Hacker News Reader

50-50 forecasts can be massive depending on your priors. If someone (who I trusted) told me there was a 50-50 chance I would get shot tomorrow if I went outside, you couldn’t pay me enough to go leave the house.

Absolutely! It didn't work that way for recent presidential elections, though. The prior odds weren't always 50:50, but they were well within the range of "both sides could win" and didn't go outside that.

In the case of an election, a 50:50 forecast is a clear reminder to the wise of the importance of voting, no matter how certain the outcome might seem from the necessarily limited perspective of your acquaintances and your consumption of possibly (or likely) biased news media.

If a modern US election went far outside that that would just mean some fool was burning a huge amount of money and probably to no purpose as it would cause a response from donors on the other side who would like an outcome but also aren't going to throw money away.

It's less likely for the presidency, but for other offices, it's often the case that an incumbent has only token competition.

You have a choice to go out or not but the example of the wedding is a given obligation, so the apples-apples comparison would be "if you had to go outside, would a 50-50 chance of being shot change anything"

This is an interesting hypothetical, and I hope people with better understanding of statistics would weigh in.

But lets change the example a bit:

- There's a 10% chance of getting shot on any day you leave the house - You are able to get 5 days of food any time you leave the house - There's a 100% chance of starving if you go 8 days without leaving your house

How long can you avoid getting shot or starving?

Presumably if you don't want to go hungry at all, you have to leave the house every 5 days to get more food. After 50 days, the odds of you having been shot are pretty darn high.

Respectfully, how is this noteworthy or add to the discussion?

I guess I was trying to get the parent comment to reconsider their position on 50/50 being useful because they would never leave the house. But I guess it's still "more" useful than 90/10 even, in their example, the chance during one day isn't reflective of long term odds.

But an election (particularly presidential) is only once every 4 years. And it's not "50/50" or "90/10" every election - it's typically a different match up of options, and a slightly different mix of percentages.

But fair criticism, I didn't successfully make a good point with my attempt.

Suppose it rains on 50% of days. One forecaster gives a 50% chance of rain every day. Another gives a 90% chance of rain/not rain and is "wrong" 10% of of time. Both stations are giving you accurate information from a probability perspective (when station A says a 50% chance, you know there's a 50% chance of rain tomorrow, when station B says 90%, you know there's a 90% chance of rain tomorrow) but the 50% chance station is less useful as the number is lower. They have effectively given up in the same way pollsters have given up.

This is the best response in thread, imo. Is there a term for this phenomenon?

I don't think so, though I often find someone has thought of the same thing I have and given it a name. The root of the issue is a matter of language. Events don't have probabilities. You don't predict a 50% chance of rain, you predict rain with a 50% certainty. (There are a lot of internet brain teasers that exploit this subtle ideological difference.) The 50% is talking about how often you're correct, not how likely the rain is to happen. Probabilities don't really exist; it either rains or it doesn't.

It's statistical information arbitrage.

Let's say you go to the futures market:

Seller pays a contract-buyer $1 if it rains tomorrow. How much will seller charge for that contract?

Bookmaker using forecaster A will sell that contract for $0.51 every day, and earn a profit of $0.01/contract, on average.

Bookmaker using forecaster B and their clients will buy that contract on probably sunny says, and bankrupt Forecaster A.

Similar result if Bookmaker B sells contracts to Bookmaker A, at $0.11 on probably-sunny days, and $0.90 on probably-rainy days.

Weddings are unusually locked in dates. Canceling a hotel a few days out is generally free and 50% odds of rain can easily be high enough to postpone a weekend trip. 10% odds however may be worth the gamble even if both possibilities exist.

With elections you need to take both a short term and long term perspective. True 50/50 odds mean doubling down at the last minute is likely more valuable vs saving that money for the next election. I think the pollsters got things wildly wrong with that estimate because their underlying assumptions where off, but that’s a different issue.

The utility of consulting weather forecasts prior to events like weddings lies in the non-trivial possibility that it will give you a clear indication of what will happen.

The problem is that there are two 50-50 odds definitions.

First one, is where you have no information about the event what so ever.

Second is where you have information, and it turns out that the outcome is still 50-50.

While the use of that information in planning (pricing things, betting, investing) is identical, they fundamentally represent 2 different sets of information.