NCAA tournament and maximizing EV

[bogey]

This just kind of occured to me and I was wondering everyone's thoughts.

It seems that generally (there a few exceptions) the higher ranked team in a tournament matchup is a favorite to beat the lower ranked team. So the max EV play on any given game is just to take the higher ranked team (no points given assumed).

So it would seem when filling out an NCAA bracket the most EV thing you can do is just to pick the higher ranked team to always beat the lower ranked, thus ending up with 4 #1 seeds in your Final Four.

However, anyone who does this is probably very unlikely to win their pool. (But maybe they are still the most likely overall? intuitively if you saw a bracket like this you would think no way they can win, but maybe that is wrong) It seems you may have to mix in -EV on singular games to come up with a greater EV for your overall bracket. This seems like a stange paradox which I can't quite understand.

Anyone have any thoughts on this? Or did I make some invalid assumptions?

[LLXC]

I forgot where I read the article but I believe there have been 7 or so upsets throughout the tourny last few years and for a few years before that, 11.

I think it depends on how the pool is structured...do you get more points as more teams get eliminated? 16, 8, 4, 2, etc.

[bogey]

Right, but that doesn't change the fact that for any single game the favorite team is the max EV bet with no points.

Just because there have been x upsets in the past doesn't mean it is EV to pick an underdog with no points and even money in any single game, which is essentially what your doing when you fill out a bracket.

So it stil appears that making some combination of -EV plays when filling out your bracket results in a greater max EV, which obviously can't really be true. So what gives?

[tech]

[ QUOTE ]
Right, but that doesn't change the fact that for any single game the favorite team is the max EV bet with no points.

Just because there have been x upsets in the past doesn't mean it is EV to pick an underdog with no points and even money in any single game, which is essentially what your doing when you fill out a bracket.

So it stil appears that making some combination of -EV plays when filling out your bracket results in a greater max EV, which obviously can't really be true. So what gives?

[/ QUOTE ]

To start with, your assumption in the first sentence above is absolutely 100% false.

However, if your pool gives the same amount of points for each win, you are right -- the best thing to do is just to pick the favorite. Under any other scoring system, you will have to look at a weighted distribution of the probabilities by the point values to determine the best theoretical play.

The reailty of it though is that since there is only one tournament per year, a theoretical advantage might never show up in the results.

[LLXC]

Yes...that's what I was trying to say...tech just did a 100000000x better job explaining it [img]/images/graemlins/smile.gif[/img]

I would answer this way:

The max EV is in picking every game right. Obviously, that's pretty much impossible. So whether the idea of picking all of the lower seeds works would depend on the number of other entries in your pool. If you are heads up, you would like your chances against someone else who picks games based on "knowledge". But if there are 100 other entries in the pool, the chances are pretty good that some sheet will have done better than you.

It's like playing A-A-3-7 in O/8. The hand plays better heads up than it does against 9 callers. Picking all of the lower seeds is like playing this hand.

Over the long run, your record of finishes in the top 1/2 of the pool will be respectable. But the idea is to win, or come in the top 3. And in a pool with a significant number of entrants, you will probably fall short almost every time, because sports does not always go according to forumla.