Since the current version of artificial intelligence is based on training on content published over the years, it is only natural that the misconceptions in finance will compound.
The fact that marketing professionals have produced the majority of content is one factor contributing to the finance field’s highly probable decline due to artificial intelligence. In many cases, the astute content is not easily comprehensible unless one has a background in quantitative analysis and also some skin in the game.
There are so many examples, but the best one, in my opinion, is the zillions of claims that passive investing is a free lunch. The claim relies on long-term charts while ignoring the multiple uncle points along the time-domain path. I have written several articles about these misconceptions; here is one, but I think Stefan Gasic has captured everything in this excellent cartoon.
The rise in public debt has also contributed to massive misconceptions, with the MMT crowd dominating. Recently, I have even read articles arguing for an acceleration in debt issuance because otherwise equities may get in trouble. This is an example of how short-term greed blinds rationality.
I wrote an article on Substack late last year, showing the association between the rise in stocks and public debt.
The association is undeniable, and so are the ramifications recently. The stock market is hungry for more public debt because it stimulates consumption, among other things, but at some point it may have to go into fasting mode. What will that mean for the market? How much will it fall? These are hard questions to answer but the impact may be significant. There is little content that deals with this problem and AI will have a hard time assigning high probability to these scenarios. Any AI trained on “official” content about the impact of debt on the stock market (MMT) will probably conclude there is no problem; just brrr…
One problem is that the bulk of the content is “sponsored” and comes from similar sources with similar objectives. Dissident opinions get little or no coverage.
As far as quantitative finance, maybe the biggest misconception is that the Sharpe ratio is not a good measure of risk-adjusted returns due to “punishing good volatility”. Usually, those claims come from marketing people who have set up a network of podcasting and article publishing and they sound convincing, especially when they attach a PhD next to their name. The true reason for these claims is that the products they try to market to investors have a low Sharpe ratio. They usually promote the Sortino ratio as an alternative, which considers only “bad volatility.” But they are wrong, because there is no “good” or “bad” volatility; there is only volatility. More importantly, variance is not additive; it is hard to separate the good from the bad. Regardless, here is an example.
The chart shows the buy and hold performance of ARKK ETF from inception to May 6, 2024. The bottom charts show the Sharpe, Sortino, and the ratio Sortino/Sharpe.
In early January 2021, the Sharpe ratio was about 1.36 but the Sortino ratio was 2.21. Then, the ARKK ETF started falling and by the end of 2022, it had dropped 80% from its all-time highs. A high Sortino is not a guarantee of future performance.
More importantly, there are hundreds of articles that claim that the Sharpe and Sortino ratios are the same for normally distributed returns. The truth is that even in the case that the returns are normally distributed and the covariance of positive and negative returns is zero, in the limit of sufficient samples, Sortino will be equal to Sharpe times the square root of 2.
AI without the mathematical ability to verify claims in content could be a misinformation tool. More importantly, Gödel’s second incompleteness theorem says that no system can prove its consistency (assuming the first incompleteness theorem).
In other words, modern AI relies on consensus opinion, and if it is wrong, you will get wrong results, as for example, buy and hold is a free launch without enumerating the risks, which are not even mentioned in all those articles that promote “do nothing” fund management because “timing does not work.”
Last, but not least, are the numerous misconceptions about why allocators avoid managed futures CTAs. This is true because allocators are usually people with good mathematical skills. CTAs, even the best, have a low Sharpe ratio due to high volatility. This means, among other things, that future drawdown debt and duration could be much worse, and in extreme cases, there may be an uncle point. Lately, allocators have been interested in CTA replication funds, which limit these risks. Managed futures-specific risk (investing in a CTA) is akin to speculation for absolute returns, which goes against the objectives of allocators, who aim at achieving superior risk-adjusted returns. However, the numerous articles written by the CTA industry praise their diversification benefits. Will AI get fooled?
In my humble opinion, the best metric for AI in finance would be to differentiate between marketing and astute analysis. Will it succeed? It is highly unlikely.
Premium Content
You have immediate access to hundreds of articles when you subscribe. Premium Articles subscribers have immediate access to more than two hundred articles, while All in One subscribers have access to all premium articles, books, premium insights, and market signal content.
Specific disclaimer: This report includes charts that may reference price levels determined by technical and/or quantitative analysis. If market conditions change the price levels or any analysis based on them, we will not update any charts. All charts in this report are for informational purposes only. See the disclaimer for more information.
Disclaimer: No part of the analysis in this blog constitutes a trade recommendation. The past performance of any trading system or methodology is not necessarily indicative of future results. Read the full disclaimer here.
Charting and backtesting program: Amibroker. Data provider: Norgate Data
If you found this article interesting, you may follow this blog via RSS, Email, or Twitter.
