Monday, March 18, 2019

2019 NCAAB Tournament - March Madness Thoughts and Power Ratings



As I begin conducting my diligence for this year's NCAA Tournament, I want to provide some initial thoughts and document my annual process of comparing power ratings to seeding. 

The chart illustrated below contains aggregate ratings from three of the most respected college basketball math models/rating systems (KenPom, Jeff Sagarin, ESPN's BPI).

It represents a quick and dirty method to identify overvalued and undervalued teams, which can be a very rewarding exercise since most people are inevitably influenced by seeding when filling out a bracket.


Underseeded: Auburn (5), Wisconsin (5), Iowa St. (6), Louisville (7)

Overseeded: Houston (3), Kansas (4), LSU (3), Kansas St. (4)

While my opinions may evolve and potentially become more informed over the next day or two as my research continues, I prefer to get my initial thoughts down on paper as soon as possible to prevent being influenced by external thoughts and opinions. 
I generally view the tournament through an Elite 8 lens, but made an effort to broaden my perspective for this exercise. Forgive me for the scatterbrained, bullet point format!

East Region
·         Michigan St. has a favorable draw to get to the E8
o    LSU (3) and Maryland (6) are both overseeded and teams I am bearish on. With that said, I'm not sold on Winston being fully healthy and am definitely hesitant to pencil them in without giving it further thought.

·         Difficult to leave Duke out of E8 or F4
o    VA Tech is a solid team and the return of their PG is intriguing, but would be catching 6 or 7 from Duke in potential S16 matchup.

·         LSU is tough to judge since they are without their coach. It probably doesn't have as much of an impact as perception may suggest but still creates plenty of uncertainty.
o    Need to take a closer look but Yale seems to be a frisky 14 seed. Early line of LSU -8 is definitely low for a 3/14 game, but realistically it's SEC athletes vs. Ivy league.

·         Honestly don't know much about Miss St. and never saw them play
o    The math models view them favorably, but to expect them to get past the S16 (by beating Duke) is probably wishful thinking

West Region
·         Gonzaga has a pretty favorable draw, but a rematch with FL State potentially looms in the S16
o    Gonzaga would still be favored by about 5 in that game, but the Seminoles should not be taken lightly. Need to give this more thought.

·         FL State should get to the S16. Marquette is vastly overrated and probably the weakest #5 out of the bunch. 

·         Bottom half is tough to forecast and should be very entertaining from a fan perspective
o    Michigan (2) and Texas Tech (3) are two of the upper-echelon teams I am a bit skeptical of
o    Nevada (7) is the highly decorated mid-major up against a toughFL (10) team in round 1. Both teams are underseeded according to the math models and it's unfortunate they have to square off in Round 1.
o    Elite 8 decision - probably going to have to split this half of the region up in the combos. No strong convictions at this point.

South Region
·         Similar to the West Region, the bottom half is tough to forecast and I can see myself mixing up the E8 combos
o    Cincy (7) is underseeded but I wasn't very high on them coming in
o    Nova (6) is overrated thanks to the Big East not being a good conference. I wouldn't be surprised to see them get knocked off by St. Mary's in the first round
o    Purdue (3) is hard for me to evaluate - over reliance on Carson Edwards and I'm not sure how good the B10 is.
o    Tennessee (2) has plenty of talent but is vulnerable and not an auto E8 team. Potential tough matchup with a Cincinnati team in Ohio that can match their physicality in Round 2.

·         Kansas State (4) is undoubtedly the weakest and most overseeded #4 of the group. I'm comfortable putting them in zero E8 combos even if I do 10+ entries.

·         Virginia (1) (similar to Duke) is an automatic E8 in my eyes - based on underlying team and favorable draw

Midwest
·         UNC (1) has the toughest road relative to the other #1 seeds
o    Auburn is drastically underseeded and takes a high variance approach of shooting 30+ 3's per game. They also like to play up-tempo and have high-level athletes. Not the ideal #5 seed to have in your region.
o    Kansas (4) is overrated/overseeded - I wouldn't be surprised to see them get bounced in Round 1 by Northeastern

·         Kentucky (2) has a pretty solid draw relative to their counterparts within the bottom half of this region.
o    Iowa St. is an underseeded, high variance team but they have two tough games just to get to the S16. It's difficult to feel comfortable putting them in E8 from a sheer math standpoint.
o    Houston (3) is overseeded and would be around a PK with Iowa St. so can't feel great about them even making the S16
o    Wofford (7) is a tough #7 and can play with upper-echelon teams but Kentucky has too much NBA talent to give it any real thought - UK probably giving 7.5 or 8 in that game

Bold Prediction: No Big East team reaches the S16

Bracket Contest Advice:
If you are entering a large-field bracket contest, it makes the most sense to avoid picking Duke as champion (or finalist). 

From a game theory perspective, you need to differentiate yourself from the field and there are teams like Gonzaga and Virginia that are just as good as Duke. The perception among the masses suggests a full strength Duke team is the unquestioned top team, but I'm not buying it. Duke may be in the top tier, but they are not alone.

As of this writing, ESPN's bracket tournament challenge suggests roughly 41% of entries picked Duke to reach the National Championship game. 

In large field tournaments, it pays dividends to be a calculated contrarian. Gonzaga (8%), Virginia (7%) and Kentucky (5%) all have similar probabilities of reaching the final game as Duke, yet public perception views these teams as 5-6x less likely to do so.

Thursday, September 6, 2018

2018-19 NFL Week 1 Power Ratings

It's that time of the year again - the NFL is back!

With the 2018 regular season set to begin tonight, it's time to discuss Week 1 power ratings.

As a reminder, I do not put any weight into W/L records when constructing these ratings. Mainstream outlets like ESPN and Bleacher Report basically just rank every team in order according to win/loss record and are heavily influenced by what happened the previous week. I'm pretty sure my eleven-year-old cousin is capable of ranking teams in order (1-32) based on win percentage.

I prefer to take a big picture approach considering I understand you can't overreact to a single game of football - the ball isn't even round! I also realize every team plays a different strength of schedule, and thus raw W/L records don't really tell us much about how good a team is compared to another.

Numerical values are assigned to each team in order to get a better estimation of how much better one is than another. New England is the best team in the NFL according to my ratings, but how much better are they than teams like Pittsburgh, Dallas and Chicago? I look to answer those types of questions when constructing power ratings.


The primary components used to generate these numbers are:
  • Yards-per-play differential from the prior year
  • Season over/under win totals blended with strength of schedule projections
  • Week 1 market expectation

Power ratings are generally used to project the beginning stages of a point spread.

For example, if Minnesota played Carolina on a neutral field, my ratings suggest Minnesota would be about a 3.5-point favorite. Other elements such as location, perception, schedule spots, etc. are then factored into the number to create a more accurate point spread.

Thursday, August 30, 2018

2018-19 NFL Preview: Strength of Schedule Projections


In addition to NFL power ratings (which will be released prior to Week 1), I also create strength of schedule (SOS) power ratings during the offseason. The formula is fairly simple and solely based on season win total numbers.

I prefer to use season win totals from reputable sports books to forecast strength of schedule for the upcoming season. It doesn't make much sense to use win/loss records from the previous year. Not only do NFL teams naturally regress/progress from year to year, but transactions take place during the offseason that can either ameliorate or worsen a roster.

The season win totals posted on Bookmaker.eu have been used to construct SOS power ratings for each team.


The formula used to create an NFL schedule consists of six games between divisional opponents, eight games against two different divisions, and two games versus two different conference opponents based on division ranking from the previous year (sometimes referred to as "interconference games").

To use Arizona as an example, the Cardinals will play the AFC West and NFC North. They also have to face San Francisco, Seattle and Los Angeles two times each because they reside in the same division. Furthermore, Arizona plays Atlanta (3rd place in NFC South) and Washington (3rd place in NFC East) in their two interconference games because each team finished third in their respective divisions last year.

I simply added up the season win total expectations for all 16 of Arizona's opponents and arrived at a strength of schedule rating of 136.4, which projects the Cardinals to have the toughest schedule.

An underrated benefit of this activity is that it adds perspective to the season win numbers and helps identify teams getting respect from the market and oddsmakers alike. New Orleans and Green Bay have similar win totals, but the Saints play a tougher schedule by a fairly wide margin according to the SOS ratings. This suggests the market holds New Orleans in a higher regard than the Packers, but the same conclusion could not be drawn by simply considering raw season win numbers.



The NFC North projects to be the toughest division with the NFC South being a close second. Contrarily, the AFC East grades out as the weakest division by a relatively wide margin.

Although this may not be a flawless exercise, it provides a decent snapshot of the easiest/toughest schedules and certainly has more predictive value than previous season's win percentage.

Monday, August 20, 2018

2018 NFL Prop Bet - First Head Coach Fired


Cliff Welch/Icon Sportswire via Getty Images

While I certainly don't enjoy speculating on the future employment prospects of other human beings, the "First Head Coach Fired" NFL prop holds a special place in my heart.

I'll be transparent from the start - there are relatively slim pickings which is probably a good thing considering we are truly speculating.

Furthermore, there isn't much room for creativity compared to the MVP evaluation that was published a few weeks ago, but nonetheless I want to provide some commentary on the candidates and go through the evaluation exercise.

Favorites

David Richard/AP

Hue Jackson (3/1) is essentially coaching on a one-year deal, but barring an early season collapse, I don't envision him being let go before the end of the season.

It feels strange to say considering his 1-31 record over the past two seasons, but Jackson is essentially playing with house money (at least for this season) considering he finally has a serviceable quarterback in Tyrod Taylor and is somewhat tied to number one pick Baker Mayfield.

Considering everything mentioned above and Jackson's short odds of 3/1, there isn't much incentive to tie up money for six months on an investment that seems fairly speculative.

Dirk Koetter (5/1) - Tampa Bay is slated to be underdogs in their first five games and possess an unfavorable early season bye. It wouldn't be out of the question to envision an early bye week transition following an 0-4 start.

With that said, 5/1 odds aren't long enough to get me particularly excited and I'm not sure it makes sense to tie up money for the entire NFL season in pursuit of a relatively slim payout. Furthermore, Koetter may get a pass at the beginning of the season considering starting quarterback Jameis Winston is suspended for the first three games.


Dark Horses



Marvin Lewis (10/1) - Cincinnati projects to be one of the worst teams in the NFL. With that said, Lewis was given an extension and seems to have the team owner in his corner so it's hard to imagine Lewis losing his job before the end of this season.

At first glance, Jason Garrett looked fairly enticing at 18/1, but the market is relatively bullish on Dallas considering their projected win total of 8, which is enough positive sentiment to keep my hand out of the cookie jar.

Final Verdict

With the exception of Koetter, there aren't any options that I realistically feel good about. However Koetter's relatively short odds aren't enticing enough.

The fact that "No Coach Fired" is the favorite out of the bunch validates my ambivalence.

When it all boils down, there doesn't appear to be any fruitful opportunities within this realm. Better luck next year!

Monday, August 6, 2018

2018-19 NFL MVP Odds - Evaluating Previous Winners and Identifying Dark Horse Candidates

Joe Sargent/Getty Images
With the commencement of training camp and preseason taking place last week, I wanted to get back in the saddle and talk some NFL. I thought it would be a fun exercise to analyze previous MVP award winners in an attempt to uncover some potential sleepers in 2018-19.

Before diving into this year's player pool, let's examine some past winners to identify common qualities and provide historical context.

Pro Football Reference

After evaluating the exhibit above, which includes MVP award winners over the past 30 years, the main takeaways are:
  • 24 / 32 were quarterbacks (75%)
  • 32 / 32 made the playoffs (100%)
  • 32 / 32 won nine or more games (100%)

With the exception of Adrian Peterson's monster 2,300-yard season in 2012, a non-quarterback has not won an MVP award since 2006. Whether it be the pass-heavy mindset or running back by committee approach that currently exists, the days of LaDanian Tomlinson scoring 30 touchdowns are in the rearview mirror.

Look no further than last year when Todd Gurley accumulated over 2,000 total yards and 19 touchdowns for an 11-5 Rams team, but only received eight (out of 50 total) MVP votes.

Additionally, there have been no wide receivers or tight ends to be crowned MVP by the Associated Press since the award originated in 1957.

Moral of the story: The 2018-19 NFL MVP award winner will play the quarterback position on a team that makes the playoffs and has a winning record.

Now that the parameters have been established, let's take a look at this year's candidates.

Source: Bookmaker.eu

All of the MVP candidates listed at Bookmaker.eu are included above for reference along with their corresponding odds. As an example, Tom Brady's MVP odds of "500" correspond to +500 or 5/1.

It is worthy to note that Bookmaker is accepting $5,000 limits on their MVP offering, which resembles a fairly liquid market in the context of player props and suggests the European-based operator has confidence in their numbers.

Since not all of these names are valid candidates to take home the trophy, a more realistic representation is depicted below.

I used current season over/under win totals to estimate the number of wins for each team. Understanding that projections are not an exact science, the condensed player pool contains quarterbacks whose team is expected to win at least eight games in an attempt to account for margin of error.

Source: Bookermaker.eu

While there are plenty of notable names highlighted above, I've identified a handful of viable contenders that are being overlooked.

Ben Roethlisberger (25/1)

Big Ben stood out like a sore thumb after evaluating the condensed player pool of quarterbacks. As previously discussed, the key factors with MVP forecasting are playoff appearances and winning records. Pittsburgh is currently -270 to win the AFC North, which makes them the most likely team to win their division outside of New England.

Considering the state of affairs involving Le'Veon Bell, it's not hard to imagine Big Ben shouldering the offensive workload for Pittsburgh this season. Furthermore, Roethlisberger has more than enough brand equity to satisfy the name recognition/popularity element that inevitably factors into the equation.
Despite showing signs of decline last season along with some retirement rumblings, Roethlisberger appears to be re-energized and ready to roll in 2018. Keep those fingers crossed as a little bit of narrative never hurt anybody!

Although Roethlisberger was listed as high as 33/1 during the initial writing of this analysis, the current price still offers some justifiable upside.

Jim Mone/AP
Kirk Cousins (27/1)

Cousins may not be the most appealing name listed above, but the former Washington Redskins QB has demonstrated the ability to put up favorable numbers. Cousins finished among top-10 quarterbacks in total yards and yards per attempt in each of his three seasons as a full-time starter.

Furthermore, Stefon Diggs and Adam Thielen are more talented than any wide receiver Cousins has played with thus far in his career. Throw in Kyle Rudolph, who ranked as Pro Football Focus' fifth best receiving tight end in 2017, and Cousins has various upper-echelon weapons to choose from.
The Vikings also have the luxury of playing nine games in a dome, which doesn't hurt in the stat padding department.
 
Although Minnesota resides in a rather competitive division, the Vikings are slight favorites to repeat as NFC North champs (+100 to win the division) and projected to win 10 games.

As evidenced above, the driving factor behind MVP candidacy is the ability to win games and make postseason appearances. Cousins fits the profile of previous award winners and presents long enough odds to make a potential investment worthwhile.

Ed Zurga/AP
Patrick Mahomes (66/1)

This one may seem silly on the surface considering Mahomes has only started one NFL game thus far in his young career, but hear me out.

After shipping Alex Smith out of town, Mahomes has been handed the car keys to a relatively high-powered offense. The framework is in place for Mahomes to put up gaudy numbers under the tutelage of Andy Reid (aka The Quarterback Whisperer). The former Texas Tech gunslinger has a myriad of dynamic weapons at his disposal in Travis Kelce, Tyreek Hill and Sammy Watkins.

From a distant view, it seems like Kansas City is going to take a high variance approach this season. The defense was far from spectacular last year, surrendering 5.6 yards per play (T-27th), and Mahomes should find himself in plenty of shootouts.

Despite not being favored to win the AFC West, the division is still somewhat up for grabs. San Diego currently sits at +120 with Kansas City not too far behind at +229 as of this writing. It would not be inconceivable for Kansas City to climb out on top of the AFC West for the second year in a row, which would bode well for Mahomes' MVP prospects.

They call them long shots for a reason, but the situation in Kansas City presents Mahomes with the theoretical upside to jettison himself into the MVP conversation.