Singapore stock prediction (SGX): the honest numbers and why it's hard

I run Trading Agent, and I publish every prediction my model makes — including the weaker ones — at /predictions. Most of the time, founders writing about their own product cherry-pick the markets where they look good. This is not one of those articles. Singapore is one of the markets where my model shows the most room to grow, and I think the honest version of that story is more useful than a sales pitch. The Singapore sample is still early and growing, and the model retrains as more predictions verify — so I expect the honest number here to firm up over time.

The number, plainly

On the Singapore Exchange (SGX), my model has logged roughly 177 verified directional predictions with an accuracy of about 42%. That's below our US benchmark, and on an early, still-growing sample a sub-50% reading is a real but modest bias — not a verdict carved in stone.

For context, my US predictions sit around 54% over roughly 830 verified calls, and my blended accuracy across every market is about 46% over roughly 2,248 predictions. Singapore sits below our average, not above it. I'm not going to dress that up. The model is still developing its edge here, and I'd be doing you a disservice if I implied otherwise.

The interesting question isn't whether it lags — the verified record shows that. The interesting question is why. And the answer is almost entirely about market structure.

Why a momentum model has a harder time in Singapore

My model, like most machine-learning systems applied to equities, is fundamentally a pattern-and-momentum engine. It looks for trends, continuation, and the statistical fingerprints of price moving in a direction and keeping moving. That approach works best in markets full of momentum-driven single stocks with deep domestic retail participation. Singapore is close to the opposite of that.

It's a small market dominated by yield and defensives. A large share of SGX's weight sits in Singapore real-estate investment trusts (S-REITs) and yield-oriented, defensive blue-chips — the big local banks being the obvious examples. These are names people hold for the distribution, not for explosive price appreciation. Structurally, they're designed to be stable. A REIT that pays a steady, attractive yield is doing its job when it doesn't move much. That stability is exactly what starves a momentum model of signal.

Flow is institutional and regional-hub driven. Singapore functions as a regional financial hub, and a lot of the order flow is institutional and cross-border rather than domestic-retail momentum chasing. Retail momentum is noisy, but it's trending noise — it's the kind of behaviour a model can latch onto. Institutional, hub-driven flow tends to be more measured and less directionally persistent at the single-stock level. There's simply less of the herd behaviour my model is built to detect.

Many names are dual-listed or run offshore operations. A meaningful chunk of SGX-listed companies are dual-listed or have substantial offshore operations, so their prices react to regional and global macro forces more than to anything happening locally on the chart. When a stock's next move is dictated by a Fed decision, a China property headline, or a shift in regional rates, the local technical pattern my model reads becomes close to noise. The signal it thinks it sees is being overwritten by exogenous macro it can't observe.

Single-stock volatility is relatively low. Put the above together and you get relatively low domestic single-stock volatility. Less movement means fewer clean directional moves to predict, and a higher proportion of what movement does happen is mean-reverting around a yield level rather than trending. Yield-driven defensive names just don't trend the way a momentum model wants them to.

So when I describe a name like an S-REIT or one of the major banks, I'm using it as a structural example of this dynamic — not as a pick. The point is the category behaviour, not the ticker.

What I actually do with a 42% market

I label Singapore signals Bearish, Neutral, or Bullish — never Buy or Sell — and on SGX I treat them with heavy skepticism, because the verified data tells me to. A signal that lands below chance isn't a contrarian goldmine you can just invert; a sub-50% reading on a noisy base usually means the model is reacting to structure it doesn't yet read well, not that it's reliably wrong in a tradeable way. The honest conclusion is narrower and less exciting: in a yield-heavy, offshore-flow market, the technical signal is faint for now, and I tell you so rather than hiding it behind an average.

This is the whole reason I built the brand around radical honesty. Plenty of AI stock tools quietly bury their losers — I wrote about that pattern in why most AI stock-picking tools are lying. I'd rather show you a 42% and explain the market structure behind it than show you a polished number I can't stand behind. If you want the full breakdown of how the model is built, scored, and verified, it's all on the methodology page.

Singapore is, for now, a market where my model's best contribution is honesty about where it stands. That's not the answer a marketer wants. It's the answer the data supports — and the sample is still growing.

---

This article is educational content about machine learning and market structure. It is not financial advice, not a recommendation to buy or sell any Singapore-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).