Friday, February 29, 2008

Worst. News. Ever.

Patrick Patterson breaks his ankle and is out for the season. Our bright hopes of making it into the tourny dim quite a bit. Note: UK's RPI is 56 today, and can only improve with 2 road games.

How long has he been playing on the stress fracture? He's been hobbling every game since before he sat the Houston game with a bum ankle. Warrior.

Wednesday, February 27, 2008

Basketball Wednesday

I'm experimenting with theme days for the blog, like other bloggers have. This week is Judgment Week on ESPN, where we start condemning teams to a post-season outside the NCAA tournament. For #1 Tennessee, Judgment Week meant they had to play in Nashville, which is like playing in hell, and so they were judged "unworthy" of being #1 any longer.

In honor of Judgment Week, I thought I'd post Kentucky's RPI progression from the last time I posted it, as calculated by Ken Pomeroy. I keep a spreadsheet updated with all the Pomeroy team stats, ratings, RPI and Sagarin ratings that I update the morning of and after every UK game. I'm the only person I know that does this.

The RPI wouldn't be so bad if they'd change their weights. You can be a team with very few road wins and still have a decent RPI simply because you played those road games.
Kentucky was blown out at Vandy, but saw their RPI increase by 4 points by virtue of having played a good team on the road. It jumped 12 points by beating a terrible LSU team on the road.

Click to enlarge. Going into the Ole Miss game, Kentucky's RPI is 65, and Strength of Schedule is 12. Given that Ole Miss has a good overall winning %, the win should bump us up about as much as the Arkansas win did.

RPI matters to the Tournament Selection Committee. The DanceCard website is run by some statisticians to predict who the committee will select, and is 93.3% accurate. They use the old RPI formula in their calculations (why?), with which Kentucky is rated #46.

DanceCard is updated weekly. On February 3rd, DanceCard rated Kentucky the 83rd most tournament-worthy team, giving them slim chances of making the tourny. The improvement has averaged about 10 spots a week, and Kentucky is now considered the 43rd most-worthy team. Removing the teams that will receive automatic bids, Kentucky is now considered IN the tourny by DanceCard.

Monday, February 25, 2008

On Casinos and Gambling in Kentucky

The Kentucky legislature potentially authorizing construction of at least 12 casinos in Kentucky has become the hottest issue in the state. Kentuckians rejected the incumbent governor whose major platform was to keep casinos out, and elected one who pledged to bring them in.

I can sum up economic analysis of casinos in two sentences:

1. The cost of casinos outweigh the potential benefits.
2. Casinos in Kentucky (and indeed all other states) are inevitable and unavoidable.

1. Probably the most heavily-quoted economist on the subject is Baylor professor Earl Grinols, who published his findings in a book while he was at the University of Illinois. Grinols found that the costs of casinos (increased addiction, higher crime and costs associated with it, like more police) eventually outweigh the monetary benefits in the long run.

2. The casino decision is a classic example of a competition game, and the outcome is a Nash equilibrium.

When Illinois passed legislation allowing casinos, Dr. Grinols testified before the state legislature that all their neighboring states (including Kentucky) would soon do the same. Why?

Observe the figurative matrix:

At the original equilibrium, both Kentucky and Illinois had a payoff of 100 representing that both states received tax revenue from their own constituents. This equilibrium is not sustainable, however, because each player has an incentive to cheat. Illinois, by legalizing casinos while Kentucky does not stands to gain tax revenue by Kentuckians going over the border to gamble. They benefit by essentially letting Kentuckians fund some of Illinois' budget. This is represented as a payoff of 150 to Illinois, and a payoff of 40 to Kentucky.

Kentucky loses by losing this tax revenue. Kentucky, knowing that Illinois has an incentive to cheat and recognizing that if it doesn't build casinos it will be at 40, also chooses to legalize casinos. Now, both states have built casinos and their payoffs are at 50,50.

Now, both states are worse off because they both have casinos but each state is better off than it would be if they had done nothing while the other state built casinos. The casino-free 100,100 is the best payoff but it's not sustainable because each state has an incentive to legalize gambling while the other does not. Got it?

This has played out in real-life out as almost every neighbor of Illinois has now legalized casino gambling. Dr. Grinols wasn't a prophet, he just understood the basic Nash solution that the situation would result in a state where neither Illinois or Kentucky would be better off making a unilateral decision.

You've heard the argument from Kentucky politicians: they don't like losing potential tax revenue to Illinois and Indiana. So, Kentucky will pass casinos and begin to take revenue from Tennessee and West Virginia, who will do the same thing, and so on until we have legalized gambling in all 50 states.

What's the best possible solution? For all states to eliminate their casinos, so that we remain at the 100,100 payoff equilibrium. But, that equilibrium is not sustainable unless the Federal government steps in and mandates it to be so.

Thursday, February 21, 2008

The Real Danger

This is what I think about when I wake up in the morning.

A few weeks ago, an anonymous member of the White House staff wrote Greg Mankiw to defend the President on Democratic accusations of how much he has (or has not) increased the federal debt. He succeeded in making his point, and then gives us this graph to show the long-term fiscal danger. Economists everywhere are making noise about this danger, but so far few people appear to be listening. This graph represents federal debt held by the public as a percentage of GDP. Every subsequent federal deficit adds to the total:
Here's the analysis:

On our current long-term policy path, the federal debt would explode beginning in about 20 years. Note that while this chart purports to go out to 2080, we would never make it that far on the light blue line on our current long-term policy path. Financial market pressures would force a change long before our federal debt got to 200 or 250% of GDP.

The real threat is not the additional 2.8% of GDP of debt we have accumulated since the President took office, during a time of recession, war, terrorist attacks, high oil prices, and burst stock market bubbles.

The real threat is that steady and steep upslope in the blue line - the explosion of future debt over the next several decades, if we don't do something about it.

So, what will happen by 2027 if no one steps up to take steps to prevent it now? America and her dollars would become a pretty bad investment. A dollar collapse would likely accompany economic collapse.
Why the explosion of debt? Partly due to deficit spending due to the dramatic increase in Social Security and Medicare outlays as America ages. And military spending has something to do with it as well. Some of my former professors at Baylor believe there will eventually be pressure on the elderly to die. Maybe mass movements to euthanize any sick old person.

Hear any candidates talking about this? Fred Thompson talked about cutting entitlements, but was blasted for it (and also tried to say he wasn't for cutting entitlements). No one wants to be honest about it. They want to promise things for people and deliver on those promises. They don't want to look at the bill. I vote for McCain because I think he's the only one who could look Americans in the eye and say "this is a real problem."

I don't want my son to be burdened with this debt and to see such a collapse. Think about all this when you spend your federal stimulus money ("Bush Money" as a Wal-Mart employee tells me his customers call it). Remember that China will want its money back.

Wednesday, February 20, 2008

Book Review (5th of 2008)

Sticking with the political theme.

The Choice by Bob Woodward (abridged audio).

This book basically picked up where the last book I read left off. It is a candid look at the months leading up to the '96 election campaign from the Dole and Clinton camps. How they reached their decisions about running, and how they shaped their campaigns.

I now think I've now read most of Woodward's books from the 90's, and most of the major books on the Clintons. This book was short and didn't have much new information in it. I think every Woodward book has one Clinton line in it where he lays out the F-bomb, and this is no different.

I give it 2 stars out of 5. Bob Dole really is a nice guy, his wife is a nice lady, and this book shows it well. A quick, easy read.

Tuesday, February 19, 2008

Book Review (4th of 2008)

We visited some outlets at Pidgeon Forge a few weeks ago and found a bookstore selling a ton of audio books for $1 each. I bought a few.

Blood Sport: The President and his Adversaries by James B. Stewart.

Stewart was a reporter brought on by the Clinton White House to independently investigate Whitewater. The story starts with the mysterious suicide of White House counsel and rumoured Hillary lover Vince Foster.
Stewart goes step-by-step through the Clintons' political career, the constant womanizing of Bill Clinton, the failed land deal of Whitewater, the friends Hillary made involved in insider commodities trading (not illegal), and the first run for the White House. The book ends about the time the Paula Jones scandal was gaining ground.

It's an interesting look at the advisors to the Clintons, their recklessness, their paranoia about the media, and just how sometimes they ended up at the wrong place at the wrong time. I found it useful to hear Whitewater explained completely, including all of its weirdness and shadiness.

I give the book 3 stars of 5. The reader did an amusing job imitating Clinton's and his Arkansas' cronies' accents. I would add it to the list of books to read about Hillary, if anyone is considering voting for her.

Monday, February 18, 2008

Book Review (3rd of 2008)

We took a trip to Cincinnati last week and spent a couple days working on my in-law's new house. There's much to blog about, but I think for the next several posts I'll catch up on book reviews.
On Becoming Baby Wise: Giving Your Baby the Gift of Nighttime Sleep. By Dr. Gary Ezzo and Dr. Robert Bucknam.
This book is also available in a more Christian format, complete with videos. Check out Growing Kids God's Way by Gary Ezzo.

Several people recommended this book to us. Those who recommended it have had 9-month old babies who sleep 12 hours a night and are happy kids to be around. They all gave credit to (or swore by) this book's methods.
We've been around other babies who don't sleep through the night, and the parents are exhausted and miserable. I decided that I wanted to avoid that situation at all costs, thus I read this book.

Parenting methods are controversial, and spark arguments. The book gives some history of different methods, and how they settled on theirs (establishing a parent-controlled routine for the baby). The studies related that show sleep's correlation with IQ, ADD, and other conditions just drove home the point for me. I agree with the book, there are no "lucky" parents with "easy" babies, it's dependent on parents establishing certain routines early.

They grip you early with data:
Of 520 infants on the parent-directed feeding (PDF) method, at least 76.8% were sleeping through the night between 7 and 9 weeks. At least 95.7% were sleeping through the night at 12 weeks. (The book breaks it down by boys/girls, and breastfeeding/formula feeding, birth defects, other problems, etc).

They err by not giving data on parents who do not use the method. But, if something performs 97% of the time, I'm highly likely to try it.

80% of the babies on the PDF routine required no help in sleeping through the night, they did it on their own. The remaining 20% cried for 3-5 nights and then slept through.

It might also cost 3-5 nights of sleep while your child learns when to sleep, but it pays off in years of full nights sleep.

I give the book 3.5 stars out of 5. It's quite repetitive (so you can remember its methods) and if you don't have a kid, then it's not worth reading. But, if you're expecting, I consider it a MUST read. If you know someone with a kid who is a handful, you might pick them up a copy.

Saturday, February 09, 2008

Hiatus

I'm taking a 7 day fast from blogging starting tomorrow. It's not that I've run out of things to blog about, just that sometimes blogging keeps you from other things that are more important to think about.

Until then, a few links of note:
HT: Rynoman. The NCAA hosted a 2-day mock Selection Sunday with sportswriters and TV commentators, and liveblogged it. A fascinating look at how the brackets are put together every March. Also, somewhat disappointing.

Many bloggers I read are quite skeptical of Rick Warren.
Matt McKee sent me a blog page of a techie who was posting from the Davos World Economic Forum. The guy met and briefly interviewed Rick Warren, and found him quite likable.
I encourage you to watch this video with Rick Warren.
I liked how he believes technology is great, and also how he talks about his church's thousands of small groups that all get to travel to do missions.

Taylor Wehrle has found a cool sermon series from Sovereign Grace Ministries. It's 4 sermons on how we can glorify God through sleeping, working, eating, and playing. I'm definitely interested in the sleeping one. Maybe I'll listen to these instead of blogging.

This is what I'm looking forward to watching tonight. I'm eager to see and listen to the radio transmissions of 3 of the 4 guys on this page for free, as advertised. Joni and I are both very eager for the Daytona 500 a week from tomorrow.

Friday, February 08, 2008

Probability and Sports (Part 3)

Remember George Mason? Who could forget? They are the prime example that randomness happens.

In 2006, George Mason tied LSU (1986) as the lowest seed (#11) to make the Final 4. At the start of the tournament, the odds of them winning a championship were 600 to 1. WhatIfSports.com does a Bracket Preview where they simulate the tournament 10,000 times.

I wrote in March of that year that in WhatIf.com's 10,000 simulations, George Mason made the Final 4 only 80 times. Even after they had won 2 games in the tourny, the simulators gave them a 3.8% chance of making the Final 4.

Their Pythagorean Winning %, as reported by Pomeroy, gave them a 1.52% chance of making it to the Final 4. The had less than a 50% chance of even winning their first game.

GM's success had many coaches and ADs calling for an expansion of the bracket to 128 teams, arguing that "obviously, small schools are becoming just as good as the big schools."

But, it's not obvious. Rare events happen. If 80 of 10,000 simulations had George Mason making it at least to the Final 4, then there's a chance. Now, if this happens so often that the long odds no longer make sense, then you can make your argument for 128 teams. But, until then, I will revel in the memory of the bracket-busting randomness that George Mason represents.

Thursday, February 07, 2008

Probability and Sports (Part 2)

When Team A beats Team B, Team A is considered to be the “better” team. But, is one game ever a good measure?

The Patriots beat the Giants in NY by 3. Then, they lost to the Giants on a neutral field by 3. Who is the better team? Shouldn’t these performances even out? What if they played 10 times?

I will make my point about 1 game not being enough by using an NFL simulator. Observe:

I’ve been playing Tecmo Super Bowl III on SNES since 1995. I have an SNES emulator on my PC, which means I can speed up the games and simulate seasons 1,000 times faster than the SNES could. TSB III lets you play with legendary players (Montana, LT, Sayers, Butkus, etc.) as well as make trades in the off-season. The computer is pretty good at randomizing play calling, knows when to go for 2, etc. Some teams are rigged to pass more or play different defenses by default, which allows me to test and see which system is most effective.

I like to build teams and see how they perform against each other, letting the computer play the games play-by-play as I watch. It’s an excellent simulator.

I recently built 2 great and evenly-matched teams with Kansas City and Carolina, loaded with the top players in the game at every position.

KC has the top offense, defense, top-rated QB and sacks leader. Carolina has the top RB, top 2 interception leaders, and a slight edge in special teams. Carolina plays a 3-4 defense by default, whereas KC plays a 4-3.

Both teams are now in the Super Bowl, having gone 18-0. I highly anticipate knowing who will win, but I think that KC’s offense will just be too much…

Final score: Carolina 34, Kansas City 7.

A total blowout. Carolina appears to be the dominant team.

If ESPN covered this game they’d say something like “There’s now no question who is better…” And ESPN would be wrong. One game doesn’t answer any questions.

The beauty of SNES emulation is that I can repeat the game, as many times as I want, as if it was never played before. 35 observations is a better (more scientific) sample size than 1 observation.

The result of 35 games:

Kansas City wins 65.8% of the games! The Chiefs average 35.2 points per game, to Carolina’s 27.1, an average spread of KC by 8.06 points per game.

Carolina’s highest score in a game was 45. KC matched or bested that 11 times, including a 75-point performance! We can now safely name Kansas City the dominant team.

The first game’s spread was 2 standard deviations above the mean, a rare event. If you simply took that first game at face value, you'd reach the wrong conclusion about which team is better. Randomness happens. The NBA eliminates some of this randomness by playing 7-game series. But, in the NFL and NCAA you don’t get that luxury. So, it’s impossible to take 1 game and declare that a team is clearly better.

(Note: In SNES simulation, teams don’t learn from each other. In real life, if two teams played 30 times, they would learn from their actions and really not be the same teams at game 30 as they were at game 1. This makes it harder to measure their attributes).

So, next time you watch a game, maybe even an “upset,” ask yourself “If this game were to be repeated 30 times who'd be the winner, on average? What can I really learn from 1 game?” It will change how you watch sports.

Wednesday, February 06, 2008

Probability and Sports (Part 1)

I’ve been a lot more aware of probability lately, mostly due to a book I recently read which I’ll talk about some other time. I’ve also noticed some bad uses of statistics and an ignorance of randomness.

I enjoy the show SportsScience on FSN, where they apply physical sciences to the sports. They have a really cool laboratory for recording physical activity, and they get all-pro athletes on to help test hypotheses. For example, they test which athletes hit harder, how fast Chad Johnson stops (on a dime), if Michael Jordan really “hangs” in midair (no), if a basketball player or a football player jumps further, how much further a warm golf ball goes than a cold one, etc.But, sometimes they get it wrong.

In one episode, they wanted to test which stressors affect free throw shooters the most. They brought in an 80% FT shooter from some college.

They had him shoot 10 free throws at a time. The first round, with no distractions, he shot 9 of 10, or 90%. For the second, they added a crowd of people screaming and clapping noisemakers, some clowns on the court (including one scantily clad female who stayed right in his ear), and some flashing strobe lights. The guy hit 7 of 10, or 70%. Thus, they determined the distractions lowered his shooting by 22.2%.

This, however, is a false conclusion. The average of 90% and 70% (the shooter's 2 scores) is 80%, which is what the guy shoots in real-life (based on a large data set of hundreds of free throws in various situations). You can’t say the distractions had anything to do with his 7 of 10, as this was probably just regression to the mean. See, if you have an 80% free throw shooter shoot sets of 10 free throws, he’s going to hit a range of 6 of 10, 8 of 10, and 10 of 10 sometimes. Scientifically, you can’t take one of those sets and determine anything meaningful about it compared to the others.

Now, if you had him shoot 500 free throws without a distraction and 500 with distractions and compared the results, controlling for all other factors, you might be able to say something.

Next, they had the guy shoot with all the previous distractions, plus a heavy metal band playing louder decibals than the loudest NBA arena. He shot 6 of 10. They concluded that sensitivity to sound must be greater than sight, thus he shot more poorly. Again, a faulty conclusion.

Thankfully, they did make one good insight: They monitored the player's brain activity and heart rate during every exercise. When the heavy metal band started, the player’s heart rate jumped considerably. This might be useful information (if more tests are run on more players).

The scientists say that if you most want to distract a player, make a loud and unexpected noise during a free throw. A bull horn, or a loud clap or something. This should make his heart rate spike and increase his chances of missing (not sure what they were basing this on). So, I hope any Kentucky students reading this will keep that in mind for when Florida comes to Rupp.

But, beware misuse of statistics by people on television.

Improvement

I thought people might enjoy a graphical look at how Kentucky is improving, and what a difference a road win makes (click the pic to enlarge). Here is the RPI since the Vandy game. Notice the 32 place improvement after Tennessee, and the 26 place improvement after Georgia, our first road win.



Here's a look at the Pomeroy ratings and Sagarin's Pure Points, as a better measure of how good the team is. The trend is improvement overall. The Pure Points is a good measure of how many points we can score vs. how many an opponent can. (Excel wouldn't line the points up right, but the first ones you see are the morning of the Vandy game, the next you see are the day after, and so on).

Tonight will hopefully bring another needed road win.

Monday, February 04, 2008

Another must-read

Yesterday was Greg Mankiw's birthday, and I sent him congrats via Facebook. He also published an Op-Ed in the NY Times that everyone should read.

Eventually, economists will bang the drum on the debt issue so loudly that people will eventually give the politicians the political will to change things... I hope. Again, John McCain is the only person who I can envision coming on television and saying it straight: We either cut benefits, raise taxes, or both.

Republican candidates are fond of saying we should cut tax rates because doing so would incentivize more rapid economic growth (true) and raise tax revenue (wishful thinking). But unless we figure out a politically acceptable way to reduce the benefits now promised to future retirees, taxes are going up in the coming decades. The national debate will have to shift from which tax cuts do the most good to which tax increases do the least harm.

Democratic candidates like to talk about expanding the social safety net with universal health insurance. But they blithely ignore the fact that the safety net we already have was bought on credit and that the bill is almost due.

Sunday, February 03, 2008

Your Super Bowl Pick (2008)

While Brady will be holding up the trophy at the end of the game, the smart money is on the Giants!



The line opened at NE by 14, then moved to NE by 12 over the past 2 weeks (as betting came in top-heavy on the Giants).

What do the predictors say? Prediction Tracker updated this morning, but is still missing a prediction from the top predictor during the second half of season, Cover81.

The average of all reporting predictors says New England by 9.90, with a standard deviation of 3.43. That gives a range of 6.47 to 13.33. So, 12 is in the upper range of this prediction.

Judging by the average of all of predictors, there is a 19.4% chance that the Giants won't beat the spread, or an 80.6% chance of the Giants beating the spread.

The season's overall top 4 predictors (excluding Cover 81) predict, on average NE by 10.44. The upper bound factoring in standard deviation is 12.22!

The top predictors for the 2nd half of the season (perhaps a better measure, since data at mid-season is stronger than in the first few weeks) average out at NE by a mere 7.91. The highest prediction is NE by 9.1 , 3 points lower than the spread! Missing Cover81's data hurts, and I sent him an email...maybe he'll send me his number.

So, the consensus among the predictors is clearly that the Giants will beat the spread. Those who have been more accurate lately say it will be a close game!

Note: I don't condone gambling, but this is as clear-cut a pick as I've seen in a long time.

Saturday, February 02, 2008

A Briefing (I guess?)

Some bloggers I've seen call a post like this a "briefing." I guess that's what this is.

"Super Delegates"= Why I still think Hillary will win the Democratic nomination. Their people are too entrenched and experienced in the convention circles.
"If Super Tuesday results in a stalemate between Obama and Clinton, the key to the nomination will likely be the "super-delegates"--796 Democratic party leaders, who make up 20 percent of the delegates at the Democratic convention" including Bill Clinton and friends of Hillary.

Clinton's grand plans don't pay for themselves (as she says they do).

People forget that the Bill Clinton administration was seen as "centrist" when elected, lurched to the LEFT quickly after election, then when the backlash led Americans to elect a Republican Congress and Senate, Clinton moved back to the center.

Hillary is running on a very populist platform, even more than her husband did. But, even some outspoken liberals think she'd make a bad President.

However, Ann Coulter says she'd campaign for Clinton rather than vote for John McCain. If Rush Limbaugh and Coulter represent the mainstream of Republican thought, then I should probably change parties.

That's probably why (former Fed chairman) Paul Volcker is endorsing Obama.
"It is not the current turmoil in markets or the economic uncertainties that have impelled my decision. Rather, it is the breadth and depth of challenges that face our nation at home and abroad. Those challenges demand a new leadership and a fresh approach.”

HT to a former Iraq blogger for finding this website that lets you see and play with the Federal Budget.
Notice how big the Department of Treasury's contribution to the deficit is. Notice that 89.5% of that is payment on debts owed, mostly to Japan, China, and Europe who lend us money while we run these huge deficits.

Remember that number when you get your tax rebate. That money was borrowed from abroad and contributes to the increasing amount of debt that will get passed onto the next generation.
Eventually, someone will have to pay. When you scroll up and look at the amount of outlays that go to Social Security and Medicare recipients, and know that number is about to increase dramatically... kind of makes you wonder what the future holds.

That's why I'm voting for John McCain. He's the only candidate who has proven he has the integrity to not spend money. Before a Ron Paul person comments and says "Ron Paul '08!" note Paul's hypocrisy as a big earmark writer himself, part of the problem and not the solution.