top of page

Artificial Intelligence





OpenAI, a Silicon Valley start-up whose original funders included Twit-in-Chief Elon Musk, has recently made headlines with a beta test launch of the latest iteration chatbot, ChatGPT. It's an updated version of its query-based content generator in the rapidly emerging AI space. The big difference with this latest version is that it can admit when it's wrong. Think of it like Google on steroids, only with humility.


As computer-generated content starts to go mainstream, there is a risk of expecting too much from technologies such as artificial Intelligence. Early versions of OpenAI's software were prone to mistakes that were not readily recognized. Additionally, dangerously provocative and politically incorrect outcomes were left unchecked. The newest version attempts to remediate these shortcomings by including the ability to admit mistakes and to self-censor. But since the invention of the computer, the programmer's credo of "GIGO" - garbage in, garbage out, still governs the utility of any artificial device made by humans.


ChatGPT can produce a college term paper, write a poem or song, and even generate a blog post. You may actually be reading one soon as I have signed up to participate in their beta test. Tuesat11 could soon be the thoughts of a piece of silicon instead of ones previously formed by my rapidly decaying grey matter. Retirement just gets better every day! But will those cyber-missives actually generate better investment recommendations simply because they are computer created?


I have often talked about how 'narratives' motivate investor actions when it comes to markets. The quest to outperform the S&P500 is a difficult and fleeting one. More often than not, we are caught unaware of the influential shift in narrative that leads to changes in accepted wisdom. That is why I studiously adhere to the mantra of trying to expect the expectations of others. Markets conform to those expectations, even though they may be wholly erroneous from conception.


But mispricing generated by groupthink and investor bias often follows from the very narratives that drive investment decision-making. Can we expect a chatbot created by software programmers that use data available to everyone, to be immune to such bias? Not likely. It's one thing to write a term paper summarizing historical events or generate a song that draws on Spotify's 250,000 examples, but quite another to try to predict the future path of investor expectations.


The latest market rally could well be a perfect example of that process. A 2023 recession that leads to a Fed pivot, that leads to lower rates - is now a 'done deal' consensus. The rally off the lows since Mid-October was aided and abetted by an under-positioning and an excess of pessimism. That was behind my recent views of a market that does better than the economy. But now, with the economy stubbornly refusing to collapse, an uneasy complacency has replaced that pessimism. Calls for a soft landing are increasing, yet history is not supportive of such an optimistic conclusion. And recent employment data has pushed out the recession call even further.


But with a yield curve so highly inverted and liquidity ebbing for the foreseeable future, I don't like the chances of a soft slowdown scenario continuing to prevail much longer. Unfortunately for the bears, markets are reflecting that soft-landing narrative as if it is a lock. And if an AI-based investment blog were to be asked for a market call, the soft landing consensus and a Fed pivot would be inevitably produced by the 'intelligent' bot. But don't look for an insightful prediction. It can't use data that hasn't been generated yet.


So bad news for those looking for AI to help them navigate the markets. The computer will still generate a bad call just like any other investor. Artificial intelligence is like an artificial Christmas tree. It may be easier to create than the organic kind, but it still looks like the same tree.




Risk Model: 5/5 - Risk On


The Model is pretty sure of itself these days. The market is not overbought enough to be dangerous. Volatility expectations have dampened down after the Fed pivot narrative has now become widely expected. The Chinese 'restart scenario' has turned Copper/Gold slightly higher - for now. And AAII sentiment has seen a dead cat bounce off a decade-low bottom endured since the 2022 sell-off began. But each of these indicators has the potential to turn on a dime here should the narrative of a benign economic outcome leading to easier money start to waiver. It wouldn't take much to restart the rush to the exits, especially when the calendar flips the page to a year fraught with risk, as next year portends.


We are in a box here. The yield curve is the critical element now. Think of the shape of the curve as a business. When your cost of carry is negative it's like your cost of goods sold is higher than your selling price. Not a sustainable business model. Something's gotta break - I just don't know what yet.


If we see a 'dis-inverting' of the curve - from either direction - it won't be easy for stocks to hold up. Higher bond yields in the long end that come from a stronger-than-anticipated economy will crimp valuations. Lower front-end rates reflective of a much worse economy would come at the expense of earnings. I'm in the 'bear market rally' camp still, and it's getting a bit long in the tooth now.


This week's chart that caught my eye is courtesy of Greg Taylor of Purpose Investments who points to a pattern that, if continued, would lead to some degree of market duress in the near future. SQQQs anyone?



VIX, SPY -Trendlines


Comments


bottom of page