I am of the opinion that every writer, analyst and expert covering any field should make specific predictions for which they should be held accountable. As it stands now, writers in particular sometimes have little incentive to make accurate predictions. With no skin in the game, what's to stop them from simply making bold claims for the sake of drumming up controversy?
If all writers were forced to make very particular prognostications and then revisit those predictions to see where they went right and wrong, there would be a pretty strong incentive (pride and reputation) to get it right.
Most important, it would help readers understand who knows what they're talking about and whose content is full of fluff. It's really easy to look back on past events and analyze them after the fact, but it's an entirely different endeavor to put yourself on the line in predicting what will take place in the future. It's easier said than done.
This idea is why I make a number of very specific preseason predictions, as well as scrutinize the predictions after the season.
I say "at the time" because I think my 2013 predictions were even more accurate. I've listed the majority of them in this slideshow, both good and bad. I actually believe the predictions that go wrong are much more valuable because 1) I can determine if I indeed made a mistake (sometimes you can make a great prediction that simply doesn't pan out) and 2) I can tweak the models or numbers I use to make predictions to enhance future accuracy.
I posted an article called Why Tony Romo Will Limit Turnovers in 2013 over at NBC in which I predicted 12 picks for the quarterback.
So what makes me so confident that Romo will limit the turnovers in 2013? Well, his interception rate was really trending downward prior to last year. It was as low as 1.6 percent in 2009 and 1.9 percent in 2011—Romo’s last two full seasons prior to 2012. And Romo’s 2012 interception total of 19 was inflated by the fact that Dallas was down so often. That forced Romo to throw more passes, which will of course naturally increase his interceptions, and throw a lower quality of passes. In games like that against the Chicago Bears on Monday night, Romo tried to fit passes into tight windows in order to mount a comeback. In my opinion, he’s always been selfless like that; he’ll take the heat for throwing too many interceptions, even if the majority of them come in situations in which he could have easily tucked the ball to save his stats.
In 2013, I’m betting on the fact that 1) Romo won’t again attempt 648 passes and 2) his interception rate will regress toward the mean.
I was right. Romo threw only 535 passes on the season (which would still be well below 2012's mark even if he played in Week 17), and he ended up tossing only 10 interceptions.
At Bleacher Report, I also projected Romo's entire stat line:
387-for-580 (66.7 percent) for 4,582 yards (7.9 YPA), 32 touchdowns, 12 interceptions, 100.4 passer rating
Romo’s always been able to post top-tier numbers, but his legacy will come down to what he does in December and January. Nonetheless, he’s probably a good bet for at least 4,500 yards and 30 touchdowns. More important, Romo will almost assuredly toss fewer than the 19 interceptions he did in 2012.
Romo's final line was 342-for-535 (63.9 percent) for 3,828 yards (7.2 YPA), 31 touchdowns, 10 interceptions and a 96.7 passer rating.
I had a pretty strong opinion on running back DeMarco Murray in the preseason, giving four reasons he was primed for a breakout and an argument for why I was leading his hype train. Among the reasons was that we couldn't yet conclude he's injury-prone:
Murray might not even be “injury prone.”
A lot of what we view as “injury proneness” is just an illusion. Injuries are a low-frequency event and, for the most part, very random. That means we’d expect the distribution of injuries to be rather random as well, regardless of whether or not some players are more susceptible to injuries than others.
Nonetheless, we’d still expect a few players to be more susceptible to injuries than others. It makes sense that some people’s genetic makeup is such that they’re unlikely to get injured and/or likely to heal quickly after getting injured. But that doesn’t mean we can predict future injuries with any sort of accuracy.
Even if injury proneness does indeed exist, it would take quite a long time to discover whether or not a player is truly more susceptible to injuries than average. Imagine that the typical player has a 10 percent chance to get injured in a given season and an injury-prone player has a 20 percent chance to get hurt. Even if that’s the case, we’d still need a pretty substantial number of seasons to pass before we could claim with any sort of certainty whether or not a player’s injuries were because he’s injury-prone or if he just got unlucky.
Predicting injuries is kind of like projecting fumble recoveries. The events are rare, and thus susceptible to randomness. And while important, using past fumble recoveries (or injuries) to predict future ones is basically useless. That means Murray might be injury prone, or he might not, but the nine games he’s missed in two seasons really can’t help us make that determination.
That led me to predict that Murray would start at least 14 games. I actually thought there was an even better chance he wouldn't miss a single game, but Murray did indeed play in 14 contests.
One of my most unpopular predictions was that defensive end DeMarcus Ware was on the verge of breaking down. At Dallas News, I wrote that he was no guarantee to bounce back in 2013:
I tracked the historic production of pass-rushers in terms of approximate value—a good measure for overall value that incorporates tackles and sacks—and sorted it by age. I did the same for Ware for both his tackles and sacks.
The idea was to see whether or not Ware’s progression has been typical. The average pass-rusher develops earlier than most positions, peaking at ages 25 and 26. There’s a small drop in play around age 27—to around 90 percent of peak production—and most pass-rushers can maintain that level of play until around age 32 or 33.
Ware’s tackle totals over the past three seasons have been 66, 58, and 56. Based on his total snaps, Ware’s tackle rate has moved from 7.0 percent to 6.4 percent to 6.3 percent. Will that drop continue? Given his age, it’s likely. So at a projected 900 total snaps, Ware would record 55 tackles at a 6.1 percent tackle rate.
We can use the same sort of methodology with pressures to project Ware’s sacks. Ware’s pressure rate—the percentage of pass-rush snaps he’s reached the quarterback—has dropped a lot over the past three years; it was 10.8 in 2010, 9.2 percent in 2011, and only 6.8 percent last season. However, I think we’ll see that number move up in 2013; Monte Kiffin will do a better job than Rob Ryan of finding unique ways to get Ware going. I’ll project his pressure rate at 8.0 percent due to the scheme switch and simple regression toward the mean.
Ware rushed the passer 454 times in 2012, but he also dropped into coverage on 64 plays. Those coverage snaps will convert to pass-rushing snaps this year, so Ware will probably rush the quarterback around 520 times or so—a number that fits with other 4-3 ends around the league. At 520 snaps, Ware would total 42 pressures with an 8.0 percent pressure rate. I’ve shown in the past that sacks typically add up to around one-quarter of all pressures, meaning Ware’s most likely sack total in 2013 is probably somewhere around 10.5.
Final 2013 Projection: 55 tackles, 42 pressures, 10.5 sacks
Even though I was told my projection for Ware was ridiculously low, it turns out I was actually too optimistic. Ware ended the year with 40 tackles, 34 pressures and 6 sacks.
If my Ware projection didn't strike a nerve, my forecast for tight end Jason Witten was sure to do the trick. I had some pretty pessimistic views of Witten throughout the preseason and regular season, due primarily to his age and the fact that his 2012 numbers were incredibly inflated due to an unusual workload.
Witten’s efficiency has plummeted.
On a per-catch basis, it was pretty evident that Witten regressed in 2012. He averaged just 9.4 yards per reception—down 2.5 yards from 2011. Part of that is probably due to the short routes that he ran; the average pass to Witten was just 8.3 yards in length.
However, if Witten’s decline in yards per reception was due solely to shorter routes, we wouldn’t expect his yards per route to decline. That’s because he’d catch a higher percentage of his targets and total more catches (which he did) and yards. But take a look at Witten’s yards per route over the past five seasons:
Notice a trend? With fewer targets, Witten is going to see his bulk stats drop significantly.
Witten's yards-per-route dipped AGAIN in 2013 to 1.58.
I projected Witten's final 2013 stats on numerous occasions:
In projecting Witten’s numbers, we need to appropriately determine his opportunities. He was targeted 150 times in 2012—by far the most of his career—after averaging 123 targets in the prior three seasons. Due to game situations, Dez Bryant’s emergence, and new faces in Dallas, we’re likely to see Witten’s targets drop precipitously to, say, 125.
Witten hauled in 72.5 percent of his targets over the past four seasons. Due to his natural regression as a receiver, that mark will likely decline a bit. If Witten catches 70 percent of his 125 targets, he’d bring in 88 passes in 2013. And even though Witten isn’t very explosive anymore, he’ll still probably average more than the 9.4 yards per catch that he posted in 2012. At 10.0 yards per reception, Witten would total 880 receiving yards.
Witten has never been an efficient red zone receiver, converting just 5.5 percent of his career catches into scores. That number might actually decline in 2013 due to the presence of Gavin Escobar, Terrance Williams, and other red zone threats. We can probably project Witten to beat his 2012 touchdown total of three, but not by much.
Final 2013 Projection: 88 receptions, 880 yards, 4 TDs
He landed in the end zone more than I figured, but Witten's final line of 73/851/8 was still unexpected by most.
At NBC, I posted a final stat line prediction for wide receiver Dez Bryant:
Bryant is a touchdown machine because he can score from any spot on the field and, due to his stature, he’s a big-time red zone threat. Over the course of his three-year career, Bryant has taken 13.5 percent of his receptions into the end zone. That’s an outstanding mark, but I think it could increase this year. The primary reason is that the Cowboys should be in the red zone more often, so a higher percentage of Bryant’s looks will come in scoring situations. At a 15.0 percent touchdown rate, Bryant’s most likely touchdown total would be 14.
Final 2013 Projection: 92 receptions, 1,411 yards, 14 touchdowns
Bryant finished the season with 93 catches, 1,233 yards and 13 touchdowns. I was off on Bryant's yards per reception by just a bit, but this projection was about as close as one could hope.
Here at Bleacher Report, I predicted who would win a number of awards for Dallas. Here is who I selected:
MVP: QB Tony Romo
Rookie of the Year: C Travis Frederick
Most Improved Player: CB Morris Claiborne
Biggest Surprise: DE George Selvie
Biggest Disappointment: TE Jason Witten
Offensive Player of the Year: RB DeMarco Murray
Defensive Player of the Year: LB Bruce Carter
This is subjective, of course, but overall, I think these predictions were quite accurate. I'd argue Romo, Frederick and Selvie were all the clear winners of their respective awards. Ware and Murray have competition for "Biggest Disappointment" and "OPOY," but those two players are at least in the discussion for their respective awards.
I missed badly on both cornerback Morris Claiborne and linebacker Bruce Carter's anticipated improved play, and I think those two represent how you can view the same result in different ways.
Looking back, I think I made a really poor prediction on Claiborne; I typically don't prefer cornerbacks without elite length, but I couldn't get off of the fact that Claiborne was a top-six pick just a year prior. Surely the scouts couldn't be that wrong on him, I thought. Oops.
Meanwhile, I still believe Carter has elite potential. Perhaps there's a mental factor at play that can't be entirely captured by measurables, but I'll still predict a bounce-back year for him in 2013.
Verdict: 5 CORRECT, 2 DEAD WRONG
Right tackle Jermey Parnell accompanied cornerback Morris Claiborne, left tackle Tyron Smith, and safety Barry Church as my four breakout candidates for Dallas. Here's what I had to say about Church:
At 6-1, 222 pounds, a good comp for Church is Seahawks safety Kam Chancellor. That’s especially true because Seattle runs a defense that’s very similar to what Monte Kiffin is going to bring to Dallas in 2013 – more zone concepts (although less Cover 2 than people think). That means Church will be in the box quite often. The safety recorded a 4.17 short shuttle before entering the NFL, suggesting he has plenty of short-area quickness to thrive in that area.
I also projected Church with numbers no one saw coming, and he ended up surpassing them:
If we view Church as a player with a skill set similar to Seahawks safety Kam Chancellor (who stands 6-3, 231 pounds with 4.62 speed) playing in a comparable scheme, then it follows that his stats should probably be similar. In 2012, Chancellor recorded 69 tackles in his third season in the NFL—down four from 2011. That’s a good baseline for Church, but I think we can bump it up just a bit based on his success as a run defender and the fact that Dallas will probably play more defensive snaps than Seattle did in 2012.
Chancellor saw 56 targets playing in the box in 2012, and Church will likely check in near that as well. It could be a slightly lower figure since the Cowboys will probably play some more Cover 2 than Seattle—meaning Church will be in the deep half.
Final 2013 Projection: 50 targets, 38 receptions allowed, 350 yards allowed, 80 tackles, 2 INT
Church finished the year with 45 targets, 35 receptions allowed, 339 yards allowed, 133 tackles, 1 INT. Money.
Verdict: 2 CORRECT, 1 DEAD WRONG, 1 DIDN'T PLAY
Projecting Smith to the Pro Bowl was a pretty straightforward prediction that I made right here. He was indeed selected to play in the game.
Smith is just 22 years old. He’s entering his third season in the NFL at an age when many players are just coming into it as rookies. Cowboys rookie wide receiver Terrance Williams will be 24 before Smith is 23, for example. That’s amazing, and it really goes a long way in trying to project Smith’s potential breakout. Offensive tackles typically have very long careers, but it takes them a few years to come into their own. Smith’s age is such a positive because, given his experience at 22, he’ll have more years playing at peak efficiency.
I think this prediction that Dallas's running game would improve significantly from 2012 was a really, really easy one to make and a pretty good indicator that most writers have little comprehension of basic statistics.
The Cowboys were so poor on the ground in 2012 that it was foolish to predict that they wouldn't improve, even if only a little, as many did. I cited those 2012 numbers and a change in running game philosophy under Bill Callahan as reasons the Cowboys would have much better rushing efficiency.
Everyone knows that the Cowboys’ running game needs to improve. Most equate that to sticking with the run, although that will likely just result in the team getting down early and, ironically, being forced to pass. Instead, the team really just needs to rush the ball more efficiently, especially in crucial situations like short-yardage, goal line, and late in games.
Speaking of short-yardage runs, they could actually be one of the reasons the Cowboys ranked so low in rushing efficiency in 2012. Dallas ranked 31st in both attempts and rushing yards, which is to be expected given how often they were down in games, but they were also just 31st in efficiency at 3.6 YPC. That’s a poor number no matter how you slice it.
However, it’s probably a bit skewed by the situations in which Dallas ran the ball. They kept the ball on the ground a lot in short-yardage and other low-upside situations, so their efficiency was bound to be relatively low. That’s one of the reasons that a stat like YPC can be really misleading; it doesn’t account for game situations. It values a two-yard run on third-and-one the same as a two-yard run on first-and-10, but there’s obviously a major difference there; one is a successful play, and one is not.
That’s why the best metric to judge rushing success is the aptly named ‘success rate.’ Success rate measures the percentage of plays that result in an offense increasing their chances of scoring. So the metric rewards a two-yard run on third-and-one, as it should. It’s a better gauge of rushing success than YPC because it accounts for such situations.
The Cowboys' rank in YPC jumped from 31st in 2012 to eighth in 2013.
This was really more of a long-term prediction, so we'll need to take a wait-and-see approach, but for now, it appears the Cowboys did indeed make a mistake on running back Joseph Randle.
After the draft, I listed some statistical reasons the Cowboys shouldn't have selected Randle and argued that both Lance Dunbar and DeMarco Murray were much superior options:
Murray is a far more talented player than Randle. Let’s take a look at the numbers:
Murray: 6-0, 215 pounds, 4.41 40-yard dash
Randle: 6-0, 204 pounds, 4.63 40-yard dash
Randle is a much leaner, slower back. That’s not good. Both have an upright running style, so if we’re going to label that as the cause of Murray’s injury woes, what does that say about the lighter Randle?
When it comes down to it, Murray should be the feature back in the Cowboys’ offense, racking up around 70 percent of the total touches. Randle can spell him, perhaps on third down and in short-yardage situations, but the Cowboys are best off with Murray leading the way in 2013.
Randle averaged 3.0 YPC in his rookie year.
Verdict: CORRECT SO FAR
Perhaps the prediction of which I'm most proud is that defensive end George Selvie will end up being an outstanding player:
If Selvie were coming out of college as a rookie, he’d represent the perfect opportunity to acquire value by exploiting a marketplace inefficiency. See, Selvie is “short” for a defensive end at just 6'3", and NFL teams as a whole still seem to “pay” for height; they draft tall pass-rushers because they’re tall, not (always) because they have long arms. But if you’re paying for a trait in height that doesn’t actually help in the NFL, you’ll eventually be disappointed with the results.
We see a similar phenomenon with quarterbacks and height. As I detailed in my article on predicting quarterback breakouts, NFL teams still very much favor tall quarterbacks because they believe the passers need height to “see over the line.” I think most quarterbacks would tell you that they don’t actually look over the offensive line, but rather through lanes, but height is indeed correlated with success for quarterbacks. However, it’s probably not the cause of the success, or at least not to the degree that NFL teams think. Instead, hand size actually seems to be more important for quarterbacks, allowing them to control the football and throw it accurately. Taller quarterbacks usually have larger hands, obviously, leading to the illusion that quarterbacks must be tall.
That means that shrewd NFL teams can acquire value by actually searching for shorter quarterbacks who have oversized hands. That way, they can generate value by emphasizing a predictive trait that the rest of the market is overlooking. The Chargers and Seahawks found some success with this strategy with Drew Brees and Russell Wilson – two short quarterbacks with massive hands.
In the same manner, teams should search for defensive ends who are actually slightly undersized but have the same trait that allows taller players to thrive: long arms. At just 6'3", Selvie doesn’t have what NFL teams consider to be “ideal” height. But he does have incredibly long 34.5-inch arms.
So why does Selvie have just three sacks in three years in the NFL? Let’s not forget that he was a seventh-round pick who dropped, at least in part, due to his height. As a late-rounder, Selvie hasn’t really had much playing time; he rushed the passer on 203 snaps in his rookie season and just 169 since then. In comparison, DeMarcus Ware rushed the quarterback 454 times last year. So Selvie doesn’t even have a full season of snaps under his belt.
What’s more likely? Selvie has generated only three sacks because he’s a poor pass-rusher, or the defensive end with the most tackles for loss in NCAA football since 2000 has underachieved on a limited number of NFL snaps? I’m choosing the latter.
I predicted Selvie would finish the year with 12 sacks. He had seven.
Verdict: MOSTLY CORRECT
One of my worst predictions was that the Cowboys' pass rush would improve under defensive coordinator Monte Kiffin. Here's how I predicted 39 sacks for Dallas:
If Ware rebounds in 2013, the Cowboys should once again be a top 10 pass-rushing unit. It might seem like a big jump from 21st to 10th, but the 10th-ranked team in terms of total sacks last year had 39—just five more than the Cowboys. That total should remain steady in 2013 because the distribution of sacks doesn’t change much, so if the ‘Boys are going to rank in the top 10 in sacks, we need to figure out how they can make up for those five.
The most obvious answer seems to be that Ware will bounce back. That’s certainly a possibility, but I wouldn’t necessarily count on it. Ware is entering his age 31 season—a time when many pass-rushers break down—and he had only 32 pressures last year. I’ve shown before that pass-rushers typically convert around one-quarter of their pressures into sacks, so Ware was actually lucky to record 11.5 sacks. His most likely total was eight.
We can also look at the Cowboys’ pressure total as a unit to see if they got unlucky in regards to sacks. The entire defense had 142 pressures in 2012, according to Pro Football Focus. Based on that, their most likely sack total was 35.5. They checked in just below that, so it’s not like they were extremely unlucky. Still, a reversal of luck alone accounts for 1.5 sacks.
So for the ‘Boys to reach 39 sacks, we need to find out how they’ll record 3.5 more. That number is the equivalent of 14 pressures, and I think it’s very possible that the defense adds that many this year. First, Ware’s pressure total was too low. Even though he got lucky to have 11.5 sacks, he’ll probably have around 10 extra pressures in 2013. He averaged well above that number in prior seasons.
Spencer’s pressure total of 29 may or may not increase, but the main area where you’ll see superior play from Dallas is in the interior defensive line. The defensive tackles will play a one-gap scheme in which they can shoot up the field to make plays. With just four more pressures from them, the Cowboys can easily reach the magic total of 39 sacks—and a top 10 finish.
You could argue that the loss of Spencer would have affected my projection, but in all honesty, it wouldn't have. I was so high on Selvie that I really didn't see a drop from Spencer (in terms of pass-rushing ability).
Dallas finished the year with 34 sacks (pretty close), but ranked 25th in the NFL.
At NBC, I projected a pretty sizable increase in total points for Dallas in 2013. Here was one reason:
Superior Field PositionThe primary reason that Dallas will score more points is that their defense is bound to give them more help. Last season, the Cowboys’ defense hauled in only seven interceptions and forced a turnover at the third-lowest rate in the league. That would have improved in 2013 regardless of Rob Ryan’s departure, but it’s especially likely with Monte Kiffin. If the Cowboys can acquire more short fields, they’ll be able to put points on the board much faster. They ranked seventh in the NFL in yards per drive in 2012; if their point total matches that, they’ll score in the neighborhood of 425 points in 2013.
I predicted a 49-point jump for Dallas from 2012 (up to 425 points), and they delivered with 439 points.
I also predicted the Cowboys would improve their point differential to at least +25. Because of the horrific defensive play, it ended up being just +7.
VERDICT: 1 CORRECT, 1 WRONG
Similar to the rushing game prediction, it was relatively easy to project more takeaways for Dallas since they were so poor in 2012. I used stats to project the Cowboys at 17 picks and 11 fumble recoveries:
To predict the Cowboys’ interception total, let’s start up front with the pass-rush. Dallas ranked only 20th in sacks and 23rd in pressures in 2012. There’s good reason to think the pass-rush will improve this year with Kiffin in town, but what can we realistically expect from Dallas?
Odds are they’ll have right around a league average pressure rate, meaning we can use the NFL average for picks, 15, as our baseline. Given the Cowboys’ switch to more zone coverage, their emphasis on undersized pass defenders, and the quality of their cornerbacks, we can probably bump up that projection just a bit. We’ll call it 17. For the record, that would have ranked them 11th in the league last year.
Fumble recoveries are extremely random because, once a fumble occurs, it’s really difficult to predict who will end up with the football. But forced fumbles aren’t that random. There’s a reason we see certain teams and players – Chicago Bears cornerback Charles Tillman comes to mind – rank high in forced fumbles. There’s a skill to it.
Last year, the average team forced 16.4 fumbles and recovered 10.3 of them. That’s a 62.5 percent fumble recovery rate for defenses, meaning they’re slightly more likely than the offense to land on the ball. Considering the nature of Kiffin’s defense, we can probably project the Cowboys to increase their forced fumble total in 2013. We’ll say a small jump to 18. If that’s the case, the most probable result would be right around 11 fumble recoveries.
The Cowboys finished the year with 15 interceptions and 12 fumble recoveries, good for 27 takeaways overall.
Like I said, predictions are important because they display comprehension. I'm so confident in my ability to predict Cowboys and NFL-related outcomes that I will issue an open challenge to any Cowboys writer at a major website/newspaper who thinks they can beat my predictions.
We'll figure out the details before the start of the 2014 season, but if you legitimately feel as though you can out-predict me and you're willing to make those predictions public, please message me on Twitter (@BalesFootball).
If I lose, I'll probably go ahead and just dismiss it as variance anyway.