TSN Predictor

I've been playing a Facebook 'game' for the last couple years now called TSN Predictor. It isn't really a game by most people's definitions but then it really isn't a Facebook game (in the sense that you can't spend money and get no benefit from spamming your friends) either. It's more of a combination between a contest and an advertisement. The basic idea is you try to predict the winner of an NHL hockey game each day and whoever put together the longest streak of correct predictions at the end of the month wins $1000. Then at the end of the year one of those winners also wins a car. I believe they can afford to do this due to the ads on the page for the car sponsor and because it advertises when the TSN network is televising NHL games.

There's also a 'double down' question each day which you can answer or ignore. If you ignore it (or if the question is a draw) then nothing special happens. You get 1 added to your streak if you pick a game correctly and get your streak reset to 0 if you picked a game incorrectly. Forget to play and nothing happens at all. Get the double down right and pick a game right and add 2 to your streak.

April was a short month as the regular season ended after 7 days. As it happened after 5 days I was sitting in 3rd place with an ongoing 8 streak. The 6th was Lounge Day and I forgot to play before we left. I ended up sneaking into the MathSoc office to play, but didn't want to spend too much time so I picked quickly and left. The situation had 2 people ahead of me with 9 points and only 1 game being played that day. Both people ahead of me picked one team to win (it's interesting that it shows me their picks) so I quickly picked the other one. I also picked the double down question (total goals scored in the game, over/under 4.5). My reasoning was that there were other people tied with me at 8 points and a bunch at 7 as well. So even though I was going to catch up to the 9s if I got the game right I could well be passed by people tied with me. Also, there was still another day to go and I figured I might well need 2 points on the 6th and 2 points on the 7th.

People at Lounge Day asked about the tiebreakers. I didn't want to catch up to the 9s, I wanted to pass them. Did I need to do that? Well, I didn't have time to read the fine print when I was sneaking my pick in but I figured I should look at it now in case the situation came up again. (I seem to recall coming 2nd in April last year...) It turns out the tiebreakers are, in order: total success %, total correct picks, random roll.

Total success percentage as the first tie breaker means I had the edge against everyone. I only had 8 points of the possible 9 after 5 days but it's because I forgot to play on April 1st. No one else was in that situation so everyone else at 8 points had a worse tiebreaker. The guys at 9 were currently at 100% but if they got the game wrong on the 6th (the only way I was tying them was by picking opposite them and being right) then they'd have a loss and I'd be beating them. The second tiebreaker wouldn't matter in such a short month but could be key in a full month. The third tiebreaker is completely random and I should pick an extra double down question rather than leave it to randomness since presumably I can pick a yes/no question with better than 50% accuracy.

So... Should I have picked the double down question on the 6th? Knowing now that I completely dominate the first tiebreaker I'm not so sure. Certainly I should have gone through everyone near me's picks. If I could end up at a 9 streak with control of the tiebreaker I could then figure out how to approach the last day. My thinking was that I needed to beat the current 9s and to guarantee I stayed ahead of all the current 7s. In reality I was fine against both of those and only really had to worry about the current 8s.

I'll try to reconstruct the day in question...

9 points - 9/9 - pick StL, under
9 points - 9/9 - pick StL, over
8 points - 8/9 - pick Pho, over
8 points - 8/8 - pick StL, under
8 points - 8/8 - pick Pho, under
7 points - 7/8 - pick Pho, over
7 points - 7/7 - pick Pho, over
7 points - 7/7 - pick StL, over
7 points - 7/7 - pick StL, over
7 points - 7/8 - pick StL, under

6 points - 6/7 - pick Pho, over
6 points - 6/6 - didn't play
6 points - 6/7 - didn't play


Assume Phoenix wins and I get the double down right... What are the outcomes of each possible choice?

OVER
10/10 - ME
10/11
9/10
9/9
8/9

UNDER
10/10 - ME
10/10

NO PICK, OVER

10/11
9/10
9/9 - ME
9/9
8/9

NO PICK, UNDER
10/10
9/9 - ME

I'm behind the 8-ball if I don't pick. Because I have people at 8 points going each way there's always going to be a 10. Then I again need to hope they get something wrong on the last day. Not picking and having the under happen is particularly bad as that guy also had a 100% pick percentage. I'm not convinced I need to pick something. And since both ways have a single other person at 10 points I don't see a huge difference between them. The over is a little better as I have the tiebreaker on that 10 pointer but either way I'm pretty sure I'd need to pick a game and the double down on the last day as well. I should pick whichever one I think is most likely to happen.

I picked the under. Unfortunately the game ended up having 5 goals scored in a Phoenix win. The final scores ended up being 11 - 10 - 9 - 9 - 9 - 9. The guy who ended up at 10 points actually didn't pick the double down on the last day! There was a guy at 9/9 at that point who could have won with a 100% tiebreaker if he'd gotten the final day right on both counts. I don't know that I would have been able to take that risk if I was him...