No, this isn't another academic paper arguing for the Efficient Market Hypothesis. This paper is actually interesting and useful.

The authors took well-known candlestick patterns and performed a statistical test on out-of-sample S&P 500 data to quantify the efficiency of patterns in determining when the market turned.

Interestingly, they found that the candlestick patterns worked extremely well, with a statistical significance of 36 standard deviations. [A 36 sigma significance means they are dead sure of their conclusion.]

A worthwhile read. I have only read the introduction and discussion, and scanned the rest.

https://www.pitt.edu/~caginalp/Paper65.pdf

I am in no way connected to the authors of this paper.

Good find. I skimmed through it and will probably go back for a more in depth read later. One thing that I find very interesting is that they essentially automated a few patterns to perform this study.