Point, dispersion and distribution shape

There is point recognition, variance recognition, and distribution shape recognition.

When you have choices A and B and you’re trying to figure out which is better. - Point Recognition recognizes each option as if it were a point on a number line. - If Recognition of dispersion is done, it looks like this figure. - - a common diagram - Comparing A and B in terms of expected value returns to point recognition. - Comparison by expectation is not the only truth. - When you try to compare the two with 95% certainty, you get a reversal of the relationship between the big and the small. - - For those who perceive it in terms of expected value, the option of reducing the variance without changing the expected value “doesn’t make sense.” - - For example, rebalancing.

On top of this recognition of dispersion [Recognition of distribution shapes

• A world with no recognition of distribution shape is a normal distribution where the distribution is symmetric
  • Expected and median values match.
  • 50% chance of exceeding expectations

• @tokoroten: By the way, I’m not saying that this company will succeed, I’m just saying that VC’s invest in venture companies with the goal of getting 1 in 10 deals, so let’s consider the upside.

  • I thought a lot of people didn’t seem to get this.
  • This is talking about “the rationale for investing in something that has only a 10% chance of exceeding expectations”
    • Or we could say, “There are cases where it is reasonable to pay for something you know will fail 90% of the time.”
  • Mode 0 in some cases
    • There are cases where it is reasonable to invest in an investment even if the “most probable outcome is that it will be a scrap of paper.
    • 

---

This page is auto-translated from /nishio/点と分散と分布形状 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I’m very happy to spread my thought to non-Japanese readers.