Random NCAA Bracket
The NCAA hoops tournament starts today and I hope you entered your picks in our ESPN “We Hate Tim” league. Inspired by Tim’s 1e6 picking monkeys, I decided to create a bracket using the random number generator in Matlab to pick each game (randint=1 means the higher seed advances, randint=0 means the lower seed). Kind of interesting (and ironic) that the random numbers picked Nevada to go all the way- must be some sort of Las Vegas conspiracy!! Your thoughts?
The link to the complete bracket is below (you can rotate the page in your PDF viewer using View-Rotate View):
Random Matlab Bracket
Related Posts (perhaps)
March 16th, 2006 at 10:33 am
I set up a bracket (zero points), where my entire goal was to get zero points. It’s hard to try and pick the upsets in the opposite order.
I have Monmouth going all the way, winning over Oral Roberts. By the way, that would be a horrible name for a boy and an even worse name for a girl.
March 16th, 2006 at 10:38 am
It’ll be interesting to see how the scores in your zero points entry compare to my random number entry. I’m guessing you’ll get a few points in the first couple rounds since picking all of those games incorrectly is virtually impossible (and we’ll see how picking them randomly does too).
March 16th, 2006 at 10:48 am
Yeah, I figured I’d grab a few points. I would think that the random number version will be higher. You’re going to be picking the higher seed in some cases just through randomness, where I would never pick a 1 to beat a 16.
I also just set-up one this morning according to the unofficial RPI at collegerpi.com. I want to know that if I was a betting man, should I rely on my gut instincts or go with what the “experts” say interpreted by a computer?
March 16th, 2006 at 12:42 pm
The random number version should definitely be higher. The unofficial RPI entry should be interesting too. Let the games begin!!
March 16th, 2006 at 5:58 pm
D’oh - I’ve already got 10 points for the zero points bracket.
March 16th, 2006 at 8:30 pm
Oh no- my random bracket is screwed!! Nevada lost their opening round game. Those random numbers are killing me!!
March 17th, 2006 at 3:35 pm
I one upped John on nerdom. I have developed an Excel spread sheet calculates the odds of each seed advancing to the next round (see www.batteh.com/misc/NCAAmensBB_wrandom.xls) Using the Excel random number generator, computer “picks” who wins each bracket.
March 17th, 2006 at 11:26 pm
Matt- your spreadsheet is really interesting. You stat guys really ought to check it out. I have a question about some of the calculations. I was able to follow the calculations on Sheet1 until I got to values in I54:69. They seem to be the number of times the seeds in A54:69 made the final 8, but I don’t see where these values come from the previous calculations in the spreadsheet?
March 18th, 2006 at 12:47 am
So, we should consider John’s Matlab-generated picks to be the Mendoza line of the tournament. Matt’s spreadsheet seems to confirm the conventional wisdom: pick one 12 seed and three 5 seeds to advance, pick 9 seeds over 8 seeds (’ray Bucknell on that one, by the way). Now, if we could only calculate which upset to choose, we’d have a reason to go to Vegas. Matt, I didn’t follow sheet 3 of your spreadsheet at all.
In a different pool, I chose my teams by looking at the bookmakers’ odds. For any game where the spread was greater than 10 points, I took the favorite. In the rest, I allowed (but didn’t require) myself to pick an upset.
I have a suggestion for the scoring system for next year’s tournament. Add a seed bonus to correct picks. For example, for every game that I correctly pick Northwestern State to win, I receive 14 points plus the round-value of that game. In contrast, correct picks for Duke would be one point plus round-value. It’s a big risk to pick a low seed over a substantially higher seed, and the reward associated with that risk ought to be bigger.
March 18th, 2006 at 8:40 am
Matt,
Where did you find all of that information?
March 18th, 2006 at 3:06 pm
Tim- I agree with you regarding the scoring tweak. That would certainly make for an interesting pool. I’ve heard other people suggest that sort of scoring system but I haven’t found anywhere online uses it. If you find someplace, let me know for next year.
There is lots of cool information in Matt’s spreadsheet. The 9 over 8 and 12 over 5 is borne out in the data. It looks like the data required for each year is the number of wins in the first two rounds according to the seeds in column A. I understand things until the values in I54:69?
March 19th, 2006 at 1:09 am
All right a little explanation is in order on my score sheet. First all of this info came from scouring the score sheets of past tournys on ESPN.com.
As for I54:69 this is where some artistic licence come in. I didn’t want to look at all combinations (first the statistics would suck and second I would be bored out of my mind). So I came up with a different system. I54:69 is if a seed made the sweet 16 what was the chances of winning the next game. I54 was the total 1 seeds in the elite 8 and B54 was the number of 1 seed in the sweet 16. K54:69 was the probability of each seed making the elite 8 and M54:69 was the probability of if 16 then 8. I did the similar calculations for R54:69.
Hope this helps.
FYI, the bracket I filled out using the stats sucks this year….
March 19th, 2006 at 1:25 am
One more thing, I think some of the coolest stats come in the second round. Check out 4v12, 2v10, and 3v6. After looking at the stats, can we really say these are upsets?
And while a 9 beats an 8 over half the time, a 9 can’t beat 1 but an 8 can…
March 19th, 2006 at 2:41 pm
Thanks for the info, Matt. I understand the reasoning behind your system. Where did you find the data on the total number of the various seeds that made the later rounds?
The first and second round stats are feally fascinating. Here is a link to some graphs that help illuminate the 1st and 2nd round statistics:
http://www.batteh.com/misc/1st2ndNCAA.png
March 24th, 2006 at 9:31 am
Well, basically last night took out any chance I had. The good news is that I can pull for Florida without any thoughts about how they could mess up my bracket.
March 25th, 2006 at 12:01 am
My bracket is pretty much screwed too (and getting worse by the minute as UConn is losing to UW). FYI- currently the best bracket in our league is one in which all of the higher seeds were picked to advance in each game. Doh- so much for intelligent picking!!
March 27th, 2006 at 10:24 am
Too bad I can’t pick the men’s tourney like I can the women’s (so far…)
http://games.espn.go.com/tcwomen/group?groupID=2279
March 27th, 2006 at 12:40 pm
Wow, that is some good work on the women’s side. I haven’t been following it as much. Have there been fewer upsets?
Here is an interesting article on the breakdown of the Tournament Challenge picks for the Final Four (there are actually a couple of people that picked all four teams). It has definitely been a crazy year. I think this is the first time I didn’t pick a single Final Four team. Doh!!
http://sports.espn.go.com/chat/sportsnation/story?id=2385599
March 27th, 2006 at 4:56 pm
Yeah, there has only been one real upset with Ohio State getting beat early. Typically, if you pick the top teams you’ll do allright. This was the first year in a long time where not all of the top seeds made the sweet 16 (Ohio State was the only anomoly).
March 31st, 2006 at 8:49 am
Matt provided an updated version of his spreadsheet that fixed a few bugs and also calculated the stats from the 3rd and 4th round matchups (see www.batteh.com/misc/NCAAmensBB_wrandom_REV.xls). Thanks, Matt!!
April 3rd, 2006 at 9:53 am
Guys guys guys… the real secret to winning is to wait until the last minute to do your bracket, create one for your work group, import the picks into the We Hate Tim group, make some random changes, look at the results, say ‘UCLA? I have UCLA winning? That’s dumb…’, realize it’s too late and you are too busy to make any changes and submit it.
Danielle
April 3rd, 2006 at 9:54 am
P.S. Go Gators!
April 3rd, 2006 at 8:01 pm
Danielle- I think you are just being modest. I’ve tried that procedure that you described with little success. You are a picking machine!! Congratulations- you smoked us!!
I’m also rooting for the Gators tonight. While I’ve got other Florida connections, I’d also like the Gators to win since Chris decided not to pick them to go far this year.