Back in the 1990’s I worked for two data platform companies, Tibco and Dow Jones Markets (Telerate). One of the advantages of working in those companies was access to real time data. As a result, I could test ideas about pricing and arbitrage writing just a little code. I learned very quickly that arbitrage opportunities were very transient and could not be capitalized on unless you could trade almost instantly. That truth has resulted in ever faster high frequency trading platforms.
At about the same time a colleague of mine was extolling chart patterns as means to divine market direction. This brought to mind the traders I used to work with who would laboriously create their charts by hand over night after the markets closed. This was laborious enough that they could only do a small number of instruments, and had no way to set alerts when a pattern, even a simple trend, was violated.
It seemed to me that if traders used chart patterns for their trades, they might want to do so using real time data. At the time I visualized a chart that displayed data for short time intervals and that ran the chart pattern recognition algorithms every period and displayed patterns as they formed. This would be using computational power to remove all that work and open up new opportunities to detect and respond to patterns as they formed in real time.
I never got the chance to test this idea and suspected that it would be a ubiquitous technique in trading rooms by the end of the first decade of the C21st. Well apparently not. Perhaps I will get another chance to try out this idea this decade.