Adam (kenjita) wrote,

Expected Value to assess risk in an either/or situation is stupid, amirite?

Okay, Joe, or anyone else who knows statistics.

Let's say you're doing a problem that wants you to figure out the expected value of a situation. You're an oil company, and you're trying to decide if you want to dig on Site A or Site B.

Directly from my (easy) homework:

Site A
Profit if oil is found: 80 million
Loss if no oil is found: 10 million
Probability of finding oil: .2

Site B
Profit if oil is found: 120 million
Loss if no oil is found: 24 million
Probability of finding oil: .1

When you do your math, you find the expected value of site A is 8 million in profit, and Site B is 9.6 in loss. The homework then asks you to find the difference.

The problem is, this is not a guaranteed situation or anything. Unlike other expected value situations, you're not going to make 8 million dollars off site A. You're either going to make 80 or you're going to lose 10. One or the other. There is no third situation, in which you blend together both answers. Your books, at the end of the day, will either reflect a loss or a gain.

And this bugs me. Is this the kind of logic that is actually used? Do they just use this number not as an expected gain, but more of a risk assessment tool? Please say yes.
  • Post a new comment


    Comments allowed for friends only

    Anonymous comments are disabled in this journal

    default userpic