Hacker News Reader

Comment by tzs | Hacker News Reader

Choosing lottery numbers strategically does make sense, because your expected value depends on both whether or not your number wins and on how many other people picked the same number.

A lot of people pick dates, or numbers that have won recently, or that form geometric patterns on the ticket, or that form interesting sequences, and so on.

Nothing you can do (assuming no bias in how the winning numbers are generated) will affect the chances your ticket wins, but by picking randomly from the set of possible tickets that doesn't include those commonly picked sets of numbers you can decrease the expected number of people you'll have to share the jackpot with.

View on HN

> A lot of people pick dates

Yes, birth dates of themselves and family members, which are randomly distributed in the population, and which you can't possibly know, because you don't know the identity of other lottery buyers.

> or numbers that have won recently

Do people do this? I'd assume they do the opposite of this.

> or that form geometric patterns on the ticket, or that form interesting sequences, and so on.

This is all hand-waving. Please list the numbers that I shouldn't choose.

> those commonly picked sets of numbers

What are those numbers?

In any case, even if you eliminated several thousand numbers from the pool of possibilities, that doesn't really help you choose a specific set of numbers from the millions of possibilities, other than not those.

> Yes, birth dates of themselves and family members, which are randomly distributed in the population

Nope - not uniformally random, there's literally a non uniform age distribution, a specific set of months, a capped number of days, and various "peak" birth seasons.

    >> > those commonly picked sets of numbers
    > What are those numbers?

One way to find out is to get a math degree and moonlight as a consultant for, say, the local uni math department and perform quasi regular spot checks on the state lottery games.

There are many lottery variations and differing games show interesting patterns of choice - although more and more people these days simply get machine generated tickets .. this expands the surface of interest as not only the game results themselves need to be checked for bias but also the various networked game terminals for their random generation and for local tampering.

It's probable that anyone who has done such work is bound by NDA's and agreements not to dump mass data aquired through work.

If lotteries could be gamed mathematically, then they would be gamed mathematically, and we'd have a bunch of mathematical millionaires as a result. I'm calling BS on all of this hand-waving. Seriously, how much do you think the miniscule odds of winning can be increased via these methods? I want a numerical answer.

And in any case, these factors do not apply to US Presidential elections and are thus irrelevant to the larger point.

Yes, that is why when lotteries can be gamed they often are, eg:

https://listverse.com/2019/06/04/10-people-who-successfully-...

> I'm calling BS on all of this hand-waving.

Maybe work on the math a little.

> Seriously, how much do you think the miniscule odds of winning can be increased via these methods?

There are multiple methods that can be and have been used, they all have different outcomes on Expected returns.

> I want a numerical answer.

Again, maybe you might want to work on the math a little yourself - on a specific ruleset (or on a family of rules using some abstraction) .. there are tools that can help you; mathematica, R, etc.

The main issue here is a single vote in a US Presidential election, i.e., one vote out of around 150 million. Thus, the relevant analogy is buying one lottery ticket in a huge lottery such as the Powerball or Mega Millions and matching every number to win the jackpot.

Everything else, including cheating and rigging the lottery, is irrelevant to the question of strategically voting in a US Presidential election. None of your examples or hand-waving have shown that you can significantly improve the odds of one ticket winning the Powerball or Mega Millions jackpot.

> None of your examples or hand-waving have shown that you can significantly improve the odds of one ticket winning the Powerball or Mega Millions jackpot.

No one has said that you can improve the odds of one ticket winning. What has been said is that you can improve your expected value of playing one lottery ticket by carefully choosing what ticket to buy.

Your expected value is the product of:

• The probability that your ticket wins

• The value of the prize

• 1/(N+1) where N is the number of other people whose tickets have the same numbers as your ticket.

The first two factors are the same for everyone. The third depends on the numbers on the ticket because a lot of players do not pick their tickets randomly.

For example 1 2 3 4 5 6 is a surprisingly common choice. I've seen several lotteries where they reported afterwards that if that had been the winning ticket the prize would have been shared by several hundred or even thousands of people.

> What has been said is that you can improve your expected value of playing one lottery ticket by carefully choosing what ticket to buy.

Carefully choosing? AFAICT the so-called strategy is "choose numbers higher than 31". That's a tiny bit helpful perhaps yet still exceeding vague. And this strategy depends on others not also adopting the same strategy, in which case it would actually become counterproductive. If all lottery players acted "rationally", the strategy would cease to exist.

I logged in simply to praise the level of patience you have

"Thus, the relevant analogy is buying one lottery ticket in a huge lottery such as the Powerball or Mega Millions and matching every number to win the jackpot."

I like your analogy, but I think it needs to account for the fact that there are two scenarios in which a single vote can influence an election:

1. If it would have been a tie without that single vote. In this scenario, the single vote breaks the tie.

2. If the candidate that the single vote is to be cast on would have lost by one vote withtout that single vote. In this case, the single vote creates a tie.

The odds of either happening are astronomically long, of course.

This short paper goes into some more detail [1].

[1] https://www.eecs.qmul.ac.uk/~norman/papers/probability_puzzl...

Ok, so, choose numbers higher than 31.

That still doesn't leave you with much guidance. And the more people who follow this advice, the less it becomes effective.

Reply on news.ycombinator.com