This Market Factors starts with a CIBC analyst’s heretical idea to take some profits on bank stocks and buy lagging financials. Section two describes the lurking risk to the AI investment story out of China and the diversion covers a fascinating new study on the origins of language.

CIBC analyst Paul Holden found last quarter’s earnings for the Canadian banks solid enough but uninspiring. He is recommending some profit taking with proceeds directed to laggard stocks, including insurers, within the broader financial services sector.

Mr. Holden called fiscal second quarter earnings results solid and noted that market reaction was minimal. He is raising his sector revenue expectations by a marginal 1.0 per cent for fiscal 2027. He sees a market environment characterized by strong capital markets, higher-than-usual credit losses, small if any improvement in net interest margins (profits on loans) and efficiency improvements.

Overall, the analyst is “finding it increasing hard to identify opportunities for upside” in profit expectations for the major banks (outside of his employer, which he does not cover). In addition to a slowing economy, an increasingly competitive loan market is limiting growth. He expects year-over-year profit growth of roughly 11 per cent for the sector in the latter half of 2026.

The bank sector has handily beat the benchmark year to date. The S&P/TSX Bank index is up 21.6 per cent versus 10.1 for the TSX. The strong returns, however, have left valuations unattractive.

The current price to book ratio of 2.3 times is well above the ten-year average of 1.6 times. The average dividend yield at 3.3 per cent is below the ten-year government bond yield and well below the ten-year average of 4.2 per cent. The forward price-to-earnings ratio of 14.4 times is 3.2 standard deviations above the ten-year average.

Mr. Holden is recommending that clients rotate out of bank stocks and into more attractively valued stocks within financial services. He writes: “We advocate for buying financials with solid fundamental stories, including double-digit EPS growth, but that have lagged the banks. Within our coverage universe, that includes [Manulife Financial] and [Element Fleet Management]”.

For investors still comfortable in banks, Mr. Holden has an “outperform” rating on Bank of Montreal.

I described one risk to the AI investment story last month – the upcoming IPOs of OpenAI and Anthropic - and then Friday Scotiabank analyst Nat Schindler identified another substantial threat: China. Chinese developers are marketing smaller AI models at prices more than 65 per cent less expensive than comparable U.S. solutions.

The Chinese software products use more limited language models that do not need to refer to massive data sets, fed by huge data centres, sitting on the cloud, for normal operation. This reduces the overhead costs for Chinese companies.

Mr. Schindler sees the potential for Chinese AI products displacing U.S. ChatGPT and Claude for simpler applications. The more expensive U.S. computing will be reserved for more complex solutions, leaving them with less dominant market shares than are currently implied by U.S. hyperscaler growth forecasts.

The argument that the AI investment story is not analogous to the 1999-era tech bubble are in large part based on valuations of the most relevant tech stocks. Nvidia Corp., Microsoft Corp., Meta Platforms Inc., and Amazon.com Inc. are all trading with current forward price-to-earnings ratios much more reasonable than companies like Cisco Systems in early 2000. The earnings expectations that make the PE ratios palatable, however, are in many cases dependent on increases in the value of private holdings in AI specialists like OpenAI and Anthropic.

The U.S. government could limit Chinese software imports to protect the investments of the major tech firms. Even in this case, the adoption of Chinese solutions in the rest of the world would limit global market share for U.S. companies.

The combination of Chinese competition and earnings growth based on unrealized gains in speculative investments makes the recent volatility in megacap technology stocks entirely understandable. The entire story is starting to feel a bit wobbly.

I’m actually excited about this one: a development in the study of language from research from the University of Vermont. The research led to a new term, ousiometrics, that leads to the heart of human psychology and the roots of language and its evolutionary purpose.

The relevant research paper is summarized in Science Advances here. The description of the methodology starts with the previously accepted framework for language with its emphasis on - in this order - evaluation (the degree of positivity), potency (dominant versus submissive) and activity (active versus passive). The corresponding emotional responses to these stimuli are categorized as valence (degree of pleasure), arousal and then degree of dominance.

Using a massive data set including a test subject reacting to individual words, transcripts of talk radio, retweets on Twitter and thousands of books published since 1900 (the latter to measure frequency of usage for evocative words), the study found that the traditional model of evaluation, potency and activity didn’t work.

The study found that language revolves around two structures. In one, the degree of bad/good is closely associated with a determination of gentle versus aggressive. In the other structure, words are oriented around an assessment of weak versus powerful and a second judgement of safe or dangerous. In both structures, words are judged by the implied degree of structure – organized versus disorganized.

The word ousiometrics is based around the Greek word ousia, meaning essence. It’s why I find this topic fascinating – it gets to the root or essence of human psychology and what matters to us most.

Looking for our updates on market movers, analyst actions, stock technicals, insider trades and other daily, weekly and monthly insight? Click here to visit our Inside the Market page.

Jamie McGeever on Friday’s big selloff and how the AI rally may have finally met its match: the Fed

Tom Bradley on the warning signs grizzled veterans are detecting in the market right now

David Berman on why the Andrew Left securities fraud trial proves small investors are not helpless against the manipulative whims of big shots

Economist David Rosenberg thinks the bond market correction has seen its peak

Lululemon Athletica’s first quarter earnings were in line with expectations but the retailer guided revenue expectations lower for the second quarter and fiscal 2026. “Weakness in yoga” was cited as the culprit by management. Lulu is among the last of the pandemic winners to fall back to earth – Zoom Communications and Peleton got crushed earlier - and it’s now down 75 per cent from the 2023 highs at U$500.

Prime Minister Mark Carney’s defence spending plans are extraordinary. Scotiabank analysts led by Patrick Bryden recounted plans for $180-billion in direct defence procurement and $290-billion in defence-related capital infrastructure (production facilities primarily) by 2030. This amounts to a $47-billion per year expansion. Companies expected to benefit include Atco Ltd., Blackberry, Bombardier, CAE, CGI Inc., MDA Space Ltd. and Calian Group Ltd.

With an approach similar to BofA Securities’ Positive Triple Momentum stock screening strategy, Citi’s Drew Pettit is focused on “ROE Trend” – companies improving return on equity through higher profit margins and asset turnover. The improving ROE stocks have outperformed the negative ROE trend group by just under 10 percentage points year to date. The list of positive ROE trend stocks is long, with (in order of size of expected ROE improvement in 2027) Ciena Corp., Caterpillar Inc., Rockwell Automation Inc., GE Aerospace, Broadcom Inc., Corning Inc. and Analog Devices among the opportunities most likely to interest Canadian investors.

Read this week’s earnings and economic calendar here