I built a scanner that finds chart patterns, and then I did the part nobody does — I followed every single one to see whether it actually worked. Most of them don't. The first pattern almost everyone learns, head and shoulders, reached target 142 times out of 531 attempts in US stocks. Taking any US pattern at all, choosing nothing, reached target 4,899 times out of 10,326. I went looking for a reason to keep it and found a better reason to drop it. That's the room. Everything here is a number that had a chance to embarrass me, and some of them did. — Three things, then you're set. Every rate here carries its n, including the ones above. "Double bottom long, 66.4%" is decoration. 1,106 of 1,665 attempts is a number. If I post one without the other, say so in the comments — I'd rather be corrected here than on Reddit. Nothing is judged on its own. Taking any US pattern at all reaches target 4,899 times out of 10,326 attempts. Forex, 50 out of 156. Commodities, 85 out of 169. So a rule that gets 45 out of 100 is bad in US stocks and good in forex, and quoting it without saying which market is meaningless. The losers stay up. 3,964 stopped out in US stocks alone, and another 1,463 went nowhere. If I ever quietly delete a row, every number above it becomes worthless, so I can't. Two things I did delete this week, on purpose. My forex list went from 778 pairs to 10, and 9,351 finished patterns went to 156. Half of that second cut was my own fault: every pair I kept had its inverse in the book as well, so every move was counted twice and six setups were clearing my 20-attempt floor on duplicates. One of them still grades A on 23 real attempts, which is the part I was glad to find. The Friday post that was going to be about forex is now about US stocks, because I deleted its evidence. What I can't tell you: whether any of this works on the right hand edge of the chart. Every number I have is measured on bars that already printed. Argue in Argue with the numbers. It's the one category you can post in.