Wednesday, May 27, 2020

Boston to win the Stanley Cup

NHL Commissioner Gary Bettman announced yesterday that the NHL plans to return with a 24-team Stanley Cup playoff, starting no earlier than July. The format involves the top four teams in each conference playing a round-robin for seeding, while the next eight teams will engage in a best-of-5 play-in series to advance, with the fifth seed taking on the twelfth and so on. From there, the series would be set by reseeding, with all series best of 7.

In the past, I have used regular season records to simulate the playoffs. So far, that method has never picked the eventual champion, and with this unique playoff set-up, the odds are even worse. But that won't stop me from trying.

Eastern Conference

The top four teams would seed as follows based on regular-season point totals against each other:
Washington (12 points)
Tampa Bay (11, 2-0 vs. Philly)
Philadelphia (11)
Boston (10)

The four play-in series would be as follows, with season series records:
#5 Pittsburgh over #12 Montreal (2-1)
#11 NY Rangers over #6 Carolina (4-0, no wonder the Hurricanes voted against this proposal)
#7 NY Islanders over #10 Florida (3-0)
#8 Toronto over #9 Columbus (1-0-1)

Using the reseeding method:
#11 NY Rangers over #1 Washington (2-1)
#8 Toronto over #2 Tampa Bay (2-1)
#7 NY Islanders over #3 Philadelphia (3-0)
#4 Boston over #5 Pittsburgh (2-1)

Again, reseeding for the conference semifinals:
#4 Boston over #11 NY Rangers (3-0)
#8 Toronto over #7 NY Islanders (1-1, Toronto advances on 7-5 goals)

Eastern Conference Final
#4 Boston over #8 Toronto (2-0-1, same record the Bruins had against the Islanders)

So Boston (who was the only team to crack 100 points) is your Eastern Champ, despite finishing last in the round robin.

Western Conference

The top four teams would seed as follows based on regular-season point totals against each other:
St. Louis (17 points)
Dallas (14)
Colorado (10)
Vegas (6)

The four play-in series would be all upsets as follows:
#12 Chicago over #5 Edmonton (2-1)
#11 Arizona over #6 Nashville (1-1, Coyotes advance on 7-5 goals)
#10 Minnesota over #7 Vancouver (2-1)
#9 Winnipeg over #8 Calgary (1-0, Heritage Classic)

#1 St. Louis over #12 Chicago (4-0)
#2 Dallas over #11 Arizona (2-0)
#3 Colorado over #10 Minnesota (2-2, 13-13 goals, Colorado advances on overall record)
#9 Winnipeg over #4 Vegas (2-0)

Again reseeding for the conference semifinals:
#9 Winnipeg over #1 St. Louis (2-1-1)
#2 Dallas over #3 Colorado (4-0)

Western Conference Final:
#9 Winnipeg over #2 Dallas (both teams went 2-1-1 with an OT loss, Winnipeg advances 13-11 on goals).

Stanley Cup Final

Boston over Winnipeg 2-0

So the Bruins will hoist the Cup. Of course, as this prediction method has never been right, about the only sure thing is that another team will win.

As for a personal prediction, I'll wait until the 16 teams are set and we have some idea how the level of play has changed. With no fans and no home advantage, and most players having recovered from injuries and the long season, it will be a bizarre and unpredictable playoff. Assuming that it actually takes place.



Tuesday, May 26, 2020

The Lockdown Mistake

It is Memorial Day and we are still in lockdown here in NYC, making it over nine weeks that residents have been asked to stay home to "flatten the curve". Although lockdown is entirely inaccurate, as people are always outside, bars are serving beer and cocktails to-go, and parks and beaches are open. The main differences from early March are that non-essential businesses are not yet allowed to serve the public, and many people are wearing masks on the street.

The city continues to tell New Yorkers to shelter in place, despite the curve having been flattened for over a week now. There are ten metrics that are being tracked and all ten must meet a certain level before the city can begin re-opening. I've taken the below chart from Gothamist's extremely detailed statistics page, which is updated daily.

There are still three of the ten that are not being met. One is specific to the city, namely the number of critical cases in public hospitals, which is dropping fairly quickly, but is still a few days away from hitting the magically arbitrary 375 mark. The other two apply to every region in state and represent the percentage of hospital beds and ICU beds available. In both cases here in NYC, the numbers are mostly flat over the past week, which is surprising since fewer people are getting sick, until you realize that the denominator keeps lowering as hospitals reduce their surge expansion. Talk about moving the goalposts.

The people of New York did what we were told, but now our paternal politicians are telling us to keep waiting because it's working. Actually, it already worked. Now a city of 8 million remains hostage to a poorly conceived metric and 52 critical care patients. Small business owners cannot afford more downtime and their livelihoods are in danger of disappearing altogether, while Cuomo and de Blasio reprimand us for our impatience. It is time to start opening the city.

The goal was always to avoid overwhelming the hospitals, and for the most part, that was achieved. Much of the additional capacity that was set up in NYC proved unnecessary. That is a good thing. But the goal seems to have morphed into trying to ensure that nobody else gets the virus. Given that New York State is the hardest hit area in the world (1 in 650 people have been confirmed as COVID-19 fatalities as I write this, and the number is certainly worse than that), it is natural that politicians who oversaw this disaster are wary, but they are making a mistake with their caution.

People need to be outside, especially those belonging to less vulnerable groups, namely the young and healthy. As long as they are wearing masks and practicing social distancing, they will be fine. The CDC shows a case fatality rate of 1/2000 (0.0005) for those under 49, and that includes co-morbidities. Businesses should be allowed to open as along as they engage in proper pandemic protocols during the early phases, limiting customers to promote social distancing and ensuring masks are being worn. That is the case right now for essential businesses in my neighbourhood, and nobody seems to mind. Time to add non-essential businesses to the mix.

The other reason that being outdoors is helpful is that sunshine provides Vitamin D, which has been shown to help reduce the effects of the coronavirus. It does not prevent you from getting the virus, but it does make it less likely that you will suffer a severe reaction. Tokyo is even denser than New York and they have managed to beat down the virus rather quickly. Widespread use of masks, better health and hygiene, and Vitamin D from fatty fish are some of the reasons. If you can't get outside, take a daily supplement. It can't hurt.

You might not have heard about this Vitamin D connection in the mainstream media, because it has yet to be proven conclusively, but do your own research. It is a lot better than hydroxychloroquine, about which you have probably heard far too much. The mainstream media has been rather unimpressive in general over the past few months, on both sides of the political fence. And that is another problem; this pandemic has become political here in the United States. The left-leaning New York Times and CNN are more concerned about blaming Trump than getting useful information out to their audience; I don't watch or read right-wing media but from what I can tell on Twitter, they are desperately trying to exonerate Trump. As I mentioned before, there is no point blaming a single individual; this was a failure of all leadership, including Governor Cuomo. The hypocrisy of the left is apparent as they fete Cuomo for being decisive and telegenic, despite his dithering and bickering with Mayor de Blasio in early March. It is also disingenuous to suggest that thousands of lives could have been saved if Trump had issued stay-at-home orders two weeks earlier - nobody would have listened. They are not listening now, with large crowds forming at beaches and lakes around the country, despite 100,000 dead across the nation, to expect that people would have stayed home with almost no cases reported is dishonest at best.

Politics have even infiltrated the mask debate, with conservatives foolishly thinking it is an affront to their liberties to be forced to wear a mask, while liberals weaponize the deaths of family members, friends and colleagues whenever they see those without a mask. In online forums, I have seen statements such as "I watched two buddies die of this and these assholes still refuse to wear a mask".  I do wish everyone would wear a mask as it protects others rather than yourself, but in any population you will have those that put themselves above all else. My 2-year-old is the same, so I understand.

It is tragic that we have lost so many, but this is not the first time this has happened, nor will it be the last. Did you know about the Hong Kong flu pandemic between 1968-70 that took about a million worldwide (with the population less than half what it is now) and 100,000 stateside (with about 62% of today's population)? Life went on as usual, but that is barely mentioned now. Yes, the diseases are different, but back then, there was not an overarching fear of death that seems to pervade every decision (except, of course, any decision that would result in fewer guns).

The death toll here is certainly crushing, but it must be stated that the vast majority of victims were elderly and/or in bad health, and largely poor and minorities. Remember Vitamin D? Nursing home residents were stuck inside over the long winter, while the poor have little time to enjoy the sunshine as they commute to and from thankless jobs. Minorities also have a natural deficiency. Yet has any government urged their population to consider the benefits of supplements? Nope. Fear rules the day, not education.

We can mourn the dead, but life goes on, as it always has and always will. It is time to move ahead, knowing what we know, and help those suffering economically get back on their feet. Yes, more will catch the virus and more will die because of it. People will also continue to die of cancer (wear a mask when buying cigarettes), heart disease, the flu, car accidents, guns, and hunger (20,000 a day worldwide, many of them children). You cannot prevent death, you can only delay it. But if the cost of delaying some COVID-19 deaths is to make life miserable for many (millions of job losses and thousands of businesses forced to close), it is time to let people decide for themselves. If you are scared or vulnerable, stay home and isolate. Let the rest of us live our lives and take those risks.

Now for the mandatory sports comment. Some governments have said that they will not allow fans to attend games, even with masks, until a vaccine is available. Such incredible overreach. Let us decide. I believe that outdoor stadiums can open soon, not in the first phase, but the second. Reduced capacity, limiting concessions to grab and go items with touchless payment, hand sanitizer everywhere, improving security so that lines are well spaced, and requiring masks when not eating should be enough to allow some fans in. Hockey rinks and basketball courts might need some additional measures, but it can be done.

In the meantime, the goal should continue to be protecting hospitals and their staff. Spending time outside (Vitamin D), wearing masks, and social distancing are the ways that individuals can help to achieve this. The government has to let businesses open up so many can get back to work. Testing and contact tracing are new measures that will remain active for several months if not longer. Past pandemics have lasted between two and four years, so this one is just getting started. Significant mistakes were made in the early going, now it is time to set things right. We cannot spend two years indoors waiting for a vaccine, led by panicking politicians. Educate the populace and let us choose our path.



Monday, May 25, 2020

The Best Undrafted Players in NFL History

This is the final post in my series on the NFL draft. I could write a detailed analysis about every draft since 1980 but there is little additional insight to be found. I have shown that drafting well is linked to on-field success and that finding three or four elite players in consecutive drafts can immediately help a team attain a title. Due to short careers and injuries however, you have a narrow window in which to have all of your recent picks play at or near peak level in the same season, which is why dynasties are rarer these days. Trades and free agents are equally important pieces of the puzzle, as the Patriots have shown over the years.

There are two types of free agents: those who have already played several seasons elsewhere and are generally known commodities; and those who have yet to play in the league but were never even drafted. It is this second group that I am discussing in this post. More than any other league, undrafted players have a major impact in the NFL. With 1,696 players on rosters and many others on practice squads, yet only about 256 players drafted every season, there are a lot of opportunities for those overlooked on draft day to make their mark. So once again using Pro Football Reference's Approximate Value metric, I created two lists. The first comprises the top ten (plus ties) best undrafted players in history, that is, the highest total Career Approximate Value (Hall of Famers denoted with an asterisk).
Rk   Player            Pos   From   To    tCAV
 1   Warren Moon*       QB   1984  2000    166
 2   Jeff Saturday       C   1999  2012    156
 3   Mick Tingelhoff*    C   1962  1978    151
 4   London Fletcher    LB   1998  2013    147
 5   Dave Krieg         QB   1980  1998    138
 6   John Randle*       DT   1990  2003    138
 7   Jim Hart           QB   1966  1984    136
 8   Jason Peters        T   2004  2019    130
 9   Antonio Gates      TE   2003  2018    127
10   Sam Mills          LB   1986  1997    126
10   Cornell Green      DB   1962  1974    126
10   Rod Smith          WR   1995  2006    126
TCAV is cumulative, so the longer you play, the higher your tCAV will go. Imagine if Warren Moon hadn't spent six seasons in Edmonton (where he won five Grey Cups). A good variety of positions, with two centers following Moon, and eras, from the 1960s to today, are represented as well.

The second list contains the ten most effective undrafted players, i.e. those with the best tCAV/G. This eliminates career length as a factor, though there are still some names that appear on both lists.
Rk   Player            Pos   From   To    tCAV
 1   Kurt Warner*       QB   1998  2009   .911
 2   Jeff Garcia        QB   1999  2009   .832
 3   Priest Holmes      RB   1997  2007   .832
 4   Warren Moon*       QB   1984  2000   .798
 5   Tony Romo          QB   2004  2016   .744
 6   Jeff Saturday       C   1999  2012   .739
 7   Willie Wood*       DB   1960  1971   .723
 8   Jim Langer*         C   1970  1981   .709
 9   Sam Mills          LB   1986  1997   .696
10   Cornell Green      DB   1962  1974   .692
Kurt Warner had a shorter career as he played some time in the Arena Football League and NFL Europe before beginning his miraculous rise to fame. Jeff Garcia spent five seasons with Calgary before getting his chance, while Priest Holmes signed with Baltimore and won a Super Bowl with them before moving to greater success in Kansas City. You can use PFR's play index to find a complete list of undrafted players and see many other familiar names on it.

Note that these lists includes only players who played their entire career in the NFL, so those who started in the AFL, such as Jim Otto (who would be second overall in tCAV) and Larry Little (8th overall) are not shown.

That ends the series on the NFL draft. I really enjoy PFR's play index and their Approximate Value metric as it allows me to gain some insight into the game that would be otherwise untapped. If you are an NFL fan and looking for things to do during the pandemic, check them out at



Sunday, May 24, 2020

Revisiting the 2012 NFL Draft

I'm continuing to post about the NFL draft because there is nothing else to post about and I find it fascinating to go back and see how teams did several years on. As I mentioned previously, giving draft grades before the players have even stepped inside an NFL facility is silly. One of the most obvious examples of this is the 2012 draft class of the Seattle Seahawks, which was nearly universally panned, coming in 30th on Pro Football Outsiders' report card that summarizes the grades from 9 NFL writers. Bleacher Report gave the Seahawks the worst grade of D, calling first-rounder Bruce Irvin a reach (and saying 3rd-round pick Russell Wilson was "messed up" as they club had signed Matt Flynn in the offseason). Two years later, the Seahawks were Super Bowl champs.

Hindsight is 20/20 of course, and I am not trying to criticize these publications; I do understand the appeal of these articles as fans want to understand how their team did. The point is that they are silly and fans get little meaningful insight. Some writers do go back and re-grade their picks after a few seasons, but now that I have discovered Pro Football Reference's Approximate Value metric, we can go one better and give a precise ranking. The 2012 NFL draft ranking is below:
Team   Picks   Avg tCAV   Grade
SEA      10     40.20      30
CAR       7     26.86      13
NE        7     26.71       5
DEN       7     26.14      29
TB        7     26.00       6
HOU       8     25.38      21
PHI       9     24.78       4
MIA       9     24.22      18
CIN      10     22.30       1
WAS       9     22.11      15
IND      10     21.30       3
STL      10     20.50      14
BUF       9     20.33      17
MIN      10     18.80      11
ARI       7     18.57      16
CLE      11     17.18      19
PIT       9     16.22       2
SD        7     16.00       9
JAX       6     15.33      27
TEN       7     15.29      24
NO        5     14.80      32
GB        8     14.75       7
DET       8     14.75      20
KC        8     14.13      23
BAL       8     13.25      12
NYJ       8     12.88      28
CHI       6     12.83      22
DAL       7     11.86      10
OAK       6      9.33      31
NYG       7      6.43       8
ATL       6      5.83      25
SF        7      4.57      26
As you can see, Seattle has by far the best performance, mostly because of Wilson and second-round pick Bobby Wagner. They are the two best players taken in the draft, Irvin spent four seasons with the Seahawks before moving around the league, re-signing with Seattle a month ago, and his tCAV/G ranks 50th in this draft class, suggesting he was a reach after all.

The final column in the table above is the draft ranking as provided by Football Outsiders. Despite Seattle and Denver being completely underestimated, there is some correlation between the tCAV ranking and the draft ranking, with the coefficient being 0.27, suggesting weak correlation. So these guys do know something, just not as much as they think.

What is interesting here is that three of the top four teams in the ranking happened to be conference finalists a season later. Obviously, there is more to a successful team than a single draft, but with NFL careers so short, a good draft or two can propel you to the top quickly. New England took Dont'a Hightower and Chandler Jones and made their team even stronger. Denver's top three picks were Derek Wolfe, Danny Trevathan, and Malik Jackson who joined Von Miller, second overall pick from 2011, on the Bronco defense. Although they were embarrassed by Seattle in 2013, they became a defensive juggernaut that stifled Carolina in the 2015 Super Bowl and gave Peyton Manning his second ring. Brock Osweiler was also chosen by Denver that year; without him they would be second in the ranking.

On the other hand, the fourth conference finalist in 2013 was San Francisco, who had the worst 2012 draft. They were still enjoying the fruits of previous drafts, having lost to Baltimore in the 2012 Super Bowl. They upset Carolina in the Divisional Round, but the Panthers had Luke Kuechly and Josh Norman to go along with first overall pick Cam Newton in 2011 and eventually made it to the Super Bowl two years later.

Cincinnati was ranked number 1 with Dre Kirkpatrick and Kevin Zeitler their first-round picks; both have had solid careers but not enough to help the Bengals to any sort of glory. Of course, football is a game that changes very quickly, due to injuries and the fact that many positions become quite fungible as you move down the talent chart. In other words, there is a large drop off between the elite players at a position and the rest, who are mostly interchangeable. As you can see by looking back at the 2012 draft, finding just a couple of elite players can go a long way to a championship season.



Saturday, May 23, 2020

The Best and Worst of the NFL Draft from 2000-09

For my next post in the series on the NFL draft, I've again used Pro Football Reference's Approximate Value (AV) metric for all drafts from 2000-09. A player is assigned an AV each season based on his statistics, and the AV over his career is CAV. PFR weights the CAV as well, so there are two metrics: total CAV (tCAV) and weighted CAV (wCAV). Playoffs are not included. AV is similar to WAR in baseball, and is for entertainment purposes only.

In this case, I have compiled a list of the top player by tCAV, sleepers (highest tCAV drafted in 3rd round or after), busts (generally the first round pick with lowest tCAV), and surprises (using tCAV/G, I added them because I was surprised to see them so high on the list).
Year    Best Player (Pick, tCAV)   Top Sleeper (Pick, tCAV)     Biggest Bust (Team, Pick, tCAV)  Surprise (tCav/G)
2000    Tom Brady (199, 280)       Tom Brady (199, 280)         R. Jay Soward (Jax, 29,2)        Chad Pennington (.697)
2001    Drew Brees (32, 267)       Steve Smith (74, 103)        Jamal Reynolds (GB, 10, 3)       Matt Light (.710)
2002    Julius Peppers (2, 185)    Brian Westbrook (91, 80)     Wendell Bryant (Ari, 12, 4)      David Garrard (.802)
2003    Terrell Suggs (10,156)     Lance Briggs (68, 124)       Charles Rogers (Det, 2,4)        Domanick Williams (.850)
2004    Phillip Rivers (4,204)     Jared Allen (126, 127)       Rashaun Woods (SF, 31, 1)        Bob Sanders (.680)
2005    Aaron Rodgers (24, 184)    Frank Gore (65, 129)         Troy Williamson (Min, 7, 8)      Logan Mankins (.801)
2006    Jahri Evans (108, 143)     Jahri Evans (108, 143)       Matt Leinart (Ari, 10, 12)       Jay Cutler (.712)
2007    Adrian Petersen (7,121)    Marshall Yanda (86, 113)     JaMarcus Russell (Oak, 1, 6)     Patrick Willis (.973)
2008    Matt Ryan (3, 179)         Josh Sitton (135, 98)        Vernon Gholston (NYJ, 6, 4)      Carl Nicks (.795)
2009    Matthew Stafford (1, 118)  Julian Edelman (232, 69)     Aaron Maybin (Buf, 11, 4)        Josh Freeman (.661)
In some years, the biggest bust might have had a higher tCAV rating than some other draftees, but name recognition and higher draft position also come into play (Leinart and Russell in particular). Some might think that Aaron Curry, drafted 4th overall in 2009, might be a bigger bust than Maybin, but Curry did play 35 games with Seattle (CAV 16), while Maybin's CAV of 1 was achieved in 27 games for the Bills. Meanwhile, the Patriots have the two biggest sleepers in Brady and Edelman, part of the reason for their incredible success.

I've spent a lot of time looking over these draft results and it can be quite fun to see who succeeded and who didn't. It is also entertaining to find old draft columns, like this one from Skip Bayless in 2006, and see how the writer did. Cutler was certainly a better pick than Leinart, so chalk one up for the Skipper.

The next post will be on undrafted players, and I'll also have a post on the 2012 draft and how Seattle dominated that day, leading to their 2013 Super Bowl win. Keep checking back for updates.



Friday, May 22, 2020

NFL Draft by Position

Continuing my series on the NFL draft using Pro Football Reference's Approximate Value metric, this post looks at positional rankings from 2000-09. PFR assigns every draftee 1 of 15 positions, I combine guards, tackles, and centers into offensive linemen, while defensive linemen comprise ends, defensive tackles, and even three nose tackles. The remaining positions are what you would expect: quarterbacks, running backs, wide receivers and tight ends on the offensive side; linebackers and defensive backs opposite, and punters and kickers on special teams.

Using PFR's Total Career Approximate Value and dividing by the number of picks over the decade, you get the career value for each position. Dividing that by the total number of games played, you get the effectiveness at each position, by which the list is ordered. As you would expect, quarterbacks generate the most value by far, but next are the all important and often overlooked offensive linemen. After all, QBs can't do much if they are not protected. Running backs follow, then defensive linemen and linebackers. Fullbacks are the least valuable position, which is not that surprising, given their responsibilities will not generate a lot of measurable stats. The table is below:
Pos     Picks   NFL   Pct    Avg   tCAV/G
QB       130    103   79.2  26.47   0.602
OL       420    350   83.3  25.56   0.402
RB       197    176   89.3  20.66   0.358
DL       427    386   90.4  23.28   0.340
LB       316    294   93.0  21.79   0.328
WR       334    286   85.6  16.97   0.319
DB       494    459   92.9  18.59   0.281
K         25     22   88.0  16.72   0.182
TE       153    135   88.2  11.51   0.165
P         21     20   95.2  19.19   0.158
FB        34     34  100.0   3.88   0.057
The number of picks includes players who did not make the NFL, so I have included the number of those that actually did play at least a game in the league. The average tCAV does include players who failed to play a game, though it doesn't make a difference in the order. Ignoring the low numbers for punters and fullbacks, 93% of linebackers made the league, followed closely by defensive backs. Surprisingly less than 80% of quarterbacks actually played a game, by far the smallest percentage. And by play, I mean play, carrying a clipboard does not count.

How is this information useful? Like everything else on this blog, not very. But if you happen to be an NFL scout or GM, you might want to pay attention to the guys battling it out at the line of scrimmage rather than focusing on defensive backs and tight ends.



Wednesday, May 20, 2020

Reviewing the NFL Draft 2000-09

Last month I wrote about the NFL draft, looking at the years from 2009-19 and determining that the New Orleans Saints were the best at picking players during that time. But many of the players drafted are still active and so that ranking may change over the next decade.

So I decided to go another 10 years back and analyze the NFL drafts from 2000-2009. During that time, 2,551 players were drafted, with 2,265 (88.8%) playing at least one game in the league. Of those, only 67 are still active as I write this, so the rankings will not change much, if at all. One interesting statistic: the average career length of those players who did play at least one game and are now retired is 72 games, or 4.5 seasons. That does not include undrafted players, of which there were 1,350 who played their first game between 2000 and 2009.

I also decided to switch to Pro Football Reference's total Career Approximate Value (tCAV) instead of the weighted version (wCAV) that I referenced in previous posts. As I have stated before, tCAV is an approximation of the value a player provides over his career and is the best statistic I have found at measuring value between positions, so that linemen, who have few easily available stats to determine their value, do not get overlooked.

The average tCAV for players drafted during that decade is 20.63. Of course, the best tCAV belongs to Tom Brady, whose total of 280 is 13 points better than Drew Brees. Third is Phillip Rivers at 204, but again, tCAV depends on career longevity and Brady and Brees have enjoyed a few extra seasons.

If you want to figure out the most effective player, you need to divide tCAV by the number of games played (tCAV/G). In a mild surprise, Aaron Rodgers is the champ, with 184 tCAV in just 181 games for a tCAV/G of 1.017. He is the only player to be over 1. Brady is second at .982, with 49ers linebacker Patrick Willis, who played only 8 seasons, third at .973. The average is .324. Again, two caveats: playoffs are not included and don't take this too seriously.

Here is a summary of each draft:
Year  Picks  NFL  Pct     Top tCAV     Worst tCAV   Avg tCAV   tCAV/G
2000   254   219  86.2%   NYJ (46.4)   Dal (6.8)      21.11     0.332
2001   246   220  89.4%   SDG (58.0)   Oak (6.0)      23.43     0.347
2002   261   227  87.0%   Pit (37.8)   TB  (5.1)      19.35     0.305
2003   262   237  90.5%   Ari (45.9)   Den (6.0)      21.47     0.328
2004   255   228  89.4%   Ari (62.9)   Phi (6.5)      21.04     0.329
2005   255   221  86.7%   Dal (49.6)   Buf (5.0)      20.88     0.343
2006   255   227  89.0%   NO  (61.0)   Stl (4.1)      23.57     0.337
2007   255   226  88.6%   NYJ (55.0)   NE  (5.3)      18.83     0.310
2008   252   230  91.3%   Atl (35.6)   Pit (7.4)      19.23     0.319
2009   256   230  89.8%   NYJ (37.7)   Dal (1.3)      17.55     0.301
The percentage of players making the NFL is reasonably consistent year to year. The tCAV for the latter years is lower because there are still a few players competing and those numbers should go up a bit, but not enough to alter the fact that 2009 was the worst year, confirmed by tCAV/G.

Overall, what you will notice is that there is no consistency from year to year. The Jets were the top team in three years, but in two of those they had few picks (4 in 2007, 3 in 2009).  Arizona did well in back-to-back years and got to the Super Bowl in 2008. Dallas was the worst team twice and the top team once (they picked up DeMarcus Ware 11th overall that year and snagged Jay Ratliff in the 7th round). The Steelers had a terrible draft in 2008, but still won the Super Bowl over the Cardinals, helped by their 2002 draft when they picked Antwaan Randle El, Larry Foote, and Brett Keisel among others (not to mention 2004, when they got Ben Roethlisberger).

In order to find the best drafting team of the decade, you need to calculate the average tCAV for that period, and surprise, surprise, the Jets top the list. Yes, the team that has not reached a Super Bowl since 1969 managed to choose the best players between 2000 and 2009. The next three teams were San Diego, Arizona, and Carolina - all of them without a Super Bowl as well. Baltimore, Pittsburgh, and Green Bay come next, and they all did win Super Bowls this century. Tampa Bay, who won it all in 2002, finished last.

If you look at tCAV/G, the Chargers were top at 0.399. They would have been even higher if they had actually drafted Phillip Rivers, who is a much more effective regular season QB than Eli Manning. In fact, if I switch those two to their actual teams, the Chargers average tCAV jumps to 28.71, just above the Jets. Of course, the Chargers also drafted Brees and get credit for that pick despite letting him go to give the job to Rivers. Meanwhile, New England jumps to second thanks to Tom Terrific. The Raiders were worst in this category, making that 2002 Super Bowl a battle between two terrible drafting teams, neither who have returned since.

Here is the whole table:
Rk  Team   Picks   Avg   tCAV/G (Rk)
 1  NYJ     67    28.70   0.351 (6)
 2  SDG     77    28.21   0.399 (1)
 3  ARI     74    26.54   0.356 (4)
 4  CAR     80    25.26   0.343 (9)
 5  BAL     81    25.16   0.348 (8)
 6  PIT     78    23.99   0.361 (3)
 7  GNB     91    23.16   0.355 (5)
 8  CIN     83    22.94   0.334 (16) 
 9  NOR     70    22.63   0.336 (13)
10  ATL     80    22.39   0.351 (7)   
11  SFO     88    22.25   0.305 (23)
12  NYG     74    22.24   0.324 (19)
13  NWE     89    22.19   0.381 (2)
14  IND     82    21.72   0.339 (10)
15  MIN     71    20.54   0.331 (18)
16  HOU*    66    20.52   0.336 (14)
17  JAX     88    20.38   0.335 (15)
18  PHI     82    20.34   0.331 (17)
19  DAL     80    20.26   0.305 (22)
20  CHI     87    19.91   0.336 (12)
21  BUF     84    19.14   0.312 (20)
22  SEA     84    18.98   0.285 (27)
23  DEN     80    18.93   0.336 (11)
24  DET     78    18.15   0.311 (21)
25  TEN     95    18.13   0.284 (29)
26  KAN     80    17.16   0.300 (24)
27  CLE     80    16.76   0.289 (26) 
28  MIA     75    16.69   0.278 (30)
29  OAK     77    16.03   0.249 (32)
30  STL     86    15.91   0.285 (28)
31  WAS     63    13.65   0.291 (25)
32  TAM     81    12.26   0.254 (31)
* Houston did not draft until 2002
There is a substantial difference between the top and bottom teams that you can't see when looking year to year. You might wonder if there is correlation between success on draft day and wins on game day, and the answer is yes. The correlation coefficient between the average number of wins from 2000-09 and tCAV/G is 0.41, which implies a moderate correlation between the two, as you can see below.

Of course, correlation does not equal causation. General managers and scouts that do well on draft day are probably going to do well in finding undrafted players, making trades, and signing free agents and those impact wins as well. And there is also the fact that for the first couple of seasons in this chart, most of the players contributing to wins were drafted before 2000. There is a lag between drafting a player and having him reach his peak and that would have to be studied next, but such a project would take far more time that I have at the moment. For now, you can rest easy knowing that drafting well helps you win, but as Jets fans have found out, there is a lot more to NFL success than just that.

I'll continue my NFL draft series with a post on positional rankings as well as one listing the best, worst, and most surprising players from each draft in the first ten years of this century. Check back for those shortly.



Saturday, May 9, 2020

The 2020 NFL Road Trip

The 2020 NFL schedule was released on Thursday and it is still too early to know if fans will be able to attend games. But it is not too early to come up with another season-long road trip plan! I used to do this every season up until I actually took my own trip in 2013, and then a couple of seasons after that, but eventually it became rather pointless to continue. But with sports still on hiatus (don't talk to me about the KBO please), there is no reason not to fantasize that there may be football this fall. Without further ado, here is the 2020 NFL Road Trip plan!
Thu, Sep 10  Houston at Kansas City 7:20 - Season Opener
Mon, Sep 14  Tennessee at Denver 8:10 
Sun, Sep 20  Washington at Arizona 1:05
Mon, Sep 21  New Orleans at Las Vegas 5:15
Sun, Sep 27  NY Jets at Indianapolis 4:05
Mon, Sep 28  Kansas City at Baltimore 8:15
Sun, Oct  4  New Orleans at Detroit 1:00
Mon, Oct  5  Atlanta at Green Bay 7:15
Thu, Oct  8  Tampa Bay at Chicago 7:20
Sun, Oct 11  Indianapolis at Cleveland 4:25
Thu, Oct 15  Kansas City at Buffalo 1:00
Sun, Oct 18  Atlanta at Minnesota 12:00
Mon, Oct 19  Arizona at Dallas 7:15
Sun, Oct 25  Green Bay at Houston 12:00
Thu, Oct 29  Atlanta at Carolina 8:20
Sun, Nov  1  LA Rams at Miami 1:00
Sun, Nov  8  Houston at Jacksonville 1:00
Sun, Nov  8  New Orleans at Tampa Bay 8:20
Thu, Nov 12  Indianapolis at Tennessee 8:30
Sun, Nov 15  San Francisco at New Orleans 3:25
Sun, Nov 22  Cincinnati at Washington 1:00
Thu, Nov 26  Baltimore at Pittsburgh 8:20 - Thanksgiving
Mon, Nov 30  Seattle at Philadelphia 8:15
Sun, Dec  6  NY Giants at Seattle 1:05
Mon, Dec  7  Buffalo at San Francisco 5:15
Thu, Dec 10  New England at LA Rams 5:20
Sun, Dec 13  Atlanta at LA Chargers 1:25
Sun, Dec 20  Tampa Bay at Atlanta 1:00
Mon, Dec 21  Pittsburgh at Cincinnati 8:15
Sun, Dec 27  Cleveland at NY Jets 1:00
Mon, Dec 28  Buffalo at New England 8:15
Sun, Jan 21  Dallas at New York Giants 1:00
Total highway miles would be 22,160,  assuming starting and finishing in NYC. That is about 3,000 more highway miles than I drove on my trip (I finished at 20,353). There is one doubleheader with Jacksonville at 1 and Tampa Bay at 8:20 on November 8, which is technically possible as the two stadiums are just over 3 hours apart. The toughest drive is Minnesota to Dallas (1,000 miles) but you would have 24 hours to do it. You would also have to get from Philly to Seattle (2,830 miles) in 5 days, but it can be done. There are 7 Thursday and 9 Monday games.

Of course, the chance of the entire season taking place with fans in attendance is low, so I can't imagine anyone actually shelling out money for hotels and tickets quite yet. Still, no harm in hoping.



Friday, May 8, 2020

Revisiting the 2000 NFL Draft

The third in my series of NFL draft posts looks back at the 2000 event. For those that are not familiar with that draft, Tom Brady was selected with the 199th pick, making him the greatest sleeper pick in sports history. Not only has he led the Patriots to six Super Bowls, Brady is the only member of that draft class still active, albeit with the Buccaneers now. The most recent retirees were both kickers drafted by Oakland: Sebastian Janikowski who retired after the 2018 season, and Shane Lechler, who officially hung up his boots in 2019, but last played two years prior.

As before, I am using Pro Football Reference's Approximate Value (AV) metric to rate the career performance of each player. In the past two posts, I used the weighted version of this statistic, but this time, I am using the total Career Approximate Value. To highlight the difference between the two, Brady's total CAV (tCAV) is 280, the highest of all-time (Peyton Manning is second at 271) while his weighted CAV (wCAV) is 176, one behind Manning. The longer a player remains active, the larger the variance between his tCAV and wCAV. For the 2000 draft as a whole, tCAV averages 20.16, wCAV is 17.89. Note that playoffs are not included in these calculations.

Obviously, New England "won" the 2000 draft, even though the rest of their picks were pretty lousy. Excluding Brady, the other nine Patriot picks averaged 8.44 tCAV. That would put them 30th out of the 31 teams that drafted that year, in front of only Dallas, whose 5 picks averaged 6.8 tCAV. With Brady, the Patriots averaged 33.6 tCAV, good for third. So yeah, grabbing the GOAT makes a bit of a difference.

The two teams that bettered them: the Jets, who had four first-round picks and topped the leaderboard at 46.38 tCAV, while Baltimore was second at 36.83, thanks to Jamal Lewis (who ran for 102 yards in their Super Bowl win that season) and Adalius Thomas.

The top 15 players selected in that draft, according to tCav:
Rnd   Pick    Team     Player             Pos  LastYr   G     tCAV
6      199    NWE      Tom Brady           QB   2019   285     280
1        9    CHI      Brian Urlacher*     LB   2012   182     150
1       13    NYJ      John Abraham        DE   2014   192     117
2       60    JAX      Brad Meester         C   2013   209     100
1       12    NYJ      Shaun Ellis         DE   2011   184      95
2       44    GNB      Chad Clifton         T   2011   165      92
1       16    SFO      Julian Peterson     LB   2010   158      90
1       30    TEN      Keith Bulluck       LB   2010   170      88
1        5    BAL      Jamal Lewis         RB   2009   131      83
1        8    PIT      Plaxico Burress     WR   2012   148      83
3       78    NYJ      Laveranues Coles    WR   2009   153      80
1       19    SEA      Shaun Alexander     RB   2008   123      79
1        7    ARI      Thomas Jones        RB   2011   180      78
1        3    WAS      Chris Samuels        T   2009   141      77
6      186    BAL      Adalius Thomas      LB   2009   135      77
* Hall of Fame
Note that the Jets did quite well with Ellis and Abraham in the first round and Coles in the third (Chad Pennington was another first rounder for Gang Green, but his tCav was 62).  They made the playoffs six times between 2001-10, but with Brady in the same division, they were never quite able to get over the hump.

The biggest bust is probably R. Jay Soward, drafted 29th by the Jaguars. Soward had a serious drinking problem and was out of the league after the 2000 season (though somehow the Jaguars still retain his NFL rights 20 years later). Soward did win a Grey Cup with the Argos in 2004, so all was not lost. The first overall pick was Courtney Brown, drafted by the Browns, but he only lasted four seasons there before finishing his career with a final campaign in Denver, so there is an argument that he was the biggest bust.

I could go on, but have a look at the PFR draft page yourself and play around. You can do it for every draft, and it is fun to look back and see who really "won" that draft, rather than rely on grades that are given out before any player has even started training camp.



Saturday, May 2, 2020

Which Colleges Produce the Best NFL Players

I am continuing my analysis of the NFL drafts, downloading every draft since 2000 (the one with Tom Brady), which comprises a total of 5,095 players. Those players went to a total of 309 institutions, including a few in Canada and overseas. So which school gets bragging rights for having the best NFL players? The answer will surprise you.

As in the previous post, I use Pro Football Reference's Approximate Value calculation to determine the overall success of an NFL player. PFR's founder Doug Drinen created a very complicated formula that assigns a number to each player season and called it Approximate Value (AV). Career Approximate Value (CAV) is the sum of a player's season AVs. But the number that is reported with the draft results and thus is the number I'm using is the Weighted Career Approximate Value, which takes 100% of the player's best AV season, 95% of his second-best, and so on. This is done to give more weight to a player's better seasons. I still use the acronym CAV for the weighted number because that is what PFR does. If you want the details, start here.

One caveat: the school listed is the school from where the player is drafted. No accounting is made if the player transferred. So Wisconsin gets credit for Russell Wilson's NFL success as he played his final year of college there, having transferred from North Carolina State, where he played three seasons.

To get started, I looked at the total number of draft picks for each school between 2000 and 2019. No surprise to see all power conference schools in the top 10:
Ohio St.     131
Alabama      118
Miami (FL)   115
LSU          112
Florida      110
USC          110
Florida St.  108
Georgia      108
Oklahoma      97
Notre Dame    85
But I wanted to see which schools had the best players in the NFL. I summed the CAV for each school and divided this total by the number of total draft picks for schools that had at least 20 picks, an average of one per year, to get that school's Draft Score. The results were not what I expected:
School        Picks     CAV   Draft Score
Purdue          43     1073      24.95
Texas           68     1576      23.18
California      63     1383      21.95
Miami (FL)     115     2383      20.72
Boston Col.     42      869      20.69
Mississippi     48      962      20.04
Tennessee       80     1538      19.23
Michigan        83     1589      19.14
Central Florida 29      511      17.62
Georgia Tech    37      645      17.43
Purdue has seen a number of excellent players graduate to the pros, most notably Drew Brees, who is second in CAV among active players behind Brady. But Purdue has also sent Matt Light, Shaun Phillips, Nick Hardwick, Rob Ninkovich, Cliff Avril, and Ryan Kerrigan, among many other solid players. You might notice something here, though. All of these players were drafted early in the century, with Kerrigan the latest, having been picked in 2011. In fact, since then, the Boilermakers have only had 9 players drafted, with Kawann Short (2nd round, 2013) the only one picked before the 5th round. The program is in decline.

Clearly, this analysis is biased against recent picks, whose career AV is still low, yet they are treated as if they are retired. Schools with several of these more recent picks will be dragged down in the rankings. To overcome this, I decided to compute Career Approximate Value per Game (CAV/G) and the results were substantially different.
School          Picks    CAV     Games   CAV/G
Boston College    42     869      2285   0.380
Alabama          118    1898      5673   0.335
Michigan          83    1589      4824   0.329
Mississippi       48     962      2934   0.328
USC              110    1917      5883   0.326
Purdue            43    1073      3293   0.326
Wisconsin         81    1334      4095   0.326
California        63    1383      4250   0.325
Georgia          108    1876      5777   0.325
Oklahoma St.      33     549      1707   0.322
Texas             68    1576      4937   0.319
Utah              53     908      2873   0.316
Miami (FL)       115    2383      7555   0.315
Oklahoma          97    1338      4267   0.314
Pittsburgh        49     825      2652   0.311
Another surprise. I don't think anybody considers Boston College a football factory. The Eagles have seen Matt Ryan and Luke Kuechly become stars, but also Chris Snee and Anthony Castonzo went here. Recent picks Justin Simmons and Matt Milano were also Eagles and their careers are off to a good start. As an aside, Boston College was also the home of one of just three players to finish with a negative CAV, namely Nate Freese, who went 3/7 in field goals for Detroit in 2014 and was cut after just three games. The other two are Ryan Lindley and Owen Pochman.

Schools with fewer picks will generally do better in these rankings. The more picks from a school, the more likely some of them will be in the lower rounds, and thus have lower CAVs. So to see Alabama second in the CAV/G ranking with 118 picks (nearly 6 per year) shows that they are a true football factory. I'm guessing the Dolphins are excited to have Tua Tagovailoa on board.

To demonstrate how having fewer picks will help your rankings, the school with the highest CAV/G is West Alabama, where Tyreek Hill went. His CAV is an impressive 47 in just 59 games (0.797 CAV/G) and no other player has been drafted from there. But NFL scouts will not suddenly start flocking to Livingston, I presume.

Finally, don't take these rankings too seriously. The key word in CAV is approximate, and I am just having a bit of fun playing around with the number. You never know what you might find. For me, I have a road trip plan arising from this analysis as Purdue travels to Boston College on September 26. Let's hope the season actually takes place.