I have been referring to the market sentiment gauge and the signals from the weekly Market Prediction Saturday Showopens in a new window as algorithms for many years now.
I have made countless modifications to the algorithm to take into account different seasonality performance chart patterns, economic indicators, technical indicators, and social indicators over the years. The algorithm became so complex, taking into account so many different signals that it constantly gave a sidelines rating for many months in a row.
In March of 2018 I added reporting over CNBCopens in a new window as a score to the algorithm and it created some fantastic movements in the indicator. That got me thinking about adding Bloomberg reportingopens in a new window as well. Folks, this really was the breakthrough in the market prediction algorithm that I was looking for. I had ignored the HUGE impact that mainstream financial media groups have on daily and weekly market action. You might be thinking “Duh, Lance”. Right. It’s so obvious that mainstream media has a big impact on stock market action that for years, I failed to account for it at a proper weighting in the algorithm. Sometimes it’s the things that are so obvious and right in front of our faces, that are the hardest things to see.
On March 9, 2018, something bizarre happened that took the market prediction algorithm to an entirely new level. On both March 8, 2018, and March 9, 2018, the algorithm did not change even though the market had a fairly large movement. For the first time, the algorithm did something that did not make sense to me. Indeed, the calculations take so many things into account that I could not find in the data the reason for why the algorithm did not move between March 8, 2018 and March 9, 2018. Whether this ultimately turns out to be correct in predicting future market direction is another discussion entirely. The point I’m trying to communicate is that I no longer understand the calculations behind why the algorithm is giving the readings that it is giving. I’m sure that somewhere, buried deep in all the signals coming from technical charts (as seen weekly on the Saturday Showopens in a new window), bond markets, daily economic reports, social postings (Twitter, Facebook, StockTwits), and mainstream financial media coverage, that a reason exists but I could not find it! The complexity of the algorithm has exceeded my own ability to understand it.
On March 10, 2018, I made the decision to no longer refer to “it” as a stock market prediction algorithm, but instead to SOFI. The term “artificial intelligence” is notorious for having a relatively amorphous definition. Like all AIs, the SOFI AI is nothing more than complex algorithms at its core. My stock market prediction algorithm has become so complex with so many inputs that it is taking on a kind of “life” of its own to where I no longer can predict what it is going to output. For me, that meets the criteria of an AI (artificial intelligence).
I will no longer be referring to “it” as the stock market prediction algorithm but instead as SOFI. The name SOFI is an acronym for Simulated Ontological Financial Intelligence.
Premium members with a U.S. phone number will receive a text message each day after market close with what SOFI thinks about the stock market.
All visitors to the website will be able to see SOFI’s daily score. The SOFI AI Market Sentiment Gauge will show in the sidebar on wide-screen desktops and tablets, and at the bottom of the page on small-screen mobile devices.