The Federal Reserve meeting this month was keenly awaited to see if the hardline message on inflation control materialized into the expected 75bps rate increase or the less probable 100bps. It was the indeed the former, accompanied by a strong message of staying the course to control inflation long-term, at the cost of short-term pain.
This was all very much in line with expectations. Yet, by the end of the week, after the dust settled on the announcement, the major US indices were down significantly: the Dow dropped -3.6%, the S&P 500 -4.2% and the NASDAQ -3.8%. The rate rise matched the consensus figure and not the more concerning 100bps increase. The Fed’s messaging has been consistent, from Jerome Powell’s Jackson Hole speech to his interview at the Cato Institute earlier in the month. So, there really were no surprises. This however leaves a question, was this information not properly priced in?
It wasn’t that the markets were blithe to the economic storm clouds ahead; a day prior to the announcement the three major indices were in the red month to date. What “caused” the penny to drop belatedly, so to speak? The markets are, of course, prone to volatile reactions, so it is always interesting to examine instances where these exaggerations supersede grounded consensus expectations and drive the market.
Dealing With Market Irrationality and Determining the Relevance of Data
From an event-based decision perspective, prior to the event there was uncertainty, despite the Fed having telegraphed its intentions more than usual. Post the event, that information ended up being correct, and to a large extent not particularly helpful, in that the market fell in a manner that would have been in keeping with worse news than that which correctly identified the rate hike. In data terms it was good data but of questionable utility.
In data analysis there is much focus placed on data quality, particularly when any inferences to be drawn from that data are consequential. Unfortunately, determining the relevance of that data is a whole other issue. If all was clearly causal; A, therefore B, therefore C, then investment decisions would be rather simple. The reality in investment decision-making is that this is rarely, if ever, the case. Instead, the illogical conundrum can occur, where data can be of good quality, the inference (the signal deduced from that data) can be logical and the end decision to act on that signal be completely wrong.
When Artificial Intelligence is being used to perform this type of analysis the usual response to this situation is to conclude that the information was lacking and therefore (logically) more information is required—hence the use of “Big Data,” where quality is replaced by quantity in the hope of getting the right answer. The problem with this approach is that it fails to take account of the irrational. If the deficiency of the signal is due in part to some illogical reason, then pumping all the data of the day at it will not make any improvement except by pure chance.
Go Beyond Big Data
At Plotinus we recognized that in order to build a successful AI trade decision-making strategy we had to embrace the aspects of irrationality contained in the market, not foolishly try to impose hard logic as though we were dealing with a mechanistic system.
To tackle this, we have chosen to use a Derived Data approach rather than use Big Data. In so doing our emphasis is on understanding the relevance of the data inputs we are using and how our AI modeling uses them to determine its signal from these inputs. In so doing our AI trade decision-making aims to encompass the knowledge that the market will at times defy logic. By seeking to integrate this into an investment trade decision-making strategy we believe there is the potential to have a more robust approach for identifying more statistically significant separation of signals from noise that can help the investor deal more consistently with a market full of incomplete information.
Investors always have to make investment decisions based on less than complete information, and seek to separate signal from noise. With the S&P 500 recently finishing at a 2022 low, and the Dow dipping into Bear market territory, prudent investors, looking ahead to the new quarter and coming year, may want to make efforts to better understand the relevance of the data inputs they use in making trade decisions, and to seek how they might include this to better manage their stock market exposures. ■
© 2022 Plotinus Asset Management. All rights reserved.
Unauthorized use and/or duplication of any material on this site without written permission is prohibited.
Image Credit: Lunamarina at Can Stock Photo.