Taiwan’s stock market has climbed to seventh in the world by market capitalisation – up from seventeenth just a few years ago – and now sits at roughly US$3.4 trillion. Taiwan Semiconductor Manufacturing Company (TSMC) alone is among the ten largest public companies on the planet. Global fund allocations are rising in lockstep with the index weight. But for alternative data practitioners, the more interesting story isn’t the market’s size, it’s the market’s structure. More than half of all trading value on the Taiwan Stock Exchange comes from retail investors, intraday activity runs above 40% of daily average turnover, and transaction taxes are a fraction of those in comparable markets. That combination of global relevance and persistent retail dominance makes Taiwan an unusually rich hunting ground for behavioural signals. And one firm claims to have the infrastructure to capture them at scale.
At the A-Team/Eagle Alpha Alternative Data Conference in New York in March 2026, Jack Yeh, Chief Operating Officer of CMoney, made the case that Taiwan’s retail investor ecosystem is a source of tradable alpha that remains underexplored by the global buy side. CMoney, a Taiwanese fintech company founded in 2003, operates what it describes as the largest retail investor ecosystem and consumer transaction platform in the country, with reach across more than 56% of the population.
A government tax hack turned alt-data goldmine
The most distinctive element of CMoney’s data offering – and the one likely to be least familiar to a Western audience – is its consumer transaction dataset. Its origins lie in Taiwan’s Uniform Invoice system, a government mechanism dating back to 1951, designed to combat merchant tax evasion. Every business transaction above a threshold generates a standardised invoice carrying a lottery number. A bi-monthly government-run prize draw incentivises consumers to collect and store these invoices, historically on paper but increasingly through digital carriers.CMoney operates one of the largest digital invoice carrier platforms in Taiwan. According to Yeh, the company processes transaction data from roughly 20% of the total consumer population in real time, at SKU level, covering around 120,000 retail locations. The SKU granularity is notable: unlike credit card transaction data, which typically captures merchant-level spend, the e-invoice data captures individual items purchased. And unlike e-commerce platform data, it spans both online and offline retail, including the substantial cash economy that remains a feature of Taiwanese commerce.
The dataset maps to around 80 of the roughly 150 listed equities in Taiwan’s consumer and retail sector, including the largest food and beverage, retail, and e-commerce names. Critically, CMoney argues, the data also provides a window into several globally listed companies for which Taiwan is a material revenue driver. Yeh cited three examples. Shopee (Sea Group), where he estimated Taiwan’s GMV at 15–20% of the global total, a figure broadly consistent with third-party analyst estimates and the platform’s status as Shopee’s second-largest market after Indonesia. Coupang, which Yeh described as having Taiwan as its largest overseas market, a claim well-supported by the company’s own earnings commentary. And Uber, where Yeh said Taiwan ranks as the second-largest market by order flow and frequency, though what is publicly documented is that this ranking applies specifically to Uber Eats rather than Uber’s operations as a whole.
In a concrete example of the data’s nowcasting potential, Yeh described a discrepancy during the most recent Double 11 shopping event (Taiwan’s equivalent of Black Friday). A major e-commerce platform publicly claimed double-digit sales growth. CMoney’s invoice transaction data showed year-on-year growth of only 3–4%. When the company subsequently filed its financial results, the actual figure came in at around 4.5%.Broker flows: the Pentagon Pizza Index, Taiwanese edition
CMoney’s second signal layer draws on the island’s dense brokerage infrastructure. According to Yeh, Taiwan has more than 800 brokerage branches and CMoney claims access to daily directional order flow data from this network, covering close to 100% of listed securities. The data provides two dimensions: branch-level order flow and ownership shifts.
The branch-level data introduces a locational signal that Yeh compared to the well-known “Pentagon Pizza Index,” the informal observation that a spike in late-night pizza deliveries near the US Department of Defense can signal an uptick in geopolitical activity. In Taiwan’s case, if order flow clusters around a company’s headquarters or an industry zone, it may flag that insiders or local participants are acting on information not yet reflected in the broader market. Meanwhile, the ownership shift data helps distinguish between names that are predominantly retail-held and those with concentrated institutional or strategic positioning, a structural signal relevant to both momentum and mean-reversion strategies.
Sentiment: from community chatter to pre-trade behaviour
The third layer is retail sentiment, drawn from CMoney’s stock community platform. Yeh described this as the largest in Taiwan, claiming it covers around 45% of monthly active traders. The platform captures discussion topics, attention levels, and trending tickers. But CMoney goes further than conventional social sentiment analytics: because it also operates trading tools that (according to CMoney’s own metrics) are used by roughly 20% of monthly active traders, it can observe pre-trade behavioural signals – actions like adding a ticker to a watchlist or analysing a stock’s chart pattern – that indicate intent before it becomes flow.
The intersection of these signal types is where CMoney claims the analytical edge sharpens. When attention spikes in the community coincide with order flow clustering in the brokerage data, for instance, the probability of sustained momentum increases. Yeh framed this as a two-by-two matrix: high attention with high order flow versus various low-signal combinations, each with distinct implications for price and volume dynamics.
So what does the buy side actually do with this?
The practical question for quant researchers, data scientists, and market data managers at institutional firms is how these signals translate into operational workflows.
CMoney is positioning its data as a multi-layer behavioural dataset for Taiwan-focused or Asia-Pacific strategies, with the consumer transaction data also serving as an earnings nowcasting tool for several globally relevant names. The firm already serves more than 400 financial institutions domestically, holding what it claims is a market share above 90% in financial data services in Taiwan. The expansion pitch to global buy-side firms, delivered explicitly at the conference, centres on three use cases: earnings prediction via consumer transaction data; momentum and reversal detection via brokerage flows; and early-stage sentiment detection via the community and pre-trade behavioural layer.
For data buyers evaluating the proposition, the usual due diligence questions apply, but the starting point is stronger than for many alternative data vendors. The e-invoice data’s provenance is unusually transparent, originating from a government-mandated system rather than a proprietary panel, and CMoney’s dominance of the digital carrier market gives it genuine scale. That said, prospective buyers will still want to understand panel representativeness, SKU-level mapping methodology, and data latency. On the brokerage flow side, the natural questions concern how granular the branch-level data is, how consistently it is reported across brokers, and what compliance frameworks govern redistribution. And for the sentiment layer, the perennial challenge of distinguishing signal from noise in community-generated content still applies, although CMoney’s ability to pair community activity with pre-trade behavioural data from its own tools is a meaningful differentiator from pure social-listening approaches.
Taiwan’s rapid ascent in global index weight means these are no longer niche questions. As the TWSE set six record highs in 2025 and the exchange publicly targets overtaking Canada for sixth place globally by capitalisation, the market’s relevance to global allocators will only increase. Whether the retail behavioural signals CMoney captures can be reliably validated and integrated into institutional workflows is an open question. But it’s the right question for the industry to be asking.
Subscribe to our newsletter
