TW stock prediction with machine learning: the honest record and why it's hard

If you trade Taiwan-listed stocks — TSMC (2330), MediaTek (2454), Hon Hai (2317) — you've probably noticed that almost every "AI stock prediction" product on the internet quietly skips your market. They cover US large-caps, maybe a few other developed names, and stop.

We don't skip Taiwan. We run the same machine-learning pipeline on TWSE-listed large-caps that we run on the S&P 500. But I'm not going to pretend it's where our model is strongest, because it isn't. This is the honest version.

And let me be clear up front about what "honest" means here. The record below isn't from a flattering backtest. Backtests are always optimistic — you're fitting and measuring on history that already happened. What matters is the live, forward record: predictions we made, timestamped, before the outcome was known, then scored. That live record came in lower than any backtest would suggest, and it's the only kind I'll cite.

The record: below break-even on TW, so far

Across the verified Taiwan predictions in our public log, our live directional accuracy on TW large-caps has been running below even — one of our weaker markets, full stop.

I'm deliberately not printing the exact figure here: a number pasted into a blog post starts rotting the day it's published. The live, always-current one — market by market, with every resolved call behind it — is on the public self-audit, and every individual scored call is at /predictions.

I want to sit with "below even" for a second, because most of our competitors would never publish it. It means that on a coin-flip question — will this stock close up or down — our model on Taiwan has been landing slightly on the wrong side of the coin.

For context, blended across all 16 markets we cover, our live directional accuracy sits roughly at break-even — which is exactly what you'd expect from an honest tool that isn't cooking its books. Our strongest markets right now are Canada and the US, where the live record shows a small real edge — nowhere near hype territory. Taiwan sits well below that blend, near the bottom of the table.

The exact same model architecture, the same features (RSI, MACD, moving averages, volume, volatility), the same walk-forward validation produces a small, real edge on Canada and US large-caps and a below-even result on Taiwan.

Why? This is the interesting part, and it tells you something real about how markets work.

Why Taiwan is harder to predict

1. Retail-dominated order flow

Taiwan's market has one of the highest retail-participation rates in the world — historically well over half of daily turnover comes from individual investors, versus a much more institution-dominated US market. Retail-heavy flow tends to be noisier and more sentiment-driven on short horizons. The technical patterns a model learns from price and volume are weaker signals when the marginal trader is reacting to a LINE group rather than a discounted-cash-flow model.

2. Day-trading mechanics and tax incentives

Taiwan has actively encouraged day trading through reduced transaction-tax rates on same-day round trips. That concentrates a lot of activity into intraday mean-reversion that washes out by the close — which is exactly the kind of move a daily-horizon model has a harder time anticipating. The signal the model is trained to find (continuation) and the behaviour the market actually exhibits (intraday reversion) are working against each other.

3. Concentration and correlation

A huge share of the TWSE's market cap and movement is TSMC and the semiconductor supply chain. When one name and one sector dominate, a lot of "individual stock" prediction collapses into "are you right about TSMC and the global chip cycle this week" — which is driven by US customer demand, foreign fund flows, and geopolitics, none of which live in a price chart. (To be clear: I'm using TSMC and its suppliers here as a structural example of why the market is concentrated, not as a pick. Nothing in this article is a recommendation to buy or sell any of them.)

4. Data quality and corporate-action noise

yfinance and other free data sources handle TWSE corporate actions (the frequent ex-dividend and capital-reduction events common in Taiwan) less cleanly than US data. Some of the model's TW weakness is almost certainly noise injected by imperfect adjusted-close handling, not pure model failure. We're investigating how much of the below-even gap is fixable data hygiene versus genuine market efficiency — but I won't pretend the data alone explains it. A below-even live record across a sample this large is a structural finding, not a rounding error.

What this teaches you

Here's the lesson worth more than any prediction: a market being "hard to predict" is itself information.

If a simple technical model lands below even on Taiwan large-caps across a sample this size, that's evidence the market is reasonably efficient on short horizons — the easy patterns have been arbitraged away by all those active retail and prop traders. That's not a bug in the market; it's the market doing its job.

It also means you should be deeply suspicious of any tool that claims high accuracy on Taiwan stocks. If our honest, forward-tested record sits below even, and someone else is advertising "90% accuracy on 台股," ask them the question from our piece on AI stock-picking tools: show me the complete, timestamped, loss-included live log. They won't have one. What they'll have is a backtest — and as I said, backtests always flatter.

So why cover Taiwan at all?

Three reasons:

1. The tool is still useful as context, not as a signal. The app never tells you to buy or sell anything — it shows indicators, analytics, and the scored public record, never an instruction and never a guarantee. On a market where that record honestly shows our patterns don't hold, the takeaway is "trust your own thesis here, not ours." That's a legitimate, honest use.
2. The other tools available to TW retail investors are technical-indicator dashboards with no honesty layer at all. A research tool that shows its own weakness on your market — and publishes the below-even record in public — is more useful than one that hides it behind a marketing figure.
3. We're a New Zealand company built for the markets the bigger tools ignore — Taiwan, Japan, Korea, Vietnam, Malaysia, Singapore, Indonesia, Thailand, Hong Kong, India, alongside the US, UK, Australia, Canada, China and our home NZX. That's the full set of 16. Covering them honestly, including where we're weak, is the whole point.

If you want the markets where our model is currently strongest, that's Canada and US large-caps — see the methodology page and the per-market breakdown at /predictions. And if our Taiwan record ever improves as we clean up the data handling, you'll see it move on that same public log, in real time, with nothing hidden — and if it stays below even, you'll see that too.

See the evidence for yourself — download the full resolved-prediction dataset, read the live public self-audit, inspect every model card, or run the research tools on your own data. No hype, just the receipts.

---

This article is educational content about machine learning and market structure. It is not financial advice, not a recommendation to buy or sell any Taiwan-listed or other security, and not directed at any individual's circumstances. Trading Agent is a quantitative research tool operated by WU Capital Limited (New Zealand).