Research note · MOEasymmetry · June 2026
Three interviews. Three experienced traders. Dozens of rules about stops, position sizing, market timing, and trade management.
Most trading content stops there — state the rule, show a screenshot, move on.
We didn't stop there. We ran the testable claims against 40+ years of SET daily data (1982–2026, 10,875 bars) and our full Thai backtest records. Six tests. Some confirmed. One didn't transfer to Thai markets at all.
Here's what the data actually says.
The Three Sources
Brian Shannon — 35 years trading US equities, author of Technical Analysis Using Multiple Timeframes, popularized Anchored VWAP. Interview: TraderLion "35 Years of Trading Wisdom in 20 Minutes" (2026-05-13).
Clement Ang — Won the US Investing Championship million-dollar division (150%+ return in 2025) after losing most of his capital in 2022. Interview: TraderLion (2026-01-17).
Financial Wisdom — "5 Signals That Tell You When the Market Is About to Turn" (2026-04-21).
The 6 Claims We Tested
Test 1: "Stage 4 = Never Buy"
Sources: Clement Ang (lesson 2) + Brian Shannon (Stage 2 setup)
The claim: never buy a stock trading below its 200-day moving average. Below the 200d MA = Stage 4 = institutional liquidation phase.
What we tested: We compared 737 backtest trades from our DI system without the 200d MA filter against 622 trades with the filter applied. Data: Thai market 2007–2026.
| Metric | Without 200d filter | With 200d filter | Delta |
|---|---|---|---|
| Trades | 737 | 622 | −115 |
| Win rate | 48.2% | 49.2% | +1.0pp |
| Mean return per trade | +9.81% | +10.81% | +1.00pp |
| Average winning trade | +26.89% | +28.28% | +1.39pp |
| Average losing trade | −6.06% | −6.11% | −0.05pp |
Result: CONFIRMED. The 200-day filter doesn't just cut losing trades out — it selects for larger winners. Removing 115 trades improved mean return by 1pp and grew average winners by 1.4pp. Clement learned this by losing most of his capital on HSBC below its declining 200d MA. The data agrees with him.
Test 2: "After a Bearish EMA Cross, Stop Trading"
Source: Brian Shannon (lesson 5) + Financial Wisdom (signal 5)
The claim: when the 10-week EMA crosses below the 20-week EMA, market regime has shifted hostile. Scale down, expect lower returns.
What we tested: Every 10-week / 20-week EMA weekly crossover on SET from 1982–2026 (2,319 weekly bars, 40 bullish crosses, 40 bearish crosses). Measured 60-day forward returns.
| Signal | N | 60d mean | 60d median | Win rate |
|---|---|---|---|---|
| Baseline (all weeks) | 2,285 | +2.42% | +1.41% | — |
| Bullish cross (10w > 20w) | 40 | +3.35% | +2.17% | 65.9% |
| Bearish cross (10w < 20w) | 40 | −0.44% | +0.03% | 50.0% |
Result: CONFIRMED — especially on the bearish side. After a bearish cross on SET, expected 60-day forward return drops from the baseline +1.41% to essentially zero. After a bullish cross, win rate lifts to 65.9%. This is the regime filter in practice: when the market repeatedly stops you out on the same side, it's not bad luck — it's a regime signal.
Test 3: "Monthly Reversal Candle = Buyers Returning"
Source: Financial Wisdom (signal 3)
The claim: a month that makes a new multi-month intraday low but closes strongly (in the top 30% of the monthly range) signals that buyers overwhelmed sellers. Forward returns are better.
What we tested: 1,100+ monthly SET bars from 1982–2026. We identified 34 months where SET hit a new 3-month intraday low but closed in the top 30% of that month's range.
| Period | Signal months | Ordinary months | Win rate delta |
|---|---|---|---|
| 30 days | +1.91% median | +0.83% median | 64.7% vs 54.5% (+10pp) |
| 60 days | +2.65% median | +1.28% median | 64.7% vs 56.0% (+9pp) |
Recent signal months on SET: Jul 2022, Mar 2023, Dec 2023, Aug 2024, Apr 2025, Jan 2026.
Result: CONFIRMED as a confluence signal. The +9pp win rate improvement is meaningful. When sellers push SET to a multi-month low but buyers close it strongly, the 60-day follow-through is significantly better. Not a standalone trade trigger — use it to increase confidence when other signals also align.
Test 4: "50-Month MA Separates Bear Markets from Corrections"
Source: Financial Wisdom (signal 4)
The claim: in deep corrections, the S&P 500 often finds support at the 50-month MA. When it breaks below and stays there, you're in a genuine bear market.
What we tested: SET monthly data 1982–2026 (490 monthly bars). Compared 6-month forward returns when SET was above vs below its 50-month MA, and identified all 14 cross-below events.
| Position vs 50-month MA | Months | 6m win rate | 6m median return |
|---|---|---|---|
| Above 50-month MA | 335 | 60.3% | +2.73% |
| Below 50-month MA | 142 | 56.3% | volatile |
The 14 cross-below events included every major SET bear market: - 1996-07: −20.2% forward 6m — Asian bubble peak - 2008-07: −35.4% forward 6m — Global Financial Crisis - 2019-11: −15.9% forward 6m — pre-COVID deterioration
When SET crossed back above the 50-month MA, forward returns were consistently positive.
Result: CONFIRMED as a bear market identifier. Above the 50-month MA = 60.3% win rate, consistent returns. Cross-below = every major SET bear market in 40 years is in this list. The MA doesn't predict the exact bottom — it tells you whether you're in a structurally hostile environment.
Test 5: "RSI Divergence at Weekly Lows = Weakening Sellers"
Source: Financial Wisdom (signal 1)
The claim: when price makes a new 13-week low but weekly RSI does not confirm (bullish divergence, RSI < 45), downside momentum is weakening. Forward returns are better.
What we tested: SET weekly data 1982–2026. Found 124 bullish divergence signals and 144 confirmed lows (price low + RSI low together). Measured 60-day forward returns.
| Signal | N | 60d mean | 60d median | 60d win rate |
|---|---|---|---|---|
| RSI divergence (bullish) | 124 | +4.31% | +0.81% | 53.2% |
| Confirmed low (no divergence) | 144 | +0.63% | −0.15% | 49.3% |
Result: CONFIRMED — directional signal, not a precision trigger. Divergence shows +0.96pp median lift and +3.9pp win rate vs confirmed lows. The article's caveat holds: "RSI divergence can persist far longer than expected." The signal says sellers are weakening — it doesn't say buyers have arrived yet.
Test 6: "Leaders Bottom 3 Days Before the Index"
Source: Brian Shannon (lesson 6)
The claim: the strongest stocks bottom approximately 3 days before the broad market index. Money slowly rotates into leaders before the index confirms.
What we tested: Major SET corrections >15% since 2005. For each correction, we identified the SET index trough date and the median trough date of the top-quartile RS stocks.
| Correction | SET index trough | Median leader trough | Leaders ahead by |
|---|---|---|---|
| 2019-07 → 2020-03-23 (−41%) | 2020-03-23 | 2020-03-23 | 0 days |
| 2023-01 → 2024-08-06 (−25%) | 2024-08-06 | 2024-08-07 | −1 day |
Result: DOES NOT TRANSFER TO THAI MARKETS. On Thai data, leaders bottom at essentially the same time as the SET index — not 3 days ahead.
Why? SET is a smaller market. Foreign institutional flows can override individual stock leadership signals in the short term. A large fund buying or selling an index basket moves everything simultaneously. Shannon's observation is accurate for the US — where thousands of institutional participants create the "rotation" effect — but the Thai market's structure doesn't produce the same early-warning timing.
What this means in practice: We still monitor individual RS strength rather than waiting for the index (the underlying principle holds), but we don't rely on individual leaders as a precise 3-day leading indicator for SET timing.
What Confirmed vs What Didn't
| Claim | Confirmed on Thai data? |
|---|---|
| Stage 4 avoidance (200d MA filter) | Yes — +1pp win rate, +1pp mean return |
| Bearish EMA cross = reduce exposure | Yes — 60d return drops to near zero |
| Monthly reversal candle | Yes — +9pp win rate boost |
| 50-month MA as bear market identifier | Yes — all 3 major SET bears in the list |
| RSI weekly divergence | Yes — +3.9pp win rate, directional signal |
| Leaders bottom 3 days before index | No — Thai market structure prevents this |
Five of six confirmed. The one that didn't transfer gave us a useful finding: Thai market regime filtering needs to lean more heavily on market-level signals (50-month MA, EMA crossovers) rather than individual stock leadership timing.
Why This Matters
Most trading content states rules without testing them. The ones that survive 40 years of data across a different market — SET vs US equities — aren't style preferences. They're describing something structural about how markets work.
Stage 4 kills accounts. Regime filters prevent you from fighting a hostile market. Monthly reversal candles tell you when buyers finally overwhelmed sellers at a major low. These patterns hold in Bangkok as well as they hold in New York.
But timing assumptions that depend on specific market microstructure — like the 3-day leader lead — need to be re-tested in each market. That's the job of quantitative validation, not assumption.
MOEasymmetry's entire framework is built on this principle: every rule, every filter, every exit must survive walk-forward testing on Thai data before it enters the system. The 3 traders in this article did the same thing intuitively over decades of real trading. We just ran the numbers to confirm it.
All tests run on SET daily data 1982–2026 (10,875 bars) and Thai backtest records 2007–2026. Statistical findings are directional — forward returns don't account for slippage or position sizing. This is educational research, not investment advice.