Uptrend Under Pressure / Market in Correction). Then I looked at the average outcome per trade within each regime. I ran this on two independent live Thai systems for convergence:
- DI Trend Thai — the main Thai trend-following system, n=799 trades, 19-year walk-forward (2007–2026)
- Contracting Breakout — TypeA/B/C pattern scanner, n=96 trades
Outcome unit is fractional return (not R, because the Thai books don't have a clean stop column). The diagnostic — sign of mean by regime and relative drawdown — is unit-agnostic.
US result (the prequel)
Average R-multiple per trade, by regime on entry day — US Nasdaq, 1,535 trades, 20-year WF:
| Regime on entry | n | Avg result |
|---|---|---|
| Confirmed Uptrend | 755 (49%) | +0.69R |
| Uptrend Under Pressure | 128 (8%) | −0.00R ← dead zone |
| Market in Correction | 652 (43%) | +0.49R ← still positive |
Folk rule refuted. Corrections were fine. The dead zone was somewhere else.
Thai result — the mirror image
Average fractional return per trade, by regime on entry day — SET, 895 trades combined:
| Regime on entry | n | Avg result |
|---|---|---|
| Confirmed Uptrend | 786 (88%) | +4.40% |
| Uptrend Under Pressure | 90 (10%) | −1.50% |
| Market in Correction | 19 (2%) | −4.87% ← dead zone |
The folk rule holds in Thailand. Both non-confirmed regimes lose money. The correction is the worst of the three, and Under Pressure is also negative. The edge is almost entirely concentrated in Confirmed Uptrend — 88% of all Thai trades fired in that state, and they averaged +4.40%.
Both sleeves agreed: the DI Trend system showed −4.87% for corrections, and the contracting-breakout system had essentially zero correction-regime trades (it already self-selects). Convergent, not a single-system artifact.
The contrast in one table
| Dead zone | Corrections | Implication | |
|---|---|---|---|
| US (Nasdaq) | Under Pressure (~0R) | +0.49R — still positive | Skip Under Pressure; keep corrections |
| Thai (SET) | Correction (−4.87%) | −4.87% — lose money | Cut corrections to zero; throttle Under Pressure |
Why does Thai correct differently?
In the US, the RS≥80 filter produces leaders that are strong enough to survive — and benefit from — a correcting index. They're the stocks that emerge as next-cycle leaders. This is O'Neil's observation, and it shows up cleanly in the data.
In Thailand, the same filters produce a different outcome. Prior research on Thai signals found that Thai signals reverse in regime DOWN — the patterns that work in a Confirmed Uptrend stop working when the index corrects. This isn't because the filters are wrong; it's because the Thai market structure, liquidity profile, and investor base produce different regime-conditional dynamics.
In practical terms: a Thai leader breaking out during an index correction doesn't behave like a US O'Neil leader. It fails more often. The filter isn't strong enough to overcome the market headwind in Thailand the way it is in the US.
What the sizing overlay shows
I tested several sizing schemes — re-weighting each trade's contribution based on the regime at entry. In Thai, the best scheme was full size in Confirmed Uptrend, half size in Under Pressure, zero in Correction (the folk-rule scheme):
| Scheme | Total return | Max drawdown | Return / DD |
|---|---|---|---|
| Flat (full size always) | +3,230% | 524% | 6.17 |
| Half/zero (folk rule) | +3,390% | 418% | 8.10 |
| US-style (cut Under Pressure, keep corrections) | +3,365% | 473% | 7.11 |
For Thai, the throttle is a genuine enhancement — it improves both return and drawdown, because the cut regimes had negative expectancy. This is different from the US, where the throttle was a drawdown smoother only (same total return, lower DD).
If you apply the US-style rule to Thai — keep corrections, cut Under Pressure — you do better than flat, but worse than the correct Thai rule. The US logic leaks when ported to SET.
What this means if you trade Thai stocks
Most of the IBD-style methodology was developed on US market data. The regime rules in particular — when to be aggressive, when to step back — are Nasdaq-calibrated. The finding here is that the Thai market gate should be stricter than the US one:
- In the US: cut size during "Under Pressure," keep trading corrections. The correction-entered leaders are the next cycle's winners.
- In Thailand: cut corrections to zero. Throttle Under Pressure. Your edge lives almost entirely in Confirmed Uptrend.
If your entry filter is already strict — RS≥80, RS Line slope positive, Stage 2, 200d above — you're already implicitly regime-gating, because most of your trades will naturally fire in Confirmed Uptrend. The Thai systems here fired 88% of their trades there without any explicit regime filter. The throttle mainly trims a small negative tail.
But if you're asking "should I be aggressive during a SET correction because O'Neil says corrections are where leaders emerge?" — the answer from this data is: not in Thailand, not with this kind of system.
Caveats
The Thai correction bucket is small — only 19 trades (2% of the total). The −4.87% mean carries sampling noise. But the direction is consistent with Under Pressure (−1.50%) and with the prior Thai regime research. The decision — cut it — is robust to whether the true mean is −3% or −7%.
This is a sizing overlay only. Entries are unchanged. It's research — nothing is wired live until after the November system arbiter. And the Thai/US contrast here applies to breakout-style trend-following systems with strict entry filters; different system types might see different results.
Personal research. Not investment advice. Thai data 2007–2026, US data 2006–2026. Walk-forward backtest on fixed trade sets — a sizing study, not a forecast. Past results do not guarantee future performance.