Posts Tagged ‘Moneyball’

Ryan Holliday, General Manager

Posted from San Antonio, Texas at 7:04 pm, Monday, February 20th, 2017

I’m pretty sure that writing a post that combines statistical analysis and the Cleveland Browns is a surefire way to drive away any remaining readers of this journal, but math and the NFL’s worst team are two of my favorite subjects, so against better judgement I’m going to indulge myself.

Browns fans apparently hate the idea of trading away the #1 pick in the upcoming draft, but here’s why I’d do it anyhow (see also Moneyball and Moneyball 2 for my past ramblings on this subject). The draft value chart says the #1 pick is worth 3000 points, a number that is almost certainly more than it should be, which means that the Browns should be able to make a trade similar to any of the deals in the table below. Note that in recent years teams have given up more than 3000 points, so the first pick may even be overvalued beyond what is shown below. The table shows the #1 pick from each of the last four drafts, and the players who were picked at positions equal to the value of that #1 pick. Pro-Bowlers are marked with an asterisk(*), and I’ve highlighted trades that I judged as “good” in green, “great” in bold green, “poor” in red, and “terrible” in bold red.

Trade #1 pick (3000 points) for:
  #2 (2600)
#50 (400)
#3 (2200)
#21 (800)
#4 (1800)
#12 (1200)
#5 (1700)
#10 (1300)
#6 (1600)
#8 (1400)
2016 Draft: #1 pick – Jared Goff
  Carson Wentz (QB)
Nick Martin (C)
Joey Bosa (DE)
Will Fuller (WR)
*Ezekial Elliott (RB)
Sheldon Rankins (DT)
Jalen Ramsey (CB)
Eli Apple (CB)
Ronnie Stanley (OT)
Jack Conklin (OT)
2015 Draft: #1 pick – *Jameis Winston
  Marcus Mariota (QB)
Ronald Darby (CB)
Dante Fowler (DE)
Cedric Ogbuehi (OT)
*Amari Cooper (WR)
Danny Shelton (DT)
*Brandon Scherff (OT)
*Todd Gurley (RB)
*Leonard Williams (DE)
*Vic Beasley (OLB)
2014 Draft: #1 pick – *Jadeveon Clowney
  Greg Robinson (OT)
Jeremiah Attaochu (LB)
Blake Bortles (QB)
*Ha Ha Clinton-Dix (S)
Sammy Watkins (WR)
*Odell Beckham, Jr (WR)
*Khalil Mack (LB)
Eric Ebron (TE)
Jake Matthews (OT)
Justin Gilbert (CB)
2013 Draft: #1 pick – Eric Fisher
  Luke Joeckel (OT)
Jonathan Bostic (LB)
Dion Jordan (DE)
*Tyler Eifert (TE)
Lane Johnson (OT)
D. J. Hayden (CB)
*Ezekiel Ansah (DT)
Chance Warmack (G)
Barkevious Mingo (DE)
Tavon Austin (WR)

I realize that teams rarely possess two high first-round draft picks, and thus that the table above is purely theoretical, but it gives an idea of how out-of-whack the valuation on the top pick actually is. In the past four drafts, the results of making the hypothetical trade would have been:

  • great: 3 times
  • good: 5 times
  • about even: 9 times
  • poor: 0 times
  • terrible: 3 times

There are exceptions – in 2012 everyone in the world agreed that it would be insane to pass on Andrew Luck – but this draft doesn’t have a can’t-miss quarterback, so if someone offered a deal I’d look at the numbers above and almost certainly make the trade, rather than risking everything on the hope that Myles Garrett won’t be the next Courtney Brown.

Moneyball 2

Posted from Culver City, California at 4:08 pm, Saturday, April 30th, 2016

Even though this topic may only be of interest to me, here’s the follow-up now that the 2016 NFL Draft is complete and the Browns have actually traded away their #2 pick. In a series of trades, they first gave the #2 pick to Philadelphia for a king’s ransom of picks that included the #8 pick, then traded that #8 pick to Tennessee for another bounty. Short summary: math won.

Pick HOFer Quality Starter Starter Occasional Starter Substitute Bench Never played
Browns trade:
#2 (1st) 0.17 0 0.67 0.17 0 0 0
2017 4th 0 0 0.05 0.20 0.45 0.05 0.25
#176 (6th) 0 0 0.04 0.04 0.22 0.39 0.30
Total 0.17 0 0.76 0.41 0.67 0.44 0.55
Browns receive:
#15 (1st) 0 0 0.50 0.50 0 0 0
2017 1st 0 0 0.50 0.50 0 0 0
2017 2nd 0 0.25 0.25 0.42 0.08 0 0
2018 2nd 0 0.25 0.25 0.42 0.08 0 0
#76 (3rd) 0 0 0.07 0.27 0.53 0.07 0.07
#77 (3rd) 0 0 0.07 0.27 0.53 0.07 0.07
#100 (4th) 0 0 0.05 0.20 0.45 0.05 0.25
Total 0 0.50 1.69 2.58 1.67 0.19 0.39

The result above is far better than in the example trade with San Francisco that I previously analyzed, and according to their historical drafting results should give the team two good players, another 2-3 decent players, and 1-2 guys who can occasionally contribute. Given the fact that the football gods hate the Browns, the two players drafted by Philadelphia (Carson Wentz) and Tennessee (Jack Conklin) will probably go on to become the greatest ever at their positions, but until that happens the statistical analysis says this was a really impressive result for the new Moneyball regime.

In addition to the two big trades, the Browns made three smaller trades, and overall turned ten draft picks into sixteen. Obviously quantity does not equal quality, but in this case the math says they got good value and, while they aren’t going to be very competitive for at least a couple more years, there might actually be some reason for optimism in Cleveland again.

Moneyball

Posted from Culver City, California at 2:04 pm, Saturday, April 16th, 2016

The Browns recently hired Paul DePodesta, whose story of bringing an analytical approach to baseball was chronicled in the book/movie Moneyball. Seeing as this hiring provides an opportunity to combine two of my favorite things – the Browns and math – I decided to make my own attempt to play football Moneyball. Since the NFL draft is the most obvious place where statistics can be applied, and since I’m a huge dork, I put together a spreadsheet of all Browns draft picks from 1999-2014, and used a formula based on career starts, Pro Bowl appearances, and Pro Football Reference’s “career value” rating to put each player on a seven point scale from “Hall of Famer” to “Never played”:

Pick HOFer Quality Starter Starter Occasional Starter Substitute Bench Never played
#1-5 (1st) 1 (17%) 0 4 (67%) 1 (17%) 0 0 0
#6-10 (1st) 0 1 (25%) 1 (25%) 1 (25%) 1 (25%) 0 0
#11-20 (1st) 0 0 1 (50%) 1 (50%) 0 0 0
#21-31 (1st) 0 1 (17%) 1 (17%) 2 (34%) 2 (34%) 0 0
#32-46 (2nd) 0 3 (25%) 3 (25%) 5 (42%) 1 (8%) 0 0
#47-61 (2nd) 0 0 1 (12%) 3 (38%) 3 (38%) 1 (12%) 0
Round 3 0 0 1 (7%) 4 (27%) 8 (53%) 1 (7%) 1 (7%)
Round 4 0 0 1 (5%) 4 (20%) 9 (45%) 1 (5%) 5 (25%)
Round 5 0 0 0 4 (25%) 5 (31%) 4 (25%) 3 (19%)
Round 6 0 0 1 (4%) 1 (4%) 5 (22%) 9 (39%) 7 (30%)
Round 7 0 0 0 0 7 (39%) 5 (28%) 6 (33%)

Teams who hold a draft pick in the top five value that draft pick as if it is certain to produce a Hall of Famer, but of the six top-five draft choices the Browns have made, Joe Thomas was the lone great pick, four of the others never made a single Pro Bowl, and the sixth (Braylon Edwards) made one Pro Bowl in his only good season and was traded after three years. Drafting in a position where they expected to find great players, the Browns instead came away disappointed five out of six times.

Since the Browns have done so poorly drafting in the top five, trading back needs to be a consideration. When considering whether to make a trade during the draft, the accepted way to “value” a draft pick in the NFL is the draft value chart. An alternate approach is to use analytics to determine the expected value of a trade, and that approach is more likely than the draft value chart to support trading back to get more picks, stating that the draft value chart over-values high picks. With the huge caveat that my table above is admittedly too small of a sample size to be fully accurate – it should include the draft history for all NFL teams, not just the Browns – here’s my attempt to use math to show why the Browns should listen to analytics and trade back.

The Browns have the #2 overall pick in the 2016 draft, which the draft value chart says is worth 2600 points. Theoretically the Browns could trade with San Francisco and get San Francisco’s first round pick (#7 – 1400 points), second round pick (#37 – 530 points), third round pick (#68 – 250 points), and second round pick in the 2017 draft. Based on what Tennessee just got from the LA Rams the haul would probably be even higher, but for the sake of argument let’s assume that’s the deal. Using the Browns draft history table above, and assuming San Francisco’s 2017 second round pick is in the top half of the second round, here’s what the odds of each of those picks panning out look like, based on past drafting history:

Pick HOFer Quality Starter Starter Occasional Starter Substitute Bench Never played
Browns trade:
#2 0.17 0 0.67 0.17 0 0 0
San Francisco trades:
#7 0 0.25 0.25 0.25 0.25 0 0
#37 0 0.25 0.25 0.42 0.08 0 0
#68 0 0 0.07 0.27 0.53 0.07 0.07
2017 2nd 0 0.25 0.25 0.42 0.08 0 0
Total 0 0.75 0.82 1.36 0.94 0.07 0.07

By giving up one player who the odds say is most likely to end up as a regular starter the Browns get four players and have excellent odds that one of them is a future Pro Bowler while another turns into a regular starter. The math seems clear: you make that trade.

With all of the above said, it’s the Browns, so expect to see them throw analytics out the window on draft night and pick another quarterback that they can then cut after 3-4 mediocre years.