By Francis Muyshondt.
Karsten Seier has been a quant portfolio manager all his professional life. And today, he co-manages all US momentum strategies* and active ETFs at asset manager ODDO BHF Asset Management. For him, all the required information that is necessary to select the best stocks is in the price development. From his office in Dusseldorf, he studies that data and the behaviour in the markets. And if Seier is to be believed, it is all characterised by observable patterns, though not reliably predictable We spoke to him in the heart of Brussels on the sidelines of a presentation to clients.
Available data
Momentum is among the best researched market patterns in financial academia. Since the 1990s, studies have consistently shown that stocks that have recently outperformed often continue to do so for some time. Yet it remains difficult for many investors to accept. The idea clashes with the intuitive view of rational markets. Seier understands that mistrust. 'It sounds too simple: look at the price and follow the trend. But just because it's simplicity doesn't mean it doesn't work in certain cases.' Jokingly, he states that even children could run his model – meaning this model is generally straightforward to apply as the underlying data is publicly available. 'However, the difference is not in the information, but in the discipline with which it is applied but it does not ensure positive results ,' he states with a laugh.
Most momentum strategies look at data and performance over 6 to 12 months. This produces portfolios with high turnover rates and high sensitivity to news. However, ODDO BHF deliberately chose a longer period, namely three years. 'A stock that outperforms the market for three years usually does so through fundamental improvement: earnings growth, product innovation, positive news. We measure fundamental momentum without using fundamentals.' As a result, stocks stay in the portfolio for two to two and a half years on average, closer to classic stock pickers than traders. This slowness is deliberate: short trends arise from rumours, long trends from a favourable underlying business trend.
The trend is your friend
Active ETFs are often presented as a middle ground between index investing and stock picking. For Seier, those trackers are just another package. 'An ETF is just a vehicle. The portfolio process is identical to that of our funds and institutional mandates.' Yet there is an irony in that. According to him, passive investing just reinforces the momentum he is betting on. When a stock rises, its weight in the index rises. So index funds automatically buy more from winners and less from losers. The herd makes the pattern stronger. 'Index investing can pushes trends further than normal and that's good for us, however, this does not imply consistent advantages for any given strategy."
Many factor strategies such as momentum are used by investors as satellites, a smaller position alongside a core portfolio. Seier correctly positions momentum as a core investment within a diversified allocation. The reason is technical: his portfolios hold a market beta around 1 and volatility similar to the S&P 500. The goal is not a style bet, but a more efficient index. 'Such an approach aims to avoid structurally weak stocks, although results may vary and are not guaranteed.'
When does it go wrong?
Momentum, however, does not always work, confesses Karsten Seier. It does less well when the market suddenly changes direction. Seier sums up this period loosely from memory. 'In 2003, the market suddenly rotated into cyclical and heavily discounted value stocks after the bear market of 2001/2002. The same thing happened in 2009 after the financial crisis: banks and economically sensitive companies surged while defensive quality companies lagged behind. In 2016, multiple rotations followed in short order (commodity recovery, Brexit and the election of Donald Trump) causing markets to change leaders three times in one year. And in 2022, the interest rate regime turned: rising inflation and higher interest rates caused growth stocks to abruptly underperform value,' explains the ODDO BHF AM manager.
'During such periods, investors suddenly rediscover laggards and sell en masse the stocks that were dominant for years. For a strategy just betting on continuity, that is the most difficult scenario: not a crash, but a reversal. Not panic, but reassessment.' Therefore, his model also includes a second component: risk management. Sector-wise, the portfolio stays close to the index and avoids concentration. Without that constraint, it would have consisted almost entirely of Big Tech in recent years and be hit hard at a sudden turnaround. 'The biggest danger to momentum is concentration risk. We systematically try to avoid that without a guarantee of eliminating it entirely.’
Uncertainty is good
Paradoxically, crisis days are often comfortable for the quant. Where fundamental managers have to make decisions, the model does nothing, and that is precisely why it works. 'During periods of stress, behavioural errors increase because when uncertainty rises, investors become more emotional. They are selling what had recent bad news and buying what remains relatively strong. That amplifies differences between stocks, and it is precisely those differences that fuel momentum strategies. So fear creates strong trends, and we can try to exploit them; however this does not ensure any positive performance outcome' He adds that it is not the direction of the market that matters, but the spread within the market. Flat scholarship can be difficult, chaotic scholarship often not.
Portfolios managed by Seier are updated only once a month. The rest of the time, the team maintains the infrastructure, checks data and talks to customers. 'Human input is mainly in the development of the model, not in the decisions. There is no discretionary override. What the model says, we implement. This does not mean that man disappears, but that he changes roles. The manager no longer decides on individual shares, but on rules, parameters and risks. He designs the process, then the process decides.'
AI use
In a sector where just about every asset manager claims artificial intelligence in its strategy, Seier remains remarkably cautious. His team is exploring AI, but is not using it in the investment process for now. 'The first reason is substantive. When AI models analyse market datasets, they often automatically arrive at patterns that closely resemble momentum: stocks that are already rising continue to rise, and vice versa. In other words, the technology is trying to detect the same behaviour that our models already measure directly. The added value is therefore limited.' In addition, he wants to avoid making the model unnecessarily complex. The more complicated the algorithm, the more difficult it becomes to understand why a stock is bought or sold.
To this is added a second, equally important factor: the investor. 'Institutional clients in particular are suspicious of so-called black boxes. They want to be able to intuitively understand how a portfolio is created. They want to see not only the result, but also the logic behind it. A simple model based on price trends can be explained. A self-learning algorithm combining thousands of variables much less so. And just that trust often determines whether capital is entrusted.'
Key indicator
At the end of the conversation, Seier formulates his philosophy in one sentence that defies any financial disclaimer: 'It is said that past performance says nothing about performance in the future. For equities, past performance can provide contextual information, as it may reflect how investors have reacted in previous market environments. However, it does not constitute a reliable indicator of future results.' He does not claim that markets are predictable, only that investors remain predictable. Investors are slow to react to information, adjust expectations gradually and follow winners longer than seems rationally logical.
'Momentum is therefore not a prediction of business results, but of human reaction patterns. Not what happens in companies, but what happens in heads. And as long as psychology evolves more slowly than technology, a simple observation will continue to work: those who win today often may continue to win for a while. However, there is not guarantee of continuing this trend. So the quant does not need a crystal ball.'
