Streak for the Math

February 19, 2009 1:43 PM

Theoretically, what would be the best way to game the ESPN 'Streak for the Cash' game? (Note: I do not actually want to do this)

ESPN recently introduced a game on their website called Streak for the Cash, which gives you a set of different picks every day that you can make on who will win a game or score more points or other 'prop' type bets. Each of these picks have odds right around the 50/50 mark, and in order to win the game you need to string together 27 corrects picks (I don't know the math exactly, but I know that's a very low probability of success-- although you can improve your odds slightly with some sports knowledge).

So the question is, how many people would you need and what strategy would you employ to effectively improve your odds to the point that you could win? Is it even possible, or would you need so many people that it wouldn't be feasible? (Again, I don't actually want to organize this, but I was having a conversation about it and am not smart enough to figure it out myself).

Key points:
- There are about 10-12 different 'bets' you can make per day in a variety of sports.
- You do NOT have to pick every day, and you can pick multiple bets per day as long as the prior game has finished. In other words, if you sat in front of the computer and picked the games furthest apart during the day, you could probably make up to 2-4 picks a day.

posted by rooftop secrets to sports, hobbies, & recreation (13 answers total)

Assuming they're really about 50-50 as you say, the odds of 27 in a row are about 10,888,869,500:1 against. That's 10 (US) billion to one.

So to win by covering a big chunk of the permutations, all you have to do is line up a few billion friends to each enter one combination.

Pretty simple!
posted by rokusan at 1:49 PM on February 19, 2009

if the odds of each are 50/50, then the probability of getting 27 straight is about
1 in 134 million

Say you have pretty good sports knowledge, and you have a 60% chance of getting each bet correct. Then your odds are about 1 in a million.

The best strategy for improving your odds is to bet on the matchups that you have some knowledge of.
posted by chrisamiller at 1:50 PM on February 19, 2009

I think your best bet is to (a) have good sports knowledge, and (b) don't many picks that you don't consider to be an absolute slam dunk. This might mean picking once a week.

With that said, once you hit a certain level (say 20+) you know the people in charge aren't about to give you any freebies.
posted by PFL at 1:56 PM on February 19, 2009

Look if you can get 27 sports bets in a row, you can win a LOT more than $1m.

A 15 team parlay generally pays out 1500:1.
posted by jckll at 2:02 PM on February 19, 2009 [1 favorite]

Only pick games you would bet a lot of money on, seriously. You can stretch this game out for a while if need-be..do it. Check out Vegas lines to see what potential picks have the best chance of winning in order to maximize your odds.
Don't bet on those 50/50 games. Bank on the chance that eventually, a better pick will come your way.
posted by jmd82 at 2:04 PM on February 19, 2009

chrisamiller's number (about 1 in 134 million, i.e., 1 in 2²⁷) is correct. I don't know where rokusan gets his figure over 10 billion. The best thing to do, IMO, is to pick only those where you're reasonably confident your answer is significantly more than 50% likely to be correct. (Is the "Game" column the percentage of players who picked each option? You might limit yourself to picking only the popular choice in picks where one is heavily favored over the other, but even that's not a guarantee.) If you can pick only those where you are at least 60% like to get each individual answer right, your chances of getting 27 in a row improve to slightly better than 1 in a million. If you limit it to those you're at least 75% likely to get right, your chances improve to better than 1 in 2400. If you can limit it to those which are at least 98% likely, you have a better than 50-50 chance to get 27 in a row.
posted by DevilsAdvocate at 2:55 PM on February 19, 2009

To add onto what cklennon said, I believe a 15 team parlay would cost you $50 to make a $1mm. A 27 team parlay is statistically impossible in reality.
posted by geoff. at 3:20 PM on February 19, 2009

Statistics aside, if you don't have any special knowledge, and can start the streak over at any point, you might feel better with several chances. That would suggest you should bet on a bunch every day, so if you lose one, you still have time to try again.

That said, 4 or 5 over 134 million is still effectively zero. :-P
posted by chrisamiller at 4:23 PM on February 19, 2009

Your wisest bet is to use Vegas lines to move those 50/50 odds more into your favor. Try betting on the one or two games where the margin is the highest, but have some threshold before you bet - this probably means you don't pick every day.

And not to be off-topic, but can you imagine the goldmine of data ESPN has to play with? It's ridiculous - worth way more than a million bucks - if they're willing to analyze top bettors a bit, they could make some serious money gambling (or selling data to supplement Vegas lines).
posted by antonymous at 4:55 PM on February 19, 2009

Thank you for all answers thus far. However, no one has addressed the question about whether creating a team of people coordinating picks (i.e. picking in each direction on any particular bet) could significantly improve their odds, or if the odds are so low you couldn't significantly increase the odds of any player in the group winning without the group being prohibitively large.
posted by rooftop secrets at 9:30 PM on February 19, 2009

There's really no way for so few of you to work together. If you only had to pick 6 in a row, then you could round up 63 friends to cover all of the possible outcomes. 27 is way too many - you'd need chrisamiller's 134 million friends.
Even if you could pick out the winners with remarkable 90% probability, you'd still only have a 6% chance of picking 27 in a row by yourself. And if you could pick winners like that, you'd have your friends line up with the next most probable outcomes: agreeing with you 26 times and betting against you once. Those 27 guys would each have a 6 in 1000 chances of winning (for a combined 17%). Your next 351 friends could each cover the next most likely propositions (2 wrong guesses, about 7 in 10000) and so on.

Also, there's no way to work together other than trying to cover all of the possibilities. If you determine (somehow) that the odds of a particular outcome are better than 50-50, there's no other reason to bet on the underdog (as long as the outcomes are independent).

(apologies if the numbers are wrong; it's late)
posted by Horselover Fat at 12:16 AM on February 20, 2009

Also, don't forget that once one person "loses" the streak, they are free to start over - if you're going to turn this into a math problem (which I hope someone does), then you have 1/2 of your initial number of bettors free to start again (assuming an even start date and an even split on the first event). An approach to basic will assume these players are "out" and unable to start again. The goal is not to have one person get 27 in a row on their first try, but to get to that number eventually.

I'd love to see someone come up with an equation - I'm not that good at it and am short on time.

Also, Magic at Bobcats is about as easy as they come tonight ;)
posted by antonymous at 7:49 AM on February 20, 2009

I don't know where rokusan gets his figure over 10 billion.

A lazy 27!

But now that you guys forced me to think about it, that makes no sense. It's not 27 permutations, it's just 27 coin tosses. So the 134 million is right.

(Answering from the hip, sorry. Math is hard enough when I do pay attention.)
posted by rokusan at 10:18 AM on February 20, 2009