⚠️ Educational research and a personal trading journal — not investment advice. การศึกษาเท่านั้น ไม่ใช่คำแนะนำการลงทุน. Past performance is not indicative of future results. The author does not provide licensed advisory services. Consult a licensed adviser before any trading decision.
Most trading content shows you the wins. This one shows you what failed — including three things I personally believed were true and turned out to be wrong.
I trade a specific breakout setup. Not VCP exactly, not a textbook cup-with-handle exactly — my own read, built from years of staring at charts: a stock pulls back, tightens, makes a recovery high, and breaks through it on a surge of volume. I can recognize it by eye in seconds. The question that keeps me honest is the one most people never ask about their own method:
Does it actually work — across decades, after costs, when I can't cherry-pick?
So I encoded the setup into rules and ran it across every Thai stock, 1990 to 2026 — 882 names, 2.87 million bars, point-in-time, no look-ahead. Here's the honest report.
The setup, in one paragraph
Above the 200-day average (the stock is in an uptrend). Relative strength in the top fifth of the market. Each pullback into the base shallower than the last — the tightening is the signal, not the depth. Volume dries up into the pivot, then expands on the breakout. Buy within 5% of the pivot; stop at the last higher-low or 7%, whichever is tighter. Then let it run on a trailing stop.
Track. Study. Wait. Strike.
What's true: a real edge — but not the kind you're sold
After realistic transaction costs, a liquidity filter (no untradeable penny names), and using my actual exit (take partial profit at 2R, trail the rest under the 21-day average), the setup returns about +0.35R per trade for leaders in a confirmed market uptrend. Filtering to the very top tier of relative strength pushes it toward +0.47R.
And I ran the test that most people skip: I bootstrapped it — re-sampled the trades 1,500 times to build a confidence interval. The 95% interval excludes zero. The edge isn't a lucky streak; it's statistically real. (When I ran the same test on US stocks, only one configuration survived — top-fifth relative strength, in a confirmed uptrend. Outside that, the "edge" was indistinguishable from noise. The market matters more than the setup.)
That's a genuine edge. But look at how it's earned:
- Win rate: ~43%. The typical trade is a small loss that gets stopped out.
- Median trade: −1R. More than half are clean stop-outs.
- The entire profit comes from the right tail — the few trades that run far. Cut your winners short and the edge evaporates.
This is the part no one wants to hear: a good breakout method is mostly small losses punctuated by occasional big wins. The discipline to take the small losses and let the rare winners run is the edge. There is no version where you're right most of the time.
What I was wrong about — three falsifications
Here's where it gets useful. I held three beliefs going in. The data overruled all three.
1. "Sell when it's stretched ~100% above the 50-day average." This is a real O'Neil-style exhaustion signal, and the level is eerily real — I watched a US leader top at +99.95% above its 50-day average and reverse the next day almost to the cent. So I added it as a sell rule and re-ran everything. Result: it slightly lowered returns. Selling into the climax caps the parabolic winners that the trailing stop would otherwise ride further. It's a giveback-avoider, not a return-booster — useful for peace of mind, not for the bottom line. I'd been treating a risk tool as an alpha tool.
2. "The longer the base, the more explosive the breakout." Trading folklore I'd repeated for years. The data says it's an inverted-U: medium bases (5–9 weeks) are the sweet spot, and very long bases (13+ weeks) actually underperform — slightly negative. The truth underneath: a clean, tight long base does explode (I have the examples). But most long bases are loose and faulty, and averaged together they drag. It was never about length. It was about quality. O'Neil said this 40 years ago; I had to lose the argument to the data to believe him.
3. "CANSLIM fundamentals will sharpen the edge." Adding O'Neil's earnings filter (strong current + annual EPS growth) on top of the price setup should lift returns 30–50%. It did the opposite — it actively hurt. The alpha lived in the small/mid-caps that don't even have clean earnings coverage. Filtering for textbook earnings excluded the winners. The edge lives in the structurally-inefficient corners, not the names everyone already screens for.
The money question: if you put in 1 million baht
I simulated the portfolio with different position sizes. The result is a clean lesson in risk:
- Risk 0.25% per trade → smoothest ride (−9% max drawdown) but modest growth.
- Risk 1% per trade → highest return, but you stomach −27% drawdowns.
- Risk 2% per trade → returns fall. You're over-betting; drawdowns compound against you faster than gains.
The sweet spot — best return for the pain — was around 0.5% risk per trade. Bet small. The edge is real but thin and lumpy; size it like it could be wrong, because half the time it is.
And one more honest flag: the method bled in 2022–2025 — the long Thai drought. It's a market-regime strategy. In a downtrend, even clean breakouts fail. The market is not a backdrop; it's a filter.
Why publish the failures?
Because the failures are the moat. Anyone can post a winning chart. Almost no one shows you the 36-year test where their favorite rule got falsified. The willingness to run the test, report the −1R median, and admit three of my own beliefs were wrong — that's the methodology. The edge isn't a secret indicator. It's the discipline to keep testing what you believe and discard what doesn't survive.
Track the patterns. Study what actually made winners work. Wait for the setup and the market. Strike when the math — net of costs, across decades — is on your side.
That's the whole game.
This is a personal research journal documenting my own process and tests. It is not investment advice and the author does not provide licensed advisory services. Markets carry risk; you are responsible for your own decisions. — MOEasymmetry