I've been suffering from too much all-day Irish Pub Crawl and running around LV booking crazy +EV derivatives to have the time do a full-write up on a bracket. You will have to settle for these simple steps and ask me to explain them at a later time.
The most important things are to 1) Pick what Vegas is picking be Vegas is rarely wrong, and if they are it's rarely by much, 2) Make picks that separate yourself from other people in your pool.
1. Look at the market-bracket at Poofler's blog. (You will notice that the first round is roughly equivalent to the Spread/MLs of the market.
2. Pick the favorites there.
3. If it's close(<60% for the 1st round for example), you want to pick against everyone else in your pool order to separate yourself from the rest of the people in your pool. The more people that you think will pick the favorite, the more inclined you should be to pick the upset. Some basic data for what people are picking can be found at Yahoo Sports, though it may not necessarily apply to the people in your pool. Sometimes you will pick both the Vegas favorite and against what other people are picking(see California, only 33.6% pick them, but Vegas has them as a slight favorite at 52.4%). This is the nuts.
-Things that people tend to pick:
Teams they are fans of.
Their alma mater.
Teams they see on national TV or in their conference.
Whoever ESPN is pimping in the last couple of weeks.
Whoever has a household name.
The higher seed.
4. ???
5. Profit!
P.S. Root for BYU/Memphis/WVa/Missouri for me. Thanx.
Wednesday, March 18, 2009
Sunday, March 15, 2009
Bracket Analysis - Kenpom Based Model Bracket
Over the next few days I will be unveiling a number of different brackets. A couple I have are Vegas lines or Vegas steam/movement, and my personal opinion + game theory optimal if you were submitting a bracket in a very large pool.
I predict that alot of them will look very, very similar. This one is just the numbers - who should be favored in each matchup according to our Kenpom-based model. The only subjective element I have added is HCA.
I predict that alot of them will look very, very similar. This one is just the numbers - who should be favored in each matchup according to our Kenpom-based model. The only subjective element I have added is HCA.
Quick Blurbs - Memphis' Defense is Really Really Good...
#1 Adj Def Eff.(pts allowed/100 possessions, adjusted for Opp.) teams in the Kenpom era:
2004: Louisville 83.9
2005: Duke 84.9
2006: Iowa 84.4
2007: Kansas 82.2
2008: Kansas 82.8
2009: Memphis 79.4
2004: Louisville 83.9
2005: Duke 84.9
2006: Iowa 84.4
2007: Kansas 82.2
2008: Kansas 82.8
2009: Memphis 79.4
Quick Blurbs - Life After Jerome Dyson
Connecticut has played 7 of it's 31 games without Jerome Dyson which is not much of a sample size, but it will have to do given that we are headed for the true post-season now. Conventional knowledge says that Dyson is a secondary offensive option, but their lockdown perimeter defender.
Kenpom says confirms that he is not as efficient of a shooter, but is involved in a high volume of possessions and shots and as a distributor. He is also good at drawing fouls, an adept rebounder for his position and forces a good amount of turnovers.
Kemba Walker and Craig Austrie, who have picked up most of the minutes that would have gone to Dyson, are no slouches in their own right, but don't have the unique contributions that Dyson provides.
How have the Huskies played since his absence? Here is a snapshot of UConn's performance just after the Syracuse game in which Dyson was hurt and their entire resume:
Kenpom says confirms that he is not as efficient of a shooter, but is involved in a high volume of possessions and shots and as a distributor. He is also good at drawing fouls, an adept rebounder for his position and forces a good amount of turnovers.
Kemba Walker and Craig Austrie, who have picked up most of the minutes that would have gone to Dyson, are no slouches in their own right, but don't have the unique contributions that Dyson provides.
How have the Huskies played since his absence? Here is a snapshot of UConn's performance just after the Syracuse game in which Dyson was hurt and their entire resume:
Saturday, March 14, 2009
Fun Fact - Memphis' FT Shooting Improves... From Within
Memphis 2008 FT: 61.4% (329th)
Memphis 2009 FT: 69.3% (153rd)
I've known for awhile that Memphis' FT Shooting has improved since last year - it's just been nothing to write home about until I realized today that Tyreke Evans isn't much of a charity stripe lock(69.6% on 161 attempts). The reason this struck me as awkward was that I assumed he was one of the causes for the upward movement given that the better shooters from last year's team (CDR and Rose) are now playing in the NBA.
So where is this improvement coming from?
Granted, it's a lot easier to improve in anything when you are already poor at it, but this kind of improvement, esp. from Taggart and Anderson are anything but typical. Without it, Memphis' mediocre offense (40th) would be much lower to the point where it would be something to worry about. Improving your FT shooting esp. in close games is anything but smexy, but PPP doesn't give a shit about where you get your points from. Just ask Jonny Flynn.
Oh yeah. It also helps to get rid of Joey Dorsey. 45-119 FT, 37.8%
Memphis 2009 FT: 69.3% (153rd)
I've known for awhile that Memphis' FT Shooting has improved since last year - it's just been nothing to write home about until I realized today that Tyreke Evans isn't much of a charity stripe lock(69.6% on 161 attempts). The reason this struck me as awkward was that I assumed he was one of the causes for the upward movement given that the better shooters from last year's team (CDR and Rose) are now playing in the NBA.
So where is this improvement coming from?
Granted, it's a lot easier to improve in anything when you are already poor at it, but this kind of improvement, esp. from Taggart and Anderson are anything but typical. Without it, Memphis' mediocre offense (40th) would be much lower to the point where it would be something to worry about. Improving your FT shooting esp. in close games is anything but smexy, but PPP doesn't give a shit about where you get your points from. Just ask Jonny Flynn.
Oh yeah. It also helps to get rid of Joey Dorsey. 45-119 FT, 37.8%
Friday, March 6, 2009
Bracket Odds - 9 Days Until Selection Sunday
Favorites Update:
[Seed]Team: Odds% (Change)
1. [2]Memphis: 17.2% (UP 10.2%)
2. [1]North Carolina: 13.3%(UP 0.2%)
3. [1]Pittsburgh: 11.1% (UP 0.4%)
4. [1]Connecticut: 8.3%(DOWN 3.4%)
5. [2]Duke: 5.2% (DOWN 6.8%)
6. [4]UCLA: 4.6% (DOWN 0.4%)
7. [2]Louisville: 4.0% (DOWN 1.0%)
8. [7]West Virginia: 3.4% (UP 0.7%)
9. [2]Michigan St.: 3.4% (DOWN 0.3%)
10. [3]Kansas: 3.3% (UP 1.6%)
Dropped Out: [4]Villanova: 1.8% (DOWN 0.8%)
Other Notables(Above 2.0%)
Gonzaga: 2.9%
Missouri: 2.9%
Oklahoma: 2.6%
BYU: 2.3%
Washington: 2.1%
It's been quite awhile since our last update, but only a couple major changes loom. Memphis has escalated to the top of the efficiency rankings. They benefit from an easy road in this update, but would nonetheless be favorites in a schedule neutral bracket. Duke, despite it's recent rebound, has seen it's efficiency numbers take a dip. BYU registers with non-trivial odds to be this year's dark horse.
Monday, March 2, 2009
Bracket Analysis - Potential March Madness Home Court Advantage
Like all sports, homecourt advantage in college basketball consists of a variety of factors:
1. Travel: actual flying or bus, time zone adjustments, not sleeping in your own bed/typical gameday routine, distractions due to site seeing/partying, loss of preparation time, being away from friends and family.
2. Fans: heckling, noise/distractions during free throws, shots, and possessions, counting down the shot clock, Duke speedo guy, etc.
3. Refs: Influence of the homecrowd, home conference refs who may show preferential treatment to the home conference or at least be used to their play(ie. physical play not being a foul, etc.)
4. Equipment: not being used to floors/backboard/nets/rims, home players being familiar with images on the floor(know when they are near out of bounds/3pt line/etc. in a scramble), shape of a stadium changing the way things look or feel.
Depending on how many of these factors are present and to what degree, a typical DI NCAA game is worth about ~4.5 pts of HCA, though some as few as 1 and some as many as 7. Oftentimes, teams playing at favorable March Madness sites - near home with a large proportion of their fans attending, or at a site they play in conference every year - are the benefactors of a few points of HCA, despite the fact that are games are considered "neutral".
Here, I go over the teams that are likely to be placed in a site that will provide them with HCA and what seeds they have to get in order to attain them based upon bracketing rules. It's important to note that because of the way brackets are created, none of these situations are guaranteed, but the higher a team makes it's way up the S-Curve, the more likely it is to attain these benefits.
It is worthwhile to look through some basic bracketing rules to understand why how some of these pods are created, but here are some keys ones:
- There are 8 first weekend locations for each 4 pack of teams:
Miami, FL
Philadelphia, PA
Greensboro, NC
Dayton, OH
Kansas City, MO
Minneapolis, MN
Boise, ID
Portland, OR
- The selection committee tries to protect the top seed in each 4 pack from being at a geographical disadvantage in the 1st weekend, even if it is unable to gain an advantage. Hence, the term "protected seeds" for #1-#4s.
-Each city can host 2 foursomes(ie. 1/16//8/9 or 3/14//6/11, etc.), so that means only 2 top teams can earn the same preferred site, even if more than 2 are protected seeds(see UNC/Duke/Wake Forest trio later). Often, this means that if you're not one of those two, you might lucksack HCA by being a #5 seed or worse instead.
I have tried to guesstimate HCA points to each team based upon proximity to school, whether or not they've played in that city/arena before, and likelyhood of fanbase to travel with some attempt to scale for size of fanbase. Most of these are between 1-3 pts as it's very difficult to simulate normal HCA levels in a gym that is not your own.
A few other cavets before I begin:
- HCA can be relative, even though both teams may travel a sizeable distance. If UCLA plays Duke in Seattle hypothetically, I would expect some minor HCA in UCLA's favor due to 1) larger travel/time zone shifts for the Duke, 2)Pac 10 fans in Washington cheering for UCLA, etc. For the purposes of this analysis, I have just listed HCAs that likely exist regardless of matchups.
- There are often some hidden HCAs attainable by the non-protected seeds even in the first 2 rounds(see (10)Davidson in Raleigh, (11)Kansas St. in Omaha last year). These often result due to the limited time and complex bracketing rules the committee is faced with. They are difficult to predict and will be noted here in brevity.
- I have only included teams that have a reasonable chance at being placed in a spot that helps them. Clemson(projected #4 seed) may benefit some from being in Greensboro but there's almost no chance they get put there given that those pods will be led by a #1-#3 seed. On the other hand, South Carolina as a #9 has a legitimate chance at being there and is therefore listed.
Notation:
#(Current projected seed) Team, Location (HCA Pts)
Round of 64/32
#1 UNC – Greensboro (3)
#3 Duke – Greensboro (3)
#3 Wake Forest – Greensboro (3)
#9 South Carolina - Greensboro (1)
==========
Of Wake/Duke/UNC, only 2 of these teams will be at Greensboro. The next closest locations are Dayton and Philadelphia for whatever team that doesn’t get to stay in-state, and those sites have competition as well(Pittsburgh/Villanova/Louisville/Memphis/Michigan St.) Needless to say, Duke and Wake Forest still have something to play for in the next few days and the Tar Heels don’t want to completely collapse in the next 3 games either.
#1 Oklahoma – Kansas City (1)
#2 Kansas – Kansas City (3)
#3 Missouri – Kansas City (3)
==========
Oklahoma, Kansas, Missouri, and Memphis are all competing to be in KC. This means that Wednesdays’(3/4) game between the Sooners and (Missouri)Tigers as well as the potential rematch between the two teams in the Big 12 donkament is definitely more than just a statement game. Kansas’ inner track to the #1 seed in the Big 12 Donkament where it can avoid both other teams is critical as well.
Each of the Big 12 teams is also hoping the Memphis Tigers’ stub their toes tomorrow @Houston as that will be the only remaining game where they aren’t double digit favorites(Memphis -9.2). Memphis plays the CUSA Donkament at home and thus are hudge favorites to win that(79%+). There is an off-chance that Memphis could be shipped to Dayton even if they end up ahead of all of these teams on the S-Curve.
#1 Pittsburgh – Philadelphia (2)
#3 Villanova – Philadelphia (4)
==========
At least in ESPN’s bracket, this pod seems to be forgotten after Lunardi slots in whichever of the 2 Big East powers is ahead at the time(Pittsburgh or UConn). I’ve seen both teams placed here, one of these teams with Villanova, or just 1 of these teams with someone else like Wake Forest, so it’s really hard to tell if there are some complex bracketing rules that are causing all this fluctuation.
This site is the nuts for Villanova if they can attain it. As far as I know, it’s the only site a team is allowed to go to that is within half an hour of the campus.
#2 Louisville – Dayton (3)
#2 Michigan St. - Dayton (1)
#5 Xavier – Dayton (3)
#10 Ohio St. - Dayton (3)
==========
Pittsburgh is slotted for Dayton as their backup if they don’t land a spot in Philadelphia. In that case, only one of these teams would be at Dayton to lead their 4-pack. Michigan St. could also be headed for Minneapolis even if it’s ahead of Louisville, so the Cardinals have a decent shot of a short trip and playing behind the home crowd regardless of what happens in their final 2 games and in MSG. Xavier is a long shot at this point given the logjam for this site(listed teams + Pitt/Memphis/3rd wheel of the Oklahoma/Kansas/Missouri trio), but has an outside chance at a large advantage, one that Lunardi has Ohio St. also having at this point.
#3 Washington – Portland (2)
#6 Gonzaga – Portland (1)
#6 Gonzaga – Boise (1)
==========
The lack of teams with elite resumes from the West Coast/Northwest means that Washington will probably be in Portland no matter what it does against Washington St. this week or in the Pac 10 Donkament next week.
The converse is also true. Very few top teams are jostling for the opportunity to be in either Oregon or Idaho. This means that you should not be surprised if the worst protected seeds(Clemson, Illinois, Purdue, Missouri in this bracket) or the 5’s just below them (Xavier, LSU, Marquette, UCLA(not as bad)) have to fly into the Pacific time zone and face someone like Utah St. or Arizona as a 12 seed and then Gonzaga or UCLA as a 5 seed.
#11 Florida – Miami (1)
#11 Florida St. – Miami (1)
#9 Wisconsin – Minneapolis (2)
==========
You have a similar phenomenon for the Miami and Minneapolis sites, but the HCA effect is less extreme as teams won’t be flying across the country through time zones.
Again, these estimates are rough and based on a rough knowledge of fan bases and GoogleMaps. If you've spotted some errors in judgment or a team I've missed, please leave a comment.
In a later update I will tackle the 2nd weekend and the Final Four.
1. Travel: actual flying or bus, time zone adjustments, not sleeping in your own bed/typical gameday routine, distractions due to site seeing/partying, loss of preparation time, being away from friends and family.
2. Fans: heckling, noise/distractions during free throws, shots, and possessions, counting down the shot clock, Duke speedo guy, etc.
3. Refs: Influence of the homecrowd, home conference refs who may show preferential treatment to the home conference or at least be used to their play(ie. physical play not being a foul, etc.)
4. Equipment: not being used to floors/backboard/nets/rims, home players being familiar with images on the floor(know when they are near out of bounds/3pt line/etc. in a scramble), shape of a stadium changing the way things look or feel.
Depending on how many of these factors are present and to what degree, a typical DI NCAA game is worth about ~4.5 pts of HCA, though some as few as 1 and some as many as 7. Oftentimes, teams playing at favorable March Madness sites - near home with a large proportion of their fans attending, or at a site they play in conference every year - are the benefactors of a few points of HCA, despite the fact that are games are considered "neutral".
Here, I go over the teams that are likely to be placed in a site that will provide them with HCA and what seeds they have to get in order to attain them based upon bracketing rules. It's important to note that because of the way brackets are created, none of these situations are guaranteed, but the higher a team makes it's way up the S-Curve, the more likely it is to attain these benefits.
It is worthwhile to look through some basic bracketing rules to understand why how some of these pods are created, but here are some keys ones:
- There are 8 first weekend locations for each 4 pack of teams:
Miami, FL
Philadelphia, PA
Greensboro, NC
Dayton, OH
Kansas City, MO
Minneapolis, MN
Boise, ID
Portland, OR
- The selection committee tries to protect the top seed in each 4 pack from being at a geographical disadvantage in the 1st weekend, even if it is unable to gain an advantage. Hence, the term "protected seeds" for #1-#4s.
-Each city can host 2 foursomes(ie. 1/16//8/9 or 3/14//6/11, etc.), so that means only 2 top teams can earn the same preferred site, even if more than 2 are protected seeds(see UNC/Duke/Wake Forest trio later). Often, this means that if you're not one of those two, you might lucksack HCA by being a #5 seed or worse instead.
I have tried to guesstimate HCA points to each team based upon proximity to school, whether or not they've played in that city/arena before, and likelyhood of fanbase to travel with some attempt to scale for size of fanbase. Most of these are between 1-3 pts as it's very difficult to simulate normal HCA levels in a gym that is not your own.
A few other cavets before I begin:
- HCA can be relative, even though both teams may travel a sizeable distance. If UCLA plays Duke in Seattle hypothetically, I would expect some minor HCA in UCLA's favor due to 1) larger travel/time zone shifts for the Duke, 2)Pac 10 fans in Washington cheering for UCLA, etc. For the purposes of this analysis, I have just listed HCAs that likely exist regardless of matchups.
- There are often some hidden HCAs attainable by the non-protected seeds even in the first 2 rounds(see (10)Davidson in Raleigh, (11)Kansas St. in Omaha last year). These often result due to the limited time and complex bracketing rules the committee is faced with. They are difficult to predict and will be noted here in brevity.
- I have only included teams that have a reasonable chance at being placed in a spot that helps them. Clemson(projected #4 seed) may benefit some from being in Greensboro but there's almost no chance they get put there given that those pods will be led by a #1-#3 seed. On the other hand, South Carolina as a #9 has a legitimate chance at being there and is therefore listed.
Notation:
#(Current projected seed) Team, Location (HCA Pts)
Round of 64/32
#1 UNC – Greensboro (3)
#3 Duke – Greensboro (3)
#3 Wake Forest – Greensboro (3)
#9 South Carolina - Greensboro (1)
==========
Of Wake/Duke/UNC, only 2 of these teams will be at Greensboro. The next closest locations are Dayton and Philadelphia for whatever team that doesn’t get to stay in-state, and those sites have competition as well(Pittsburgh/Villanova/Louisville/Memphis/Michigan St.) Needless to say, Duke and Wake Forest still have something to play for in the next few days and the Tar Heels don’t want to completely collapse in the next 3 games either.
#1 Oklahoma – Kansas City (1)
#2 Kansas – Kansas City (3)
#3 Missouri – Kansas City (3)
==========
Oklahoma, Kansas, Missouri, and Memphis are all competing to be in KC. This means that Wednesdays’(3/4) game between the Sooners and (Missouri)Tigers as well as the potential rematch between the two teams in the Big 12 donkament is definitely more than just a statement game. Kansas’ inner track to the #1 seed in the Big 12 Donkament where it can avoid both other teams is critical as well.
Each of the Big 12 teams is also hoping the Memphis Tigers’ stub their toes tomorrow @Houston as that will be the only remaining game where they aren’t double digit favorites(Memphis -9.2). Memphis plays the CUSA Donkament at home and thus are hudge favorites to win that(79%+). There is an off-chance that Memphis could be shipped to Dayton even if they end up ahead of all of these teams on the S-Curve.
#1 Pittsburgh – Philadelphia (2)
#3 Villanova – Philadelphia (4)
==========
At least in ESPN’s bracket, this pod seems to be forgotten after Lunardi slots in whichever of the 2 Big East powers is ahead at the time(Pittsburgh or UConn). I’ve seen both teams placed here, one of these teams with Villanova, or just 1 of these teams with someone else like Wake Forest, so it’s really hard to tell if there are some complex bracketing rules that are causing all this fluctuation.
This site is the nuts for Villanova if they can attain it. As far as I know, it’s the only site a team is allowed to go to that is within half an hour of the campus.
#2 Louisville – Dayton (3)
#2 Michigan St. - Dayton (1)
#5 Xavier – Dayton (3)
#10 Ohio St. - Dayton (3)
==========
Pittsburgh is slotted for Dayton as their backup if they don’t land a spot in Philadelphia. In that case, only one of these teams would be at Dayton to lead their 4-pack. Michigan St. could also be headed for Minneapolis even if it’s ahead of Louisville, so the Cardinals have a decent shot of a short trip and playing behind the home crowd regardless of what happens in their final 2 games and in MSG. Xavier is a long shot at this point given the logjam for this site(listed teams + Pitt/Memphis/3rd wheel of the Oklahoma/Kansas/Missouri trio), but has an outside chance at a large advantage, one that Lunardi has Ohio St. also having at this point.
#3 Washington – Portland (2)
#6 Gonzaga – Portland (1)
#6 Gonzaga – Boise (1)
==========
The lack of teams with elite resumes from the West Coast/Northwest means that Washington will probably be in Portland no matter what it does against Washington St. this week or in the Pac 10 Donkament next week.
The converse is also true. Very few top teams are jostling for the opportunity to be in either Oregon or Idaho. This means that you should not be surprised if the worst protected seeds(Clemson, Illinois, Purdue, Missouri in this bracket) or the 5’s just below them (Xavier, LSU, Marquette, UCLA(not as bad)) have to fly into the Pacific time zone and face someone like Utah St. or Arizona as a 12 seed and then Gonzaga or UCLA as a 5 seed.
#11 Florida – Miami (1)
#11 Florida St. – Miami (1)
#9 Wisconsin – Minneapolis (2)
==========
You have a similar phenomenon for the Miami and Minneapolis sites, but the HCA effect is less extreme as teams won’t be flying across the country through time zones.
Again, these estimates are rough and based on a rough knowledge of fan bases and GoogleMaps. If you've spotted some errors in judgment or a team I've missed, please leave a comment.
In a later update I will tackle the 2nd weekend and the Final Four.
Friday, February 27, 2009
Quick Blurbs - Why Conventional Arguments Can't Discredit Memphis for a #1 Seed
Teams behind Memphis in Pythag SoS, that u will never hear "they played a soft schedule" about simply because of conference affiliation:
Memphis: .6793(47th)
---
Wake Forest: .6779(48th)
Purdue: .6710(54th)
Oklahoma: .6693(56th)
Clemson: .6541(69th)
Marquette: .6458(73rd)
Missouri: .6090(90th)
You can do the same exercise with RPI and find the same(if not more teams) result.
The other main argument I hear is that these numbers have been reached without any marquee wins. Winning @Gonzaga by 18 certainly qualifies as that, though conventional wisdom not valuing Gonzaga(even though the market does) or even OOC road games is not something that surprises me.
It's worth noting that there are legitimate arguments at this point for keeping them as a #2 seed - Oklahoma can have it's losses without Blake Griffin reduced in weight, and otherwise it's W-L/SoS resume meets the test and Connecticut has not yet played worse without Dyson. That's not the point. The point is that to throw out qualifiers without going through the numbers is embarrassing, especially when you're an analyst on a large sports network.
Will try to do a favorites update after Poofler updates odds today.
Memphis: .6793(47th)
---
Wake Forest: .6779(48th)
Purdue: .6710(54th)
Oklahoma: .6693(56th)
Clemson: .6541(69th)
Marquette: .6458(73rd)
Missouri: .6090(90th)
You can do the same exercise with RPI and find the same(if not more teams) result.
The other main argument I hear is that these numbers have been reached without any marquee wins. Winning @Gonzaga by 18 certainly qualifies as that, though conventional wisdom not valuing Gonzaga(even though the market does) or even OOC road games is not something that surprises me.
It's worth noting that there are legitimate arguments at this point for keeping them as a #2 seed - Oklahoma can have it's losses without Blake Griffin reduced in weight, and otherwise it's W-L/SoS resume meets the test and Connecticut has not yet played worse without Dyson. That's not the point. The point is that to throw out qualifiers without going through the numbers is embarrassing, especially when you're an analyst on a large sports network.
Will try to do a favorites update after Poofler updates odds today.
Tuesday, February 24, 2009
Quick Blurbs - Update On Louisville's Offense
I commented on Louisville's offense a week ago in this post.
Here is an update via Basketball Prospectus:
"Louisville looked really good last night, winning at Georgetown 76-58. (Any team looks good when they make their first seven threes.) In fact the Cards seem to have turned over something of a new leaf since their debacle at Notre Dame 12 days ago. Rick Pitino's team has rebounded from their 33-point loss in South Bend to score 1.28 points per trip over their last four games, an impressive figure even given the uneven competition (DePaul, Providence, Cincinnati and the Hoyas). This miniature success story has had many authors, but it certainly hasn't hurt matters that Preston Knowles has gone 13-of-20 on his threes over that span. Needless to say, if Louisville can match their excellent defense with good offense, they will be a very tough out next month, no matter how highly-seeded the opponent."
The Cardinals did play Depaul(224th in Def. Eff.) and Providence (139th) in that stretch so the numbers could be somewhat inflated, but performances against Cincinnati(78th) and Georgetown (38th) are legitimate. Overall, their offensive eff. numbers have improved a bit from 56th to 47th at 110.9 points/100 possession.
Here is an update via Basketball Prospectus:
"Louisville looked really good last night, winning at Georgetown 76-58. (Any team looks good when they make their first seven threes.) In fact the Cards seem to have turned over something of a new leaf since their debacle at Notre Dame 12 days ago. Rick Pitino's team has rebounded from their 33-point loss in South Bend to score 1.28 points per trip over their last four games, an impressive figure even given the uneven competition (DePaul, Providence, Cincinnati and the Hoyas). This miniature success story has had many authors, but it certainly hasn't hurt matters that Preston Knowles has gone 13-of-20 on his threes over that span. Needless to say, if Louisville can match their excellent defense with good offense, they will be a very tough out next month, no matter how highly-seeded the opponent."
The Cardinals did play Depaul(224th in Def. Eff.) and Providence (139th) in that stretch so the numbers could be somewhat inflated, but performances against Cincinnati(78th) and Georgetown (38th) are legitimate. Overall, their offensive eff. numbers have improved a bit from 56th to 47th at 110.9 points/100 possession.
Monday, February 16, 2009
Fun Fact - Notre Dame Takes Care of the Ball, and it's Opponent's Ball Too?
Notre Dame leads the nation in both having the least number of turnovers and is last in the nation in forcing turnovers.
Unsurprisingly, the Fighting Irish also get to the line relatively infrequently(3rd to last in the nation) and send opponents very few times as well(11th). Just one example of how an aggressive defensive style is oftentimes correlated with fouls(strips that are sometimes reach-in's, etc.). A quick glance at TO% leaders confirms this with a few exceptions.
Unsurprisingly, the Fighting Irish also get to the line relatively infrequently(3rd to last in the nation) and send opponents very few times as well(11th). Just one example of how an aggressive defensive style is oftentimes correlated with fouls(strips that are sometimes reach-in's, etc.). A quick glance at TO% leaders confirms this with a few exceptions.
Saturday, February 7, 2009
Bracket Odds - 33 Days Until Selection Sunday
Updated based on games and bracket projection through Sunday
Favorites Update:
[Seed]Team: Odds% (Odds% Last Update/Change)
1. [1]North Carolina: 13.1%(16.5%/DOWN 3.4%)
2. [2]Duke: 12.0% (13.0%/UP 1.0%)
3. [1]Connecticut: 11.7%(16.2%/DOWN 4.5%)
4. *[1]Pittsburgh: 10.4% (7.9%/UP 2.5%)
5. [3]Memphis: 7.0% (3.0%/UP 4.0%)
6. [4]UCLA: 5.0% (4.0%/UP 1.0%)
7. [2]Louisville: 5.0% (4.2%/UP 0.8%)
8. [2]Michigan St. 3.8% (2.3%/UP 1.5%)
9. *[8]West Virginia 2.7% (1.7%/UP 1.0%)
10. [4]Villanova: 2.6% (1.5%/UP 1.1%)
*These numbers were calculated before Pittsburgh defeated West Virginia by 11 points today. Kenpom's numbers had Pittsburgh -5.3 as fair, so one can expect both teams to move in the appropriate directions. The Panthers just knocked off the 3rd hardest game remaining in their schedule, so if they can hold serve elsewhere and win 1 of 2 against Connecticut, they will solidify their hold on a #1 seed. West Virginia becomes the latest victim of the monster that is the Big East, playing well against tough competition and blowing out mediocre teams, but not developing a good resume in the process, making them a ripe team to fall into bubble territory and surprise some people come March(Something I will comment on more later).
Dropped out:
11. [1]Oklahoma: 2.4% (2.6%/DOWN 0.2%)
12. [3]Clemson: 2.3% (3.1%/DOWN 0.8%)
14. [6]Gonzaga: 2.0% (5.7%/DOWN 3.7%)
Memphis didn't just flip places with Gonzaga. They flipped them on their belly and raped them. Again, losses like this are damaging in 2 ways. They hurt your profile, most likely hurting your odds at an easier draw, and also make your team worse by both the market and most metrics when they are major underperformances.
From this point on, Memphis is favored in every remaining game, is only expected to drop one, and has an outside chance at running the table in the rest of CUSA play. They definitely have the inside track at picking up another seed bump when the one of the #2 seeds begin to falter(Marquette did much earlier than expected). A #1 seed is not out of the question. The winner of the Memphis/Gonzaga game is the real Oklahoma that people are talking about(team with an easy schedule avoiding the gauntlet of ACC/BE play) slated for a spot on the top line.
The other fluctuations are largely due to the rearrangement of the bracket. In being the #1 overall seed, Connecticut does get the easiest #2 seed in Wake Forest(17th)(bracket construction success!), but also gets Memphis as the #3(5th), UCLA as the #4(6th), Missouri as the #5(9th). This grouping - Keith and I are calling it "the bracket of death" - was actually in North Carolina's side last week and was moved together.
Tomorrow I comment on teams that will be ripe for making an unexpected deep run in March. I will pick teams outside of the protected seeds, and possibly avoid teams I've already commented on(Gonzaga).
Favorites Update:
[Seed]Team: Odds% (Odds% Last Update/Change)
1. [1]North Carolina: 13.1%(16.5%/DOWN 3.4%)
2. [2]Duke: 12.0% (13.0%/UP 1.0%)
3. [1]Connecticut: 11.7%(16.2%/DOWN 4.5%)
4. *[1]Pittsburgh: 10.4% (7.9%/UP 2.5%)
5. [3]Memphis: 7.0% (3.0%/UP 4.0%)
6. [4]UCLA: 5.0% (4.0%/UP 1.0%)
7. [2]Louisville: 5.0% (4.2%/UP 0.8%)
8. [2]Michigan St. 3.8% (2.3%/UP 1.5%)
9. *[8]West Virginia 2.7% (1.7%/UP 1.0%)
10. [4]Villanova: 2.6% (1.5%/UP 1.1%)
*These numbers were calculated before Pittsburgh defeated West Virginia by 11 points today. Kenpom's numbers had Pittsburgh -5.3 as fair, so one can expect both teams to move in the appropriate directions. The Panthers just knocked off the 3rd hardest game remaining in their schedule, so if they can hold serve elsewhere and win 1 of 2 against Connecticut, they will solidify their hold on a #1 seed. West Virginia becomes the latest victim of the monster that is the Big East, playing well against tough competition and blowing out mediocre teams, but not developing a good resume in the process, making them a ripe team to fall into bubble territory and surprise some people come March(Something I will comment on more later).
Dropped out:
11. [1]Oklahoma: 2.4% (2.6%/DOWN 0.2%)
12. [3]Clemson: 2.3% (3.1%/DOWN 0.8%)
14. [6]Gonzaga: 2.0% (5.7%/DOWN 3.7%)
Memphis didn't just flip places with Gonzaga. They flipped them on their belly and raped them. Again, losses like this are damaging in 2 ways. They hurt your profile, most likely hurting your odds at an easier draw, and also make your team worse by both the market and most metrics when they are major underperformances.
From this point on, Memphis is favored in every remaining game, is only expected to drop one, and has an outside chance at running the table in the rest of CUSA play. They definitely have the inside track at picking up another seed bump when the one of the #2 seeds begin to falter(Marquette did much earlier than expected). A #1 seed is not out of the question. The winner of the Memphis/Gonzaga game is the real Oklahoma that people are talking about(team with an easy schedule avoiding the gauntlet of ACC/BE play) slated for a spot on the top line.
The other fluctuations are largely due to the rearrangement of the bracket. In being the #1 overall seed, Connecticut does get the easiest #2 seed in Wake Forest(17th)(bracket construction success!), but also gets Memphis as the #3(5th), UCLA as the #4(6th), Missouri as the #5(9th). This grouping - Keith and I are calling it "the bracket of death" - was actually in North Carolina's side last week and was moved together.
Tomorrow I comment on teams that will be ripe for making an unexpected deep run in March. I will pick teams outside of the protected seeds, and possibly avoid teams I've already commented on(Gonzaga).
Quick Blurbs - Why the Sooners Aren't Elite
Oklahoma is a 1 loss team, undefeated in conference play, and ranked 2nd in the nation in both the AP and coaches' polls. Spearheaded by a NBA lottery pick in Blake Griffin who KenPom loves on the basis of his overall efficiency, terrific Rebounding% numbers and an almost Hansborough-like ability to get to the foul line, the Sooners find themselves as the disappointing 18th most efficient team in the nation and the 10th team most likely to win it all despite being projected as a coveted #1 seed. The Sooners have played a legitimate schedule(52nd), with more of it to come so it's not as if they haven't gotten these wins against Dance-level competition. What is the disconnect?
Favored by 20 in Vegas(I will start to include these as well in my analysis, I think it's a good secondary metric to my model) and by 20.6 in our model, the Sooners beat Colorado(144th) by only 5 pts today - a 15 pt underperfomance of already low expectations. This underperfomance would had been equivalent to Depaul(192nd) upsetting Pittsburgh(4th) at home today.
Granted, you can cherry pick bad games from any elite team's resume, but the reason Oklahoma is ranked surprisingly low is that they have significantly more of these good, but not top 5 performances in their history. It just happens that in their case, these mediocre performances are disguised as close wins against middling or weak competition:
Date Opponent(Rank), Expectation(Result, Amount Underperfomance)
11/22 v.Gardner Webb(208th), Oklahoma -26.9 (Win by 4, 22.9 pt Underperfomance)
12/22 @Rice(258th), Oklahoma -22.1 (Win by 12, 10.1 pt Underperfomance)
12/30 @Arkansas(114th), Oklahoma -11.2 (Lost by 8, 19.2 pt Underperformance)
2/7 v.Colorado(144th), Oklahoma -20.6 (Win by 5, 15.6 pt Underperfomance)
*Some would argue that it doesn't matter what you do against bad teams as long as you win, but that is the topic of another article that has already been written many times. Here, it is suffice to say that it does matter to the point where if you're only edging out bad teams, it does not predict well of your chances against elite teams.
It's important that these expectations are for Oklahoma or teams similar to the Sooners such as Clemson, Purdue, or Arizona St. If Griffins' and Co. really wanted to be graded on an elite scale - Pittsburgh/Connecticut/Duke/North Carolina, you can tack on 3-5 pts(depends on the pace of the team) to the expected win amount.
None of the aforementioned teams have a loss as bad as @Arkansas, and the data shows they don't have as many underperfomances either.
Worst Losses:
Duke: @Michigan(67th)
UNC: v.Boston College(63rd)
Connecticut: v.Georgetown(23rd)
Pittsburgh: @Villanvoa(15th)
Memphis: v.Syracuse(35rd)
UCLA: N-Michigan(67th)
It's important to remember that the Sooners have a slew of good wins(N-UAB, N-Purdue, v.Utah, @Kansas St, v.Texas), and have their share of blowouts against mediocre teams[29 pt win v. American(115th)] and even good teams as well[(19 pt win v.Baylor(45th)]. I wouldn't be surprised if they started to exceed these expectations(I would lean on the over) or even made a deep run in the tournament. However, their current overall body of work and thus expectations indicates they are closer to teams like Clemson, Wake Forest, Washington, and Xavier, than they are to the North Carolinas and Connecticuts of the world. And like those first 4 teams, making it to the Elite 8 would be running good - not bad.
Favored by 20 in Vegas(I will start to include these as well in my analysis, I think it's a good secondary metric to my model) and by 20.6 in our model, the Sooners beat Colorado(144th) by only 5 pts today - a 15 pt underperfomance of already low expectations. This underperfomance would had been equivalent to Depaul(192nd) upsetting Pittsburgh(4th) at home today.
Granted, you can cherry pick bad games from any elite team's resume, but the reason Oklahoma is ranked surprisingly low is that they have significantly more of these good, but not top 5 performances in their history. It just happens that in their case, these mediocre performances are disguised as close wins against middling or weak competition:
Date Opponent(Rank), Expectation(Result, Amount Underperfomance)
11/22 v.Gardner Webb(208th), Oklahoma -26.9 (Win by 4, 22.9 pt Underperfomance)
12/22 @Rice(258th), Oklahoma -22.1 (Win by 12, 10.1 pt Underperfomance)
12/30 @Arkansas(114th), Oklahoma -11.2 (Lost by 8, 19.2 pt Underperformance)
2/7 v.Colorado(144th), Oklahoma -20.6 (Win by 5, 15.6 pt Underperfomance)
*Some would argue that it doesn't matter what you do against bad teams as long as you win, but that is the topic of another article that has already been written many times. Here, it is suffice to say that it does matter to the point where if you're only edging out bad teams, it does not predict well of your chances against elite teams.
It's important that these expectations are for Oklahoma or teams similar to the Sooners such as Clemson, Purdue, or Arizona St. If Griffins' and Co. really wanted to be graded on an elite scale - Pittsburgh/Connecticut/Duke/North Carolina, you can tack on 3-5 pts(depends on the pace of the team) to the expected win amount.
None of the aforementioned teams have a loss as bad as @Arkansas, and the data shows they don't have as many underperfomances either.
Worst Losses:
Duke: @Michigan(67th)
UNC: v.Boston College(63rd)
Connecticut: v.Georgetown(23rd)
Pittsburgh: @Villanvoa(15th)
Memphis: v.Syracuse(35rd)
UCLA: N-Michigan(67th)
It's important to remember that the Sooners have a slew of good wins(N-UAB, N-Purdue, v.Utah, @Kansas St, v.Texas), and have their share of blowouts against mediocre teams[29 pt win v. American(115th)] and even good teams as well[(19 pt win v.Baylor(45th)]. I wouldn't be surprised if they started to exceed these expectations(I would lean on the over) or even made a deep run in the tournament. However, their current overall body of work and thus expectations indicates they are closer to teams like Clemson, Wake Forest, Washington, and Xavier, than they are to the North Carolinas and Connecticuts of the world. And like those first 4 teams, making it to the Elite 8 would be running good - not bad.
Friday, February 6, 2009
Bracket Odds - 37 Days Until Selection Sunday
ESPN decided to start updating on Friday as well, therefore we will be bringing you 2 bracket odds updates per week. You can find the odds here, updated through Thursday's games.
Keith's Update
Favorites Update:
[Seed]Team: Odds% (Odds% Last Update/Change)
1. [1]North Carolina: 16.5%(15.2%/UP 1.3%)
2. [1]Connecticut: 16.2%(10.2%/UP 6%)
3. [2]Duke: 12.0% (23.1%/DOWN 11.1%)
4. [1]Pittsburgh: 7.9% (8.7%/UP 0.8%)
5. [4]Gonzaga: 5.7% (3.9%/UP 1.8%)
6. [2]Louisville: 4.2% (5.8%/UP 1.6%)
7. [4]UCLA: 4.0% (2.6%/UP 1.4%)
8. [3]Clemson: 3.1% (0.8%/UP 2.3%)
9. [3]Memphis: 3.0% (4.4%/DOWN 1.4%)
10. [1]Oklahoma: 2.6% (2.8%/DOWN 0.2%)
Dropped out:
Wake Forest 1.9% (3.2%/DOWN 1.3%)
Duke's loss is everyone's gain, but mostly UConn and Clemson's, despite the fact that neither improved it's seeding in this update. The model we use doesn't directly show any recency bias, so all games are weighted equally. This illustrates how 1 large overperfomance against another quality team can change even though it's only 1 game in 20 by this point.(UConn's 17pt win @Louisville, pre-game line was Louisville -2.1, Clemson's 29 pt win v. Duke, pre game Clemson +3.7)
It's interesting to note the absence of the other 2 seeds - Michigan St. and Marquette. Even with better seeding than some of the favorites on the list, Kenpom doesn't consider either to be amongst the most efficient teams in CBB nation - MSU/16th, and Marquette/22nd respectively.
As a result of this and the tough schedules each team is about to face, neither is a favorite to hold onto it's #2 seed come Selection Sunday unless either team performs above expectation. Marquette has the more firm resume thus far, but the Golden Eagles also have the more daunting schedule.
Michigan St.(8 games remaining)
2/10 Michigan St. -4.9 @Michigan
2/17 Michigan St. +3.8 @Purdue
3/1 Michigan St. +0.9 @Illinois
3/8 Michigan St. -3.2 v.Purdue
Marquette (9 games remaining)
2/10 Marquette +4.3 @Villanova
2/21 Marquette +3.7 @Georgetown
2/25 Marquette +3.1 v.UConn
3/1 Marquette +6.2 @Louisville
3/4 Marquette +8.9 @Pittsburgh
This is exactly what I mean when I say that Lunardi does a poor job in predicting the future. If today was Selection Sunday, his picks would likely be right, but if Marquette ran at expectation in these 5 games, it would come out with 1.4 wins. Let me say this again. The Marquette Golden Eagles will be running good by going 2-3 in this stretch. It doesn't take an expert bracketologist to realize that the they will drop out of their current position given that record.
Who then is in the best position to snatch up these coveted 2 seeds that these teams may vacate? We've already broached the outside possibility of UCLA getting up there, but who else?
Xavier, Gonzaga, Clemson, Wake Forest, Missouri(outside shot).
What do these teams have in common? They are all just behind the #2 seeds listed above and have easy schedules the rest of the way. Some even have the inside track on their conference or will be heavy favorites in their conference tournament, both of which the selection committee seems to value based upon looking at past years. MIGHT INSERT SOME DATA LATER TONIGHT. PLAYING FTOPS DONKAMENT GOGO.
Keith's Update
Favorites Update:
[Seed]Team: Odds% (Odds% Last Update/Change)
1. [1]North Carolina: 16.5%(15.2%/UP 1.3%)
2. [1]Connecticut: 16.2%(10.2%/UP 6%)
3. [2]Duke: 12.0% (23.1%/DOWN 11.1%)
4. [1]Pittsburgh: 7.9% (8.7%/UP 0.8%)
5. [4]Gonzaga: 5.7% (3.9%/UP 1.8%)
6. [2]Louisville: 4.2% (5.8%/UP 1.6%)
7. [4]UCLA: 4.0% (2.6%/UP 1.4%)
8. [3]Clemson: 3.1% (0.8%/UP 2.3%)
9. [3]Memphis: 3.0% (4.4%/DOWN 1.4%)
10. [1]Oklahoma: 2.6% (2.8%/DOWN 0.2%)
Dropped out:
Wake Forest 1.9% (3.2%/DOWN 1.3%)
Duke's loss is everyone's gain, but mostly UConn and Clemson's, despite the fact that neither improved it's seeding in this update. The model we use doesn't directly show any recency bias, so all games are weighted equally. This illustrates how 1 large overperfomance against another quality team can change even though it's only 1 game in 20 by this point.(UConn's 17pt win @Louisville, pre-game line was Louisville -2.1, Clemson's 29 pt win v. Duke, pre game Clemson +3.7)
It's interesting to note the absence of the other 2 seeds - Michigan St. and Marquette. Even with better seeding than some of the favorites on the list, Kenpom doesn't consider either to be amongst the most efficient teams in CBB nation - MSU/16th, and Marquette/22nd respectively.
As a result of this and the tough schedules each team is about to face, neither is a favorite to hold onto it's #2 seed come Selection Sunday unless either team performs above expectation. Marquette has the more firm resume thus far, but the Golden Eagles also have the more daunting schedule.
Michigan St.(8 games remaining)
2/10 Michigan St. -4.9 @Michigan
2/17 Michigan St. +3.8 @Purdue
3/1 Michigan St. +0.9 @Illinois
3/8 Michigan St. -3.2 v.Purdue
Marquette (9 games remaining)
2/10 Marquette +4.3 @Villanova
2/21 Marquette +3.7 @Georgetown
2/25 Marquette +3.1 v.UConn
3/1 Marquette +6.2 @Louisville
3/4 Marquette +8.9 @Pittsburgh
This is exactly what I mean when I say that Lunardi does a poor job in predicting the future. If today was Selection Sunday, his picks would likely be right, but if Marquette ran at expectation in these 5 games, it would come out with 1.4 wins. Let me say this again. The Marquette Golden Eagles will be running good by going 2-3 in this stretch. It doesn't take an expert bracketologist to realize that the they will drop out of their current position given that record.
Who then is in the best position to snatch up these coveted 2 seeds that these teams may vacate? We've already broached the outside possibility of UCLA getting up there, but who else?
Xavier, Gonzaga, Clemson, Wake Forest, Missouri(outside shot).
What do these teams have in common? They are all just behind the #2 seeds listed above and have easy schedules the rest of the way. Some even have the inside track on their conference or will be heavy favorites in their conference tournament, both of which the selection committee seems to value based upon looking at past years. MIGHT INSERT SOME DATA LATER TONIGHT. PLAYING FTOPS DONKAMENT GOGO.
Thursday, February 5, 2009
Bracket Odds - 38 Days Until Selection Sunday
We're just over a month away from Selection Sunday. There have been enough games played that the bulk of most team's resumes have been developed and a good or bad game doesn't change the efficiencies of a team that much.
Using the ESPN bracket(which has been accurate in past years in determining which teams dance) and efficiency data from KenPom(which is a good predictor of future success), Keith has been generating the chances each team wins each matchup, advances to each round, or wins it all.
These projections are good enough for gambling purposes once you factor in 1)a team being under/over rated by Pom, 2)the numbers assume that teams play who is listed in this bracket, so just because a team has a hard(easy) road here doesn't mean they will have one in 6 weeks, 3)semi-homecourt advantage that might happen in the dance(more on this later). It's also worth noting that the win %'s are not taken directly from Kenpom because the numbers on his site are largely flawed.
These are based on data from Monday. Teams have moved in terms of both expected seeding and efficiency due to extreme under/over performances in the last 4 days.
Comments
Favorites: % are based as of Monday. Listed Efficiencies are updated as of Thursday.
1. [1]Duke(23.1%) - 1st in Kenpom Eff.
Before this week's slaughter at Clemson, Duke was such a sizeable favorite to win it all that it didn't matter that Duke's region had numerous teams Kenpom had as underseeded teams(West Virginia as the #8, Illinois as the #5, Louisville as the #2).
Despite the loss and the big drop in efficiency that come with it, Kenpom still has Duke as the most efficient team in the nation. Likewise, the odds for them to win it all are expected to drop from Kansas 2008-like levels(also >20%), but I wouldn't be surprised if Duke was still the favorite when next week's projected bracket is revealed.
2. [1]North Carolina(15.2%) - 3rd in Kenpom Eff.
Kenpom has operated under numbers that indicate North Carolina is among the favorites to win it all, but not way ahead-of-the-pack type numbers that many preseason predictions currently had. To think that they are still overwhelming favorites to cut down the nets in late March as many current betting markets still have is delusional.
3. [1]UConn(10.2%) - 2nd in Kenpom Eff.
This bracket actually had UConn as the #1 overall seed. Thus, the double digit win at Louisville doesn't provide them any better potential seeding in the next bracket, though it does solidify their holding in case they lose a game or two against Pittsburgh, Syracuse, or Marquette the rest of the way. It also improved their efficiencies to the point where it's on par with Duke and North Carolina now.
4. [2]Pittsburgh(8.7%) - 4th in Kenpom Eff.
5. [2]Louisville(5.8%) - 8th in Kenpom Eff.
Louisville is an interesting case as it's the only team in Kenpom's top 15 to be so weak on one side of the floor(Offense), but making up for it as the #1 team on the other side(Defense).
Of the 15 most efficient teams in the nation it is the only team to not have both of it's components not be ranked in the top 50 as the team's adjusted offense clocks in at 94th. The Cardinals have problems getting to the line(.329 FTA per FGA, 253rd) and making free throws(65.9%, 248th) once they get there and are only mediocre in the other Four Factors.
As expected, their defensive is excellent, but not in the way you would expect if you are familiar with their inclination to successfully full-court trap. While they are good at forcing turnovers (they force turnover's on 23.3% of opponent's possessions, 45th), they are superb at getting opponent's to miss shots (43.1% DeFG, 10th) and grabbing those misses (71.3% DReb%, 26th).
It will be interesting to see whether or not this relatively 1-sided team can continue their success through the post-season. Conventional wisdom says that a team that relies on one side of the ball will fail once it runs into a team that is competent on both sides, but the Cardinals have notched wins against such teams already in Big East play:
1/10 61-60 Win @Villanova (34th Off Eff./15th Def Eff.)
1/17 69-63 Win v.Pittsburgh (2nd Off Eff./27th Def Eff.)
1/31 69-63 Win v.West Virginia (36th Off Eff./6th Def Eff.)
6. [3]Memphis(4.4%) - 7th in Kenpom Eff.
7. [5]Gonzaga(3.9%) - 5th in Kenpom Eff.
Memphis and Gonzaga are both teams that are agreed to be good by the public/top 25 polls/betting markets, but Kenpom has them even better as top 7 teams. Thus, despite mediocre seeding that is probably only to improve for 1 of them(the H2H game this Saturday is the only remaining game against significant opposition), they are among the favorites to win it all.
Both programs are quite storied, so how is it that they are still under the radar compared to the polls? I have a couple answers:
Both teams dropped early games in the NC schedule when they were in the public spotlight. Memphis lost all of it's major NC games, @Georgetown(21st) in OT, N-Xavier(20th), and v.Syracuse(28th), though has since reeled of 13 straight wins in dominating fashion(look at MoV) and won @Tennessee(40th).
Gonzaga lost 4 of 5 games including a major slipup at home against Portland St.(100th) despite 2 wins against Tennessee(neutral and Away), @Washington St.(54th) and N-Oklahoma St.(48th).
8. [2]Wake Forest(3.6%) - 19th in Kenpom Eff.
Demon Deacon odds will drop significantly next week after a 30+ pt underperfomance @Miami. At least Duke could rely on a stronger opponent and a larger efficiency gap it had created in previous games.
It's worth noting that only 2 of the 3 Chapel Hill teams(Duke, UNC, Wake Forest) can be placed at at NC Greensboro for the first 2 rounds, so these teams aren't only competing for seeding, but also for 2-3 extra pts of Homecourt Advantage for 2 games. Wake Forest's H2H wins against both teams was a pivotal tiebreaker in case resumes were similar at season's end, but consecutive losses to Georgia Tech and Miami FL are starting to put them a step behind those two.
9. [1]Oklahoma(2.8%) - 12th in Kenpom Eff.
It's true that Oklahoma's low odds areas a result of Kenpom not having them as among the most efficient teams in the nation, but it's also because their half of the region is stacked with efficient teams in this bracket. Arizona St. as the #8, Gonzaga as #5, Georgetown as #12.
The ESPN talking heads have been mentioning how Oklahoma has a good chance to realize a #1 seed because they don't face the gauntlet that is the ACC or Big East, but the Sooners are through the easy part of their schedule, and face tough tests in half of their 8 remaining games. (Predicted spreads come from a model using Kenpom data)
2/11 Oklahoma -1.9 @Baylor
2/21 Oklahoma +0.3 @Texas
2/23 Oklahoma +5.2 @Missouri
3/4 Oklahoma -4.3 v.Kansas
If they drop 2 of these games, which is far from unthinkable, do we really expect them to hold onto the top line?
10. [5]UCLA(2.6%) - 6th in Kenpom Eff.
UCLA is finally performing at efficiencies mirroring the team's preseason expectations as a top 10 team. The problem for UCLA is that it's unlikely to improve it's seeding by much the rest of the way. This seems strange as they have a stranglehold on a BCS conference and have higher odds on winning the conference tournament as its hosted in Los Angeles. However, the rest of the conference schedule is riddled with possible losses:
2/12 UCLA +1.1 @Arizona St.
2/14 UCLA -4.1 @Arizona
2/19 UCLA +6.2 v. Washington
2/28 UCLA -3.2 @California
The losses against Texas(25th) and Michigan(83rd) are looking progressively worse as both teams continue to plummet, hurting the Bruins' RPI in the process. Also, the non-conference game against Notre Dame this Saturday doesn't even look like a legitimate opportunity to shore up it's non-conference resume now that the Irish(54th) have fallen in Big East play. Outside of that, UCLA's best NC win is a 64-59 win at home against Miami OH(67th)
I will try to do updates on Fridays(lots of games being played on Wednesday and Thursday) and Tuesday(weekend games + new projections released). I will probably comment on underrated/seeded teams that are likely to cause upsets next week. Stay posted.
Using the ESPN bracket(which has been accurate in past years in determining which teams dance) and efficiency data from KenPom(which is a good predictor of future success), Keith has been generating the chances each team wins each matchup, advances to each round, or wins it all.
These projections are good enough for gambling purposes once you factor in 1)a team being under/over rated by Pom, 2)the numbers assume that teams play who is listed in this bracket, so just because a team has a hard(easy) road here doesn't mean they will have one in 6 weeks, 3)semi-homecourt advantage that might happen in the dance(more on this later). It's also worth noting that the win %'s are not taken directly from Kenpom because the numbers on his site are largely flawed.
These are based on data from Monday. Teams have moved in terms of both expected seeding and efficiency due to extreme under/over performances in the last 4 days.
Comments
Favorites: % are based as of Monday. Listed Efficiencies are updated as of Thursday.
1. [1]Duke(23.1%) - 1st in Kenpom Eff.
Before this week's slaughter at Clemson, Duke was such a sizeable favorite to win it all that it didn't matter that Duke's region had numerous teams Kenpom had as underseeded teams(West Virginia as the #8, Illinois as the #5, Louisville as the #2).
Despite the loss and the big drop in efficiency that come with it, Kenpom still has Duke as the most efficient team in the nation. Likewise, the odds for them to win it all are expected to drop from Kansas 2008-like levels(also >20%), but I wouldn't be surprised if Duke was still the favorite when next week's projected bracket is revealed.
2. [1]North Carolina(15.2%) - 3rd in Kenpom Eff.
Kenpom has operated under numbers that indicate North Carolina is among the favorites to win it all, but not way ahead-of-the-pack type numbers that many preseason predictions currently had. To think that they are still overwhelming favorites to cut down the nets in late March as many current betting markets still have is delusional.
3. [1]UConn(10.2%) - 2nd in Kenpom Eff.
This bracket actually had UConn as the #1 overall seed. Thus, the double digit win at Louisville doesn't provide them any better potential seeding in the next bracket, though it does solidify their holding in case they lose a game or two against Pittsburgh, Syracuse, or Marquette the rest of the way. It also improved their efficiencies to the point where it's on par with Duke and North Carolina now.
4. [2]Pittsburgh(8.7%) - 4th in Kenpom Eff.
5. [2]Louisville(5.8%) - 8th in Kenpom Eff.
Louisville is an interesting case as it's the only team in Kenpom's top 15 to be so weak on one side of the floor(Offense), but making up for it as the #1 team on the other side(Defense).
Of the 15 most efficient teams in the nation it is the only team to not have both of it's components not be ranked in the top 50 as the team's adjusted offense clocks in at 94th. The Cardinals have problems getting to the line(.329 FTA per FGA, 253rd) and making free throws(65.9%, 248th) once they get there and are only mediocre in the other Four Factors.
As expected, their defensive is excellent, but not in the way you would expect if you are familiar with their inclination to successfully full-court trap. While they are good at forcing turnovers (they force turnover's on 23.3% of opponent's possessions, 45th), they are superb at getting opponent's to miss shots (43.1% DeFG, 10th) and grabbing those misses (71.3% DReb%, 26th).
It will be interesting to see whether or not this relatively 1-sided team can continue their success through the post-season. Conventional wisdom says that a team that relies on one side of the ball will fail once it runs into a team that is competent on both sides, but the Cardinals have notched wins against such teams already in Big East play:
1/10 61-60 Win @Villanova (34th Off Eff./15th Def Eff.)
1/17 69-63 Win v.Pittsburgh (2nd Off Eff./27th Def Eff.)
1/31 69-63 Win v.West Virginia (36th Off Eff./6th Def Eff.)
6. [3]Memphis(4.4%) - 7th in Kenpom Eff.
7. [5]Gonzaga(3.9%) - 5th in Kenpom Eff.
Memphis and Gonzaga are both teams that are agreed to be good by the public/top 25 polls/betting markets, but Kenpom has them even better as top 7 teams. Thus, despite mediocre seeding that is probably only to improve for 1 of them(the H2H game this Saturday is the only remaining game against significant opposition), they are among the favorites to win it all.
Both programs are quite storied, so how is it that they are still under the radar compared to the polls? I have a couple answers:
Both teams dropped early games in the NC schedule when they were in the public spotlight. Memphis lost all of it's major NC games, @Georgetown(21st) in OT, N-Xavier(20th), and v.Syracuse(28th), though has since reeled of 13 straight wins in dominating fashion(look at MoV) and won @Tennessee(40th).
Gonzaga lost 4 of 5 games including a major slipup at home against Portland St.(100th) despite 2 wins against Tennessee(neutral and Away), @Washington St.(54th) and N-Oklahoma St.(48th).
8. [2]Wake Forest(3.6%) - 19th in Kenpom Eff.
Demon Deacon odds will drop significantly next week after a 30+ pt underperfomance @Miami. At least Duke could rely on a stronger opponent and a larger efficiency gap it had created in previous games.
It's worth noting that only 2 of the 3 Chapel Hill teams(Duke, UNC, Wake Forest) can be placed at at NC Greensboro for the first 2 rounds, so these teams aren't only competing for seeding, but also for 2-3 extra pts of Homecourt Advantage for 2 games. Wake Forest's H2H wins against both teams was a pivotal tiebreaker in case resumes were similar at season's end, but consecutive losses to Georgia Tech and Miami FL are starting to put them a step behind those two.
9. [1]Oklahoma(2.8%) - 12th in Kenpom Eff.
It's true that Oklahoma's low odds areas a result of Kenpom not having them as among the most efficient teams in the nation, but it's also because their half of the region is stacked with efficient teams in this bracket. Arizona St. as the #8, Gonzaga as #5, Georgetown as #12.
The ESPN talking heads have been mentioning how Oklahoma has a good chance to realize a #1 seed because they don't face the gauntlet that is the ACC or Big East, but the Sooners are through the easy part of their schedule, and face tough tests in half of their 8 remaining games. (Predicted spreads come from a model using Kenpom data)
2/11 Oklahoma -1.9 @Baylor
2/21 Oklahoma +0.3 @Texas
2/23 Oklahoma +5.2 @Missouri
3/4 Oklahoma -4.3 v.Kansas
If they drop 2 of these games, which is far from unthinkable, do we really expect them to hold onto the top line?
10. [5]UCLA(2.6%) - 6th in Kenpom Eff.
UCLA is finally performing at efficiencies mirroring the team's preseason expectations as a top 10 team. The problem for UCLA is that it's unlikely to improve it's seeding by much the rest of the way. This seems strange as they have a stranglehold on a BCS conference and have higher odds on winning the conference tournament as its hosted in Los Angeles. However, the rest of the conference schedule is riddled with possible losses:
2/12 UCLA +1.1 @Arizona St.
2/14 UCLA -4.1 @Arizona
2/19 UCLA +6.2 v. Washington
2/28 UCLA -3.2 @California
The losses against Texas(25th) and Michigan(83rd) are looking progressively worse as both teams continue to plummet, hurting the Bruins' RPI in the process. Also, the non-conference game against Notre Dame this Saturday doesn't even look like a legitimate opportunity to shore up it's non-conference resume now that the Irish(54th) have fallen in Big East play. Outside of that, UCLA's best NC win is a 64-59 win at home against Miami OH(67th)
I will try to do updates on Fridays(lots of games being played on Wednesday and Thursday) and Tuesday(weekend games + new projections released). I will probably comment on underrated/seeded teams that are likely to cause upsets next week. Stay posted.
Sunday, February 1, 2009
How To Make Money in Super Bowl 43
In the Super Bowl, it is oftentimes profitable to bet the favorite MoneyLine(to win) as a result of an market inefficiency caused by a particularity in human psychology.
People don’t like to risk a lot of money to win a little money. The flip side is also true. People like to risk a little money to win a lot of money.
I’ve read quite a few studies/theses on this in fields outside of gambling, but a typical donk’s obsession with betting parlays and longshot futures should convince you of this. This usually manifests itself in the portion of the public that wants to bet the underdog(and there’s a lot of them) placing wagers on the ML and moving it artificially lower to a point where the favorite ML can be +EV.
How is it possible that this happens with the Super Bowl – the most efficient and transparent market in the sporting world – you ask? How can something be this well-known and open to millions of bettors be available? I have a few theories, though this is the strongest:
The inefficiency is localized: This phenomenon mainly exists on sports books that have a larger proportion of square/public/recreational clientele. This seems evident if we compare Vegas MLs with Online MLs with the assumption that online markets are sharper than Vegas ones just as online poker is.
Online:
http://www.sbrlines.com/LinesWebsite/Lines.aspx?l=2#/odds/nfl/whole_game/moneyline.html
Vegas:
http://www.vegasinsider.com/nfl/odds/las-vegas/money/
After gazing at these lines, one may wonder, what would be a fair ML? During the regular season a -6.5 -106/+6.5 -102 spread with ~47 total(current Pinny lines) would equate to a +/- 255 ML. It is possible that the game being played on a neutral field decreases the distribution of scores in the Pittsburgh -0.5 to Pittsburgh -6.5 range somewhat, but it is unlikely that it could be large enough to move it to more than a handful of cents.
LOL @ knowing having to know anything about sports – Steelers -200/-210/-220 - BIG.
(Also find yourself some a morsel of Cardinals +7 -115/Cardinals +230 for the middle/arbitrage)
People don’t like to risk a lot of money to win a little money. The flip side is also true. People like to risk a little money to win a lot of money.
I’ve read quite a few studies/theses on this in fields outside of gambling, but a typical donk’s obsession with betting parlays and longshot futures should convince you of this. This usually manifests itself in the portion of the public that wants to bet the underdog(and there’s a lot of them) placing wagers on the ML and moving it artificially lower to a point where the favorite ML can be +EV.
How is it possible that this happens with the Super Bowl – the most efficient and transparent market in the sporting world – you ask? How can something be this well-known and open to millions of bettors be available? I have a few theories, though this is the strongest:
The inefficiency is localized: This phenomenon mainly exists on sports books that have a larger proportion of square/public/recreational clientele. This seems evident if we compare Vegas MLs with Online MLs with the assumption that online markets are sharper than Vegas ones just as online poker is.
Online:
http://www.sbrlines.com/LinesWebsite/Lines.aspx?l=2#/odds/nfl/whole_game/moneyline.html
Vegas:
http://www.vegasinsider.com/nfl/odds/las-vegas/money/
After gazing at these lines, one may wonder, what would be a fair ML? During the regular season a -6.5 -106/+6.5 -102 spread with ~47 total(current Pinny lines) would equate to a +/- 255 ML. It is possible that the game being played on a neutral field decreases the distribution of scores in the Pittsburgh -0.5 to Pittsburgh -6.5 range somewhat, but it is unlikely that it could be large enough to move it to more than a handful of cents.
LOL @ knowing having to know anything about sports – Steelers -200/-210/-220 - BIG.
(Also find yourself some a morsel of Cardinals +7 -115/Cardinals +230 for the middle/arbitrage)
Sunday, January 25, 2009
A New Direction - College Basketball Analysis
The "I GET NO RESPECT", LOL SCHOOL and brag posts are getting kind of boring, so I'm going to do something constructive here. Namely give you enough information about college basketball to embarrass your all-talk, but don't know shit friends.
Here's what you can expect to come in the coming months:
- We're halfway through the college basketball regular season, so I will start commenting on the projected field for the Big Dance. To be fair, the major bracketologists out there are competent at evaluating resumes. That is to say, if you give Joe Lunardi the result of every game for the regular season, he can predict the 35 at large teams(and their seeding) very accurately. However, where he fails is his inability to predict the future - specifically, games remaining in the season(that will have a monstrous affect on who gets in/seeding) and odds of winning games once they get in. This is where sports betting markets come in.
Some tentative ideas I want to discuss are - How this year's cross section of teams will lead to easy/hard roads to the F4/Championship, potential upsets/mis-seedins/teams left out, partial home court advantages[some obvious(MSU as a top 4 seed), some not(Missouri as possibly an 11 seed!)].
- When Selection Sunday comes out, I will explain how to use the betting markets and basic game theory to get as much equity as you can in your March Madness Pool. In other words, How To Win Your Bracket Pool.
- For those that like money - Keith and I will be releasing plays together for March Madness. We may start as early as conference tournament time. These plays are generated via a Excel-based model using pace/efficiency data from KenPom, and then adjusted for injuries, matchups, *shudders* trends and intangibles.
If you want to be more literate you should read some of the links to the right regularly.
As an introduction to Basketball Sabermetrics you should start with Dean Oliver's Four Factors to understand what really wins basketball games and an introduction to advanced stats.
If you want to get a grip on the data I use, read Kenpom. If you're trying to determine which team has been playing better so far this year, check Kenpom. If you're trying to determine who's the more efficient player, James Harden or Stephen Curry, you should check Kenpom.
If you just like reading fuzzy stuff cause you've been weaned on ESPN garbage your whole life, you can just check Basketball Prospectus regularly - at least their content is mathematically grounded.
Here's what you can expect to come in the coming months:
- We're halfway through the college basketball regular season, so I will start commenting on the projected field for the Big Dance. To be fair, the major bracketologists out there are competent at evaluating resumes. That is to say, if you give Joe Lunardi the result of every game for the regular season, he can predict the 35 at large teams(and their seeding) very accurately. However, where he fails is his inability to predict the future - specifically, games remaining in the season(that will have a monstrous affect on who gets in/seeding) and odds of winning games once they get in. This is where sports betting markets come in.
Some tentative ideas I want to discuss are - How this year's cross section of teams will lead to easy/hard roads to the F4/Championship, potential upsets/mis-seedins/teams left out, partial home court advantages[some obvious(MSU as a top 4 seed), some not(Missouri as possibly an 11 seed!)].
- When Selection Sunday comes out, I will explain how to use the betting markets and basic game theory to get as much equity as you can in your March Madness Pool. In other words, How To Win Your Bracket Pool.
- For those that like money - Keith and I will be releasing plays together for March Madness. We may start as early as conference tournament time. These plays are generated via a Excel-based model using pace/efficiency data from KenPom, and then adjusted for injuries, matchups, *shudders* trends and intangibles.
If you want to be more literate you should read some of the links to the right regularly.
As an introduction to Basketball Sabermetrics you should start with Dean Oliver's Four Factors to understand what really wins basketball games and an introduction to advanced stats.
If you want to get a grip on the data I use, read Kenpom. If you're trying to determine which team has been playing better so far this year, check Kenpom. If you're trying to determine who's the more efficient player, James Harden or Stephen Curry, you should check Kenpom.
If you just like reading fuzzy stuff cause you've been weaned on ESPN garbage your whole life, you can just check Basketball Prospectus regularly - at least their content is mathematically grounded.
Thursday, January 22, 2009
Belated NFL Futrues Grading
WINNARZ
2.5u Dallas Cowboys to MISS the playoffs +180
1.3u Dallas Cowboys under 10.5 Wins +125
2.2u Balitmore Ravens over 6 Wins -115
2u Detroit Lions under 6.5 Wins +125
1.8u St. Louis Rams under 6.5 Wins +115
1.3u Tampa Bay Bucs over 8 Wins -115
1.3u Oakland Raiders under 6 Wins -115
1.3u Houston Texans over 7.5 Wins -105
1.3u Cleveland Browns under 8 Wins +100
1.1u Minnesota vikings over 8.5 Wins -135
1u Carolina Panthers over 8 Wins -115
FAIL
2.5u Green Bay Packers over 8.5 Wins +125
2.5u Seattle Seahawks to win NFC West -110
1.3u Pittsburgh Steelers under 9 Wins +110
1.3u Atlanta Falcons under4.5 Wins +120
1.3u San Francisco 49ers under 6 Wins +145
1u Baltimore Ravens to win AFC North +850
0.8u Philadelphia Eagles to win NFC East +450
.5u Baltimore Ravens to win AFC North +1350
SISTER KISSER
2.2u Philadelphia Eagles over 9 Wins -115
10-4-1 on season o/u, 1-3 on futures
+10.03u on 32.71u wagered. 30.7% ROI.
2.5u Dallas Cowboys to MISS the playoffs +180
1.3u Dallas Cowboys under 10.5 Wins +125
2.2u Balitmore Ravens over 6 Wins -115
2u Detroit Lions under 6.5 Wins +125
1.8u St. Louis Rams under 6.5 Wins +115
1.3u Tampa Bay Bucs over 8 Wins -115
1.3u Oakland Raiders under 6 Wins -115
1.3u Houston Texans over 7.5 Wins -105
1.3u Cleveland Browns under 8 Wins +100
1.1u Minnesota vikings over 8.5 Wins -135
1u Carolina Panthers over 8 Wins -115
FAIL
2.5u Green Bay Packers over 8.5 Wins +125
2.5u Seattle Seahawks to win NFC West -110
1.3u Pittsburgh Steelers under 9 Wins +110
1.3u Atlanta Falcons under4.5 Wins +120
1.3u San Francisco 49ers under 6 Wins +145
1u Baltimore Ravens to win AFC North +850
0.8u Philadelphia Eagles to win NFC East +450
.5u Baltimore Ravens to win AFC North +1350
SISTER KISSER
2.2u Philadelphia Eagles over 9 Wins -115
10-4-1 on season o/u, 1-3 on futures
+10.03u on 32.71u wagered. 30.7% ROI.
Subscribe to:
Posts (Atom)