In a Financial Times opinion piece yesterday, Gillian Tett (I’m a fan) calls attention to the apparently extreme near-term bet in the oil derivatives market that the crude price will continue to rise. Referencing oil maven Philip Verleger, she points out that this outsized position has been built up, not directly by humans, but indirectly through AI-driven computers. The motivation for continuing the buildup? …price and volume momentum built up by previous AI activity. The fact that prices are rising, and on high volume, are enough for others to join the crowd.

Put a different way, the fact that prices are, say, 20% higher than a week ago and that many have already piled in, are both buy signals. This even though the winter heating season is for all intents and purposes over and the spring driving season has not yet begun–why February/March is traditionally the weakest season for oil demand.

This is the world we live in today. I presume at least some AI strategies make money. They may not make the most money, and their gains may come at the expense of other, slower AI rather than you and me. On the other hand, AI doesn’t need high-priced industry-expert humans, steeped for years in the lore of a given investment domain, so costs must be low. So AI-induced extra market volatility is likely here to stay. The safest–and most sensible, I think–course is to try not to be influenced by AI forays into and out of areas we’re interested in.

Still, at the very least we probably want to think out in advance what we’ll do if, out of the blue, AI moves into an area we’re heavily invested in and pushes prices to what we judge is an unsustainable high. Do we hold no matter what? What, if anything would cause us to sell a portion of our holdings, intending to buy back later at a lower price? How much would prices need to decline for this to be an acceptable course of action?

I’ve also been thinking about the apparently relentless year-long selling that’s been going on in early stage tech stocks. I think there’s a wheat vs. chaff issue with this sector. And investor sentiment continues to be bad. On one hand, I’ve been noticing names that have declined enough to be trading at close to tangible asset value. On the other, while surprising, this is no guarantee that such stocks won’t continue to decline–my guess is that AI isn’t considering asset value at all. The British don’t call trying to time a bottom “catching a falling knife” for nothing.