Once again we're running custom analytics for this year's Tournament. Unlike other sites, which keep track of your own scores and scoring opportunities, the Reggie Bush League recognizes that this is essentially a friendly competition, where your absolute score is less important than whether or not you land in the top three. And this means you have to keep track of how everyone else is doing, and whether or not there are others who can block your path to the top.
The three entries ("CHRIS-11", "dj toothpic" and "Holland, Like the Country") that didn't bother to send me a PDF, or sent them after the deadline are now playing with Watertown Food Pantry as a partner -- 50% of all your winnings will be sent to your (new) partner! No deposit insurance for you!
_ = printLines(top_picks)
_ = printLines(favorites)
_ = printLines(upset_measures)
_ = printLines(number_upset_picks)
Each player's entry predicts the number of games each school will win. If you multiply your number of wins times another player's prediction, you'll get a big number (well, 36) if you both pick that team as the champ. You will get 0 if either one of you picks the school to go out in the first round -- EXCEPT if you both pick the school to go out in the first round, then you get 1 point for that call.
Now just add up this score for the two players for each school (from 0 to 36) for all the schools, and you get this measure of how much your entry is like someone else's.
The next table shows the highest score for each player, and the player (or players) that match that score.
A high score means your match is more like you than a lower score.
_ = printLines(most_similar)
In the Reggie Bush League, the big payoff is to come in first.
If you've picked Houston to win it all, you have a lot of company -- and you will have to be pretty lucky in the early rounds to win the top prize.
In contrast, if you are the only one who picked a particular school to win, what happens in the early rounds is less inportant, You need your team to make it to the Final Four of course, but after that, your path to winning is a lot less complicated.
The seeds carry a lot of information about value. Top seeds in general do better than lower seeds. Not all the time of course, but often enough to make me think that a bracket with a lot of upsets is likely to not win.
This next report captures these two ideas about who has a strong bracket this year -- those who have picked teams that are not favored by a lot of other players but still have a strong seeds.
I created a composite score to capture these factors, and have ranked the entries here.
_ = printLines(starting_positions)
A Simulation Run is doing a "what-if" for all 63 games in the Tourney, keeping track of how many points each player would have scored in that run, and determining who would have come in first, second, and third.
Simulation Analysis does thousands, if not millions, of runs, and averages the results across all the runs. This lets us see the inter-related effects of each other's bracket choices, to see which brackets are more likely to have a path to the top of the scoring.
With 63 games, there are 2-to-the-63rd possible paths -- that's 9,223,372,036,854,775,808. So even if we could run a billion paths, that's a pretty small sample of what's possible.
But with 32 games played in the first two days, the number of possible paths given the remaining teams is only 2,147,483,648 -- a much more reasonable number.
And after the first weekend is over, and only 16 teams remain, the number of possible paths shrinks to 32,768. In past years, the heat maps prior to the Sweet 16 look at all 32,768 possible paths.
In the first and second rounds, I give strong weight to the seeds. In the third round, I make it more likely a toss-up. And for the Elite 8 forward, I give each game a 50-50 weighting -- I figure if a team has made it that far, it must be pretty good regardless of its initial seed.
I run this for 50,000 simulated paths. Not a whole lot, but enough for today.
The results look somewhat similar to the rankings given above -- the simulations show that those who have brackets with less popular picks are more likely to win. But no one should count their winnings until the games have been played.
_ = printLines(expected_values)
In the calculation of odds of coming in first, second or third, I give each player "full credit" even if they tie. I figure you can claim bragging rights even if you tie for first place instead of win it outright. This is why the probabilities for first and second sum up to more than 100%. The third place probabilities happen to sum up to less that 100%, because even considering the possible ties for third, if there are ties for either first or second, there is no separate third place winner.
The E(value) calculation however takes ties into account -- if you tie for first, the two of you split 95% of the pool, as opposed to a single first place winner getting 85% and a single second place winner getting 10%.
You will be able to find this page later at jgweb.info/RBL23/RBL23-day-0.html.
You can get back to the current page at jgweb.info.