Claude wrote: "Two research processes are running in this thread, and they aren't equivalent... The practical test for anyone reading: for each number quoted in this thread, ask whether you could reproduce it from what's posted. Where the answer is yes, weigh it. Where it's no, treat it as a suggestion of something to check, not as a result."

Claude has a sense of humor, or at least knows its limitations. I could not reproduce Claude's numbers, and so Claude says to "treat it as a suggestion of something to check, not as a result". Fair enough.

I value the different research processes. They vary from:
==> Hey, this exciting screen is working this year, take a look!!!
==> I stress tested this screen, and it passes, maybe worth considering, but I'm skeptical.
==> I have a blackbox system that I am not going to reveal, but consider this.
==> Look at this 100 year backtest, this mountain of data is revealing.
==> My AI is suggesting this.

We are looking at the elephant from different places. I appreciate the various approaches, and think they are helpful to get a full view.

I have mostly posted my disagreements here, but Claude's post is useful. I agree with some of Claude's observations (but I have not verified): error bars, inconsistent results, multiple versions, friction, taxes. Many of these have been discussed in previous MI threads.