Sports Media Analysis

While compiling all of the NFL season predictions and Week 1 picks from writers across the country late last week and over the weekend, there were two common themes. One of these themes was IT’S ABOUT TIME FOOTBALL SEASON IS HERE.

The other repetitive phrase I remember reading constantly was “Rookie QB + Rookie head coach = Opening Day loss.” Of course, if there were two teams in the NFL who would be most likely to succumb to a rookie coach and QB pair, the Lions and Bengals would have to be pretty high on the probability meter. Nonetheless, the thrashing of the Lions by Michael Turner and company and the Ravens’ win at home against the Bengals were just two of the surprises from Week 1. I’ll have the complete analysis up on Wednesday or Thursday, but it’s safe to say that the Panthers and Bears were, according to the so-called experts, the biggest surprises of the opening weekend.

I’ve also put together a tally of the predictions by the sportswriters around the country on which teams will be in the playoffs, which teams will be in the Super Bowl and which players figure to take home some hardware at the end of the season.  That should be up within the next couple of days as well.  It’s safe to say that Tom Brady’s injury will wreak some havoc on most predictions.

By Jacob Crawford

Because I was surprised at how poorly ESPN.com’s Chad Ford performed on his 2008 mock NBA Draft compared to his peers when we did our analysis of 28 mock drafts, I decided to take a look back at Mr. Ford’s final mocks over the last few years to see if this year was an aberration or the norm. I didn’t know this until I did the research, but Mr. Ford only began the mock drafts prior to the 2005 NBA Draft – the year the Hawks stuck a knife in their stomach by passing on Deron Williams and Chris Paul. Therefore, we only have four mock drafts for our data set – which would not be considered a large amount of information – but it’s enough to make some conclusions.

Using the same methodologies as we used in the 2008 NBA mock Draft analysis, we can quickly summarize Mr. Ford’s performance in the 2005 – 2008 mocks. As you can see from the table below, Mr. Ford’s mock drafts from both 2006 and 2007 were excellent, while his mocks from 2005 and 2008 were rather ordinary.

You’ll notice I added some data on the international draft picks. I have a hunch that as the international talent pool erodes (more on that some other time) predicting when international players will be selected is going to be more difficult. Anyway, as you can see from the data there, Chad Ford’s 2008 mock NBA draft was not nearly as impressive as his 2007 and 2006 versions. As a point of reference, the pick difference analysis on Mr. Ford’s 2007 and 2006 mock drafts showed a more accurate mock draft overall than the 2008 mock NBA draft from SI.com’s Ian Thomsen, who had the best 2008 mock according to our analysis. Of course, there is a chance that the 2007 and 2006 drafts were more predictable, but we would not be able to conclusively tell that unless we examined other mocks from 2007 and 2006.

So… from the last four NBA Drafts, Mr. Ford has provided what appear to be two excellent mock drafts, most likely better than a vast majority of his peers, as well as two pretty poor mock drafts. Overall, Mr. Ford does not seem to be any better or worse at predicting the draft than other national writers, team beat writers at newspapers or other basketball writers on the Web.

Click here to see the complete data analysis of Mr. Ford’s mock drafts.

Click here for the complete analysis of the mock drafts: 2008 NBA Mock Draft Analysis

Our first analysis will be around the 2008 NBA Draft. We gathered 28 mock drafts from 27 different outlets, focusing mostly on the newspapers in the home media markets of the NBA teams as well as major national sports Websites like ESPN.com and SI.com. Additionally, the cutoff time for the mock drafts we analyzed was the morning of this year’s draft – Thursday, June 26, 2008. That evens the prognosticating field between newspaper writers like the folks at USA Today, the New York Sun and Detroit Free Press and folks like ESPN.com’s Chad Ford who actually updated his mock draft twice on the day of the draft.

METHODOLOGY

Only the first round of the NBA Draft was used for this analysis – primarily because that’s the maximum depth of most mock drafts. Initially, we only examined the number of first round picks that were guessed correctly. Under this piece of analysis, the only thing that matters is how many of the first round picks the sportswriter correctly guessed in the correct spots. For example, let’s say sportswriter A’s first five picks were:

1) Derrick Rose
2) Michael Beasley
3) Brook Lopez
4) Russell Westbrook
5) JJ Hickson

In this scenario, sportswriter A earns three points, one for each correct guess. Let’s say sportswriter B’s first five picks were:

1) Derrick Rose
2) OJ Mayo
3) Michael Beasley
4) Russell Westbrook
5) Kevin Love

Sportswriter B would also receive three points, one for each correct guess. The major drawback of this analysis is that it does not account for the prognosticator slotting the players in the correct range. The writers who picked like sportswriter B should be given more credit than sportswriter A because they not only picked several players exactly as it played out during the draft, but when they did get a pick wrong they were not far from the actual order.

To account for this issue, we also added an analysis for the average amount of distance between where the prognosticators picked players and the actual draft slot from which they were chosen. For example, if we used the sportswriters A and B examples from above, sportswriter A picked his players on average 4.2 draft slots away from where they were actually taken, even though he got three of his guesses exactly right. The seven-spot difference between where he picked Brook Lopez at #3 and his actual draft slot of #10 plus the difference between where he picked JJ Hickson #5 and Hickson’s actual draft slot of #19 was 21. Spread across five picks, that is an average of 4.2. Sportswriter B, however, picked his players on average 0.4 picks away from where they were eventually drafted.

And instead of weighting every single pick in the first round the same, we broke down the picks into these categories: Top 10 (picks 1 – 10), Lottery (1 – 14), Second 10 (11 – 20), Final 10 (21 – 30) and Total (1 – 30). Essentially, every single pick is weighted twice, with the first 14 picks known as the Lottery portion of the draft were weighted thrice. This weighting was done for both the correct pick analysis and the pick difference analysis.

Let’s look at two sportswriters from the mock draft pool: Krista Jahnke of the Detroit Free Press and Tim Povtak of the Orlando Sentinel. From the 1st round of the NBA Draft, Ms. Janke and Mr. Povtak correctly guessed eight of the 30 picks, tying for 5th best amongst the 28 mocks we analyzed. Both Ms. Jahnke and Mr. Povtak correctly guessed six of the Top 10 picks, but Mr. Povtak correctly guessed one of the picks between #11 and #14 and one pick between #21 and #30 while Ms. Jahnke did not correctly guess anything between #11 and #20. While both ranked 5th overall for the first round with eight correct guesses out of 30 chances and 5th overall for the Top 10 segment of the analysis, Mr. Povtak had higher rankings in the Second 10 and Lottery segments of the analysis. Ms. Jahnke performed better on the Final 10 segment, but the sum of Mr. Povtak’s rankings for each segment (5 for Top 10, 5 for Lottery, 6 for Second 10, 6 for Final 10 and 5 for Total) was 27 (5+5+6+6+5) was better than the 36 posted by Ms. Jahnke (5 for Top 10, 7 for Lottery, 18 for Second 10, 1 for Final 10 and 5 for Total).

We used the same ranking system for the pick difference analysis, but Ms. Jahnke was quite a bit more accurate on the pick differences than Mr.Povtak; accurate enough to overtake him in the overall analysis. In short, we summed the rankings from each of the different analysis. This allows us to identify the sportswriters who were most consistently top performers in both correctly guessing the picks but also identifying the ranges the different players would be drafted in.

WINNERS

Sports Illustrated’s Ian Thomsen claimed the overall top spot after ranking 4th in the correct pick portion of the analysis and topping the pick difference analysis. Closely trailing Mr. Thomsen were Rick Bonnell of the Charlotte Observer, Sam Smith of Sporting News, Jonathan Feigen of the Houston Chronicle and Krista Jahnke of the Detroit Free Press. The Orlando Sentinel’s Tim Povtak, MyNBADraft.com’s Todd Robinson, InsideHoops.com, the Chicago Sun-Times’ John Jackson and the Boston Herald’s David Moore rounded out the top 10. Although he performed well overall, Mr. Feigen deserves some criticism here for completely forgetting (I assume) about Eric Gordon. Not only did he not have Gordon going in the Top 10, as almost every other mock draft did, he completely left him out of the first round.

Mr. Thomsen’s win was fueled by his dominance in the Second 10 picks of the draft. He was the only sportswriter to correctly guess more than two of the Second 10 picks – Robin Lopez to the Suns at #15, Marreese Speights to the 76ers at #16 and JJ Hickson to the Cavaliers – and his picks in the Second 10 averaged only 3.1 draft slots away from where they were actually drafted. No other prognosticator had an average less than 5.0 during the Second 10. Additionally, Mr. Thomsen’s overall average pick difference for the 1st round was 3.667, the only average under 4.0. Finally, Mr. Thomsen finished in the top seven of every single category, with only Mr. Bonnell joining him as a Top 10 finisher in every category. Dominant indeed.

LOSERS

On one hand, I hate to highlight the people who did not perform particularly well on the mock draft, but that is exactly why it is important that someone do this analysis. At some point, we as readers have to trust our sources of information, and it only makes sense that we analyze them to find out who among them is consistently full of it. Much to my own personal surprise, two of my favorite NBA writers – Chad Ford and John Hollinger – were woefully bad on their mock drafts.

It’s not a huge surprise that Mr. Hollinger was off but I had always presumed that Mr. Ford would be at or near the top in an objective analysis such as this considering the amount of coverage he gives the NBA Draft. Mr. Ford was especially dreadful in the Second 10, in which he did not pick a single player correctly and he averaged a second-worst 8.2 draft slots away per player. In addition to picking two players in the Second 10 who were not drafted until the 2nd Round – Mario Chalmers and DeAndre Jordan – he slotted both Kosta Koufos and Donte Greene towards the top of the Second 10.

SPORTSWRITERS STRUGGLE WITH THEIR LOCAL TEAMS’ PICKS

Of the 12 newspaper writers who completed mock drafts only two – the Arizona Republic’s Paul Coro and Mr. Jackson of the Chicago Sun-Times – correctly picked their local teams’ 1st round pick. The Suns picked Robin Lopez at #15 and the Bulls selected Derrick Rose first overall. Given that 27 of the 28 writers in our analysis selected Derrick Rose first, it’s hard to make the case that Mr. Jackson really had some insight into how the Bulls were leaning. Mr. Coro, on the other hand had this to say when explaining his Robin Lopez selection for the Suns, “the Suns might focus on Lopez and Florida’s Marreese Speights. Lopez was a backup but is seen as a reliable, energetic big man good for rebounding and defense.”

The Toronto Star’s local team, the Raptors, did not have a first-round selection, so we can exclude them from the remaining analysis. The other nine writers seemed to do a pretty good job of at least selecting a player from the right position. For example, Marc Spears of the Boston Globe and David Moore of the Boston Herald, picked wing players DeVon Hardin and Courtney Lee in their hometown Celtics’ slot. The Celtics selected athletic G/F JR Giddens with their 1st round pick. Of the nine writers who missed on their local team’s 1st round selection(s), only Mr. Povtak of the Orlando Sentinel selected a player from a completely different position than his local team actually drafted. Mr. Povtak selected PF JJ Hickson in the Magic’s slot, but the Magic actually selected SG Courtney Lee.

Unfortunately, I do not believe much can be concluded from such a small sample of information, especially with only three of the 12 newspaper writers working in the media market of a team picking in the top 10 of the draft. Every mock draft we analyzed had more accuracy in the top 10 than in any other segment of the draft, so you can understand why the nine writers whose local teams were picking in the teens and twenties had difficulty getting their picks exactly right. That they were consistently able to get the positions right at least shows that they have some understanding of how their hometown team appears to be thinking.

DRAFT IS MORE PREDICTABLE AT THE TOP

It might seem obvious, but when we analyze how each player was predicted across the 28 mock drafts, the players projected towards the top of the draft more much easier to predict than the players taken in the middle 10 picks or last 10 picks of the 1st round. For example, the average mock draft slot for Derrick Rose was 1.04 – meaning he was picked first overall in 27 of the 28 mocks. Second overall pick Michael Beasley’s average mock draft slot was 1.96 – meaning he was taken second overall in 27 of the 28 mocks. Both Rose and Beasley were, on average, selected within 0.04 of their actual draft slot. That’s pretty damn accurate. Additionally, SG Eric Gordon, who was drafted 7th by the Clippers, had an average projection of 7.19 even though one writer – Mr. Feigen of the Houston Chronicle – left him out of the draft completely.

The only three players who were drafted in the Top 10 who had decent variances between their average projected draft slot and their actual draft position were Russell Westbrook, Danilo Gallinari and Brook Lopez. The sportswriters thought Westbrook (taken 4th; average projection was 6.75) and Gallinari (taken 6th; average projection was 9.00) would be drafted a little later, they thought Lopez would be taken earlier (taken 10th; average projection was 6.14).

According to the pundits, the major losers of the draft were DeAndre Jordan and Chris Douglas-Roberts, both of whom slipped from either unanimous 1st round selections (Jordan) or near unanimous (CDR) 1st round selections to 2nd round picks. Although CDR was selected in the 1st round in 27 of the 28 mocks, his average pick selection was still 25.70. Despite that low average, being picked 40th in the draft means the difference between his average mock draft slot and his actual slot was 14.70. Jordan was the only player who was selected in a vast majority of the mocks to have a larger difference between his perceived value and his actual value, with a difference of 16.79.

In addition to Jordan and CDR, two Kansas players fell pretty far from their perceived value on draft day. Mario Chalmers fell to the 34th selection, 14.04 slots away from his average perceived value of 19.96, while Darrell Arthur fell to #27, 9.50 spots from his perceived value of 17.50.