"Football Outsiders Almanac" will be available for download by the time you read this. As usual, it is packed with insights, information, predictions, fantasy advice, jokes, geek-chic references and enough numbers to bury an accounting firm. It also contains four chapters by yours truly, so consider this an official plug.

The Football Outsiders gang is celebrating its (our) 10th anniversary this summer, and NFL stat analysis has come a long way in the last decade. The analysis is deeper, more widespread and more accepted by teams, the media and fans. Football stats have not earned the cachet that baseball stats have earned, and they probably never will, because the sports are just too different. But everyone from fantasy drafters to general managers take a closer look at the numbers these days, and that's mostly a good thing.

Unfortunately, the rise in high-tech analysis can make you, the casual fan, feel like you need overclocked spreadsheet software, a degree in economics and a knack for awkward acronyms to discover your own statistical insights. That's not true at all! You don't need to create something as elaborate as Football Outsiders' DVOA metric to get more analytic bang for your buck out of the stats you find at NFL.com or Pro Football Reference. All you need is a basic toolkit. Apply a few simple rules to the NFL's simple stats, and you will be excavating deeper meaning in no time.

Many of the rules listed below operate under the hood of statistics like DVOA, as well as the advanced metrics used at other websites. Some have been common knowledge among stat-heads for at least a decade. All of them will help you approach both pro football and its tables of numbers with more confidence. None of them require an advanced database. Just apply these rules to the widely-available numbers, and you will be on your way to better analysis, better predictions, and a level of expertise that will impress and annoy the guys at the bar on Sunday afternoon.

Rule One: Everything is Interconnected, Dude. The quarterback depends on his line. The receivers depend on the quarterback. The defense depends on the offense to not keep giving opponents the ball at midfield. Everything depends on the opponent, so AFC West numbers come with a different set of problems than NFC West numbers. In baseball, the pitcher and batter enjoy a degree of isolation from teammate effects. In football, about the only person who works in isolation is the kickoff specialist. More on him later.

The interconnectivity can lead to analysis paralysis. You hear it all the time in circular arguments. Tom Brady will be nothing without his receivers! His receivers will be nothing without Brady! This guy is a product of a system! This system only works when that guy is running it! And so it goes. Football statistics will never sort out these problems. Neither will scouting: Five experienced scouts can watch the same game film and come away with six separate opinions.

When performing NFL statistical analysis, you should always recognize that you are using an incomplete picture and trying to hit a moving target. That doesn't mean you should give up, but that we all must proceed with caution and humility, avoid jumping to conclusions, and be ready to revise opinions based on better evidence. In other words, we should follow good scientific practice. We will never know everything, but the next few rules show how easy it can be to know more.

Rule Two: Take Sacks Into Account When Evaluating Quarterbacks. Alex Smith was sacked on 9.9 percent of his pass attempts last year. That's a whopper of a percentage, and it is just as meaningful a statistic as his 70.2 completion percentage or 8.0 yards per attempt. In fact, ignoring his sack percentage makes the other figures highly misleading. Smith did not really complete a pass on 70.2 of his attempts, but 63.2 percent of them, if we count a sack as a failed attempt to pass (which is precisely what a sack is). Factor in the 137 yards he lost on 24 sacks, and his yards per attempt drop to 6.6. Those sacks turn Smith into an entirely different quarterback, one who lost his starting job, yet sack totals are usually tacked onto the end of a quarterback's stat line, when they are provided at all.

Offensive lines have a lot to do with sacks, but then again, receivers have a lot to do with receptions, yards and touchdowns (Rule One). Quarterbacks have more impact on their own sack totals than most fans realize. Kevin Kolb's sack rate last year was 12.9 percent, while teammate John Skelton's was 6.9 percent. Kolb has become the object lesson for sack percentage, so let's pick on someone else. When Brady Quinn replaced Matt Cassel last year, the Chiefs' sack rate went up from 6.4 percent to 9.6 percent. Quinn dropped back to pass 80 fewer times than Cassel but still managed to suffer two more sacks. Colin Kaepernick replaced Smith, of course, and the 49ers' sack rate fell by one third: Kaepernick had the same number of official attempts as Smith, but he was sacked only 16 times to Smith's 24.

Kaepernick's scrambling did not have as much to do with that difference as you might think. Scrambling quarterbacks are usually sacked more often than pocket passers, for a variety of reasons (some sacks are really aborted scrambles, scramblers hold the ball longer and take more risks, scramblers are usually younger and less experienced, etc.). The five quarterbacks with the best sack percentages in the league last year -- Eli Manning (3.4 percent), Peyton Manning (3.5 percent), Drew Brees (3.7 percent), Matthew Stafford (3.8 percent) and Tom Brady (4.1 percent) -- are all somewhere between reluctant to and incapable of running. The seven Super Bowl rings on that list highlight the importance of sack percentage as a measure of quarterback quality, and the presence of both Manning brothers show that sack percentage is more tied to the quarterback's style and ability than to his line, coaches or other circumstances.

Philip Rivers lost 311 yards to sacks last year, Eli just 136. The difference between their sack rates amounts to more than 175 yards, however. About 38 percent of sacks occur on third down, so players like Smith and Rivers lose that yardage in critical situations. Last season, about 15 percent of sacks resulted in a fumble (including those that the offense was lucky enough to pounce upon), so a high sack rate is likely to lead directly to a high turnover rate.

Add it up, and sack totals and percentages are incredibly important in quarterback evaluation, yet they are often overlooked or only discussed in extreme situations like Kolb's. Whenever seriously evaluating quarterbacks, make sure you invite sack data to the party.

Rule Three: Massage Your Rushing Data. Let's perform an experiment. We will start with an awful rushing team. This team rushes 20 times for 70 yards every week, averaging 3.5 yards per carry, a rate that would have ranked 31st in the NFL last year (the Cardinals were worse at 3.4 yards per carry). But something fluky happened in this team's season opener: A running back got loose on a 21st carry and reeled off a 90-yard touchdown. Let's examine how this team's total rushing data looks on a week-by-week basis:

Game Rushes Yards Avg.
1 21 160 7.62
2 41 230 5.61
3 61 300 4.92
4 81 370 4.57
5 101 440 4.36
6 121 510 4.21
7 141 580 4.11
8 161 650 4.04
9 181 720 3.98
10 201 790 3.93
11 221 860 3.89
12 241 930 3.86
13 261 1000 3.83
14 281 1070 3.81
15 301 1140 3.79
16 321 1210 3.77

 

The NFL rushing average last season was 4.26 yards per attempt. Our awful rushing team does not fall below average until after Week 6. By the midpoint of the season, it is still above the four-yards-per-carry boundary that serves as a red flag for many of us when we search for bad rushing teams. Even at the end of the season, this ineffective rushing team has a per-carry average that rounds up to 3.8. That 90-yard run is still providing a 7 percent edge to the team's rushing statistics, even though it has been plugging around at a dreary clip for months.

Of course, we cannot just erase 90-yard runs from a team's record, because long runs can be the repeatable result of a big-play running game. Cue Chris Johnson! He had runs of 94, 83 and 80 yards last season; those three runs accounted for 15.2 percent of the Titans rushing production. Johnson is a boom-or-bust back, and it is unfair to take away his boom. At the same time, trimming these big runs away and seeing what's left can be instructive. In Johnson's case, 986 yards and 3.6 yards per carry were left last season -- not as bad as you may have expected or as good as the Titans would prefer. In many cases, particularly when making evaluations in mid-year, chopping a big run or two away can create a swing of half a yard per carry and 10-15 slots on the NFL leader boards.

Another hunk of meat that can use statistical tenderizing is scramble statistics. Aaron Rodgers rushed for 259 yards and 4.8 yards per carry in 2012, and only a few of those runs were designed plays. (Football Outsiders lists six Rodgers sneaks or bootlegs.) Rodgers added 0.1 yard per carry to the Packers rushing average, and the figure was higher in the middle of last season, before the Packers sorted out their running game somewhat. Michael Vick is a notorious source of statistical noise: He raised the 2011 Eagles' per-carry average from 4.5 to 5.1, not with any significant read-option production, but by running for his life. Scramble yardage is really misplaced passing data, and in extreme cases it gums up our ability to evaluate rushing offenses. (As for college football, where sack yards are considered rushing yards lost, all I can say is grrrrr.)

There is no one right way to account for the fact that a few long runs or a bunch of scramblers can result in misleading rushing data. At Football Outsiders, they prorate long runs to reflect the fact that the only difference between a 60-yard run and an 80-yard run is often that one rusher started at his own 40 and the other at his own 20. That's the high- tech approach. The low-tech approach is to examine the "longest" column in the rushing tables, tease the numbers a little and be always on the lookout for rushing data that is not what it seems.

Rule Four: Remember Rules One, Two, and Three When Evaluating Offensive Lines. There are many scientific tools for evaluating offensive lines. Football Outsiders uses metrics like Adjusted Sack Rate and Line Yards, and then supplements the data with lots of tape and film study. Other stat websites have similar methods. If you want to draw your own conclusions but lack the math skills to create custom stats or the time and stomach to watch a Raiders-Chiefs game a dozen times, you can rely on easy-to-find statistical standbys like a team's sack totals and rushing production, IF you remember the caveats listed above. Quarterbacks like the Mannings can make their offensive lines look good, sack prone quarterbacks make their lines look bad, a breakaway run can skew rushing data and so on.

Rule Five: Targets are on Target. Targets (the number of times a receiver is thrown to) are not yet an official stat, but they are widely available. Ten years ago they were the Holy Grail of receiver data, but now they can be found all over the Internet. Targets provide lots of information in one easy-to-understand number.

Last year, Larry Fitzgerald was targeted 156 times but caught just 71 passes. Those two numbers sum up the Cardinals season: A great receiver got stuck watching passes by Ryan Lindley-types sail this way and that. The Broncos' Brandon Stokley was targeted just 59 times but caught 45 passes, a remarkable 76 percent catch rate. That's what happens when a legendary quarterback is reunited with one of his favorite possession receivers. Stokley did not have the highest catch rate among receivers in the NFL: Randall Cobb topped him, with 80 catches among 104 targets (77 percent). As the Packers receiving corps transitions from the Greg Jennings-Donald Driver era to a Cobb-Jordy Nelson-James Jones era, that high catch rate is a reason to believe that Aaron Rodgers won't miss a beat.

Targets provide an incredible amount of information about the shape of a team's passing game. The table below shows the primary receivers targeted by the Carolina Panthers, a team whose passing game is usually oversimplified as a Cam Newton Must Do Better headline:

Player Targets Catches
Steve Smith, WR 138 73
Greg Olsen, TE 104 69
Brandon LaFell, WR 76 44
Louis Murphy, WR 62 25
Mike Tolbert, FB 39 27
Jonathan Stewart, RB 23 17
DeAngelo Williams, RB 20 13

 

The target data offers a deeper understanding of the Panthers passing game than we get from reception totals. Right away, we zero in on Murphy, who caught two fewer passes than the fullback despite 60 percent more targets. Murphy's targets show what a black hole the number three receiver spot was for the Panthers, a hole they will try to fill with free agents Ted Ginn and Domenik Hixon (plus returnees Joe Adams and David Gettis) this year. We see how important Olsen is and that Tolbert had a bigger role than most modern fullbacks have in an offense, and we get confirmation of just how important Steve Smith is to the Panthers. The targets tell us a little about Newton: He needs to be more accurate so he can get Smith the ball more than 52 percent of his throws, but an upgrade over Murphy's 42 percent catch rate would certainly make Newton's life easier.

Target data, like all data, needs interpretation. Deep receivers like Mike Wallace (119 targets, 65 catches) will catch a lower percentage of their passes than possession receivers like Stokley. The lone truly good receiver in an offense, like Fitzgerald or Smith, is going to have a lot of incomplete passes forced to him. Running backs catch about 70-75 percent of the (mostly very short) passes thrown to them, receivers 55-65 percent and tight ends some percentage in between, so it is important to adjust when switching positions. The bottom line is that evaluating receivers based only on their catches is almost like comparing quarterbacks based on only their completions. The incomplete passes tell the other half of the story, and that half is often more important than the half we normally see.

Rule Six: Quarterback Rating is a Pretty Good Stat … For Defenses. When analytics types get together and play Bang-Marry-Kill with statistics, quarterback rating gets almost unanimous "Kills." It's a bad stat, for reasons I love to write about, and most football analysts have built some kind of improvement over the odd little relic of late 1970 actuarial reasoning that still sits at the end of each quarterback's stat line.

But here's the thing: as an analytic tool for defenses, quarterback rating against isn't bad. It ignores the biggest problem that plagues defensive passing stats: great defenses give up extra passing yards while protecting leads, while bad defenses or teams get a stat boost when opponents squat on the ball in the fourth quarter. Quarterback rating bundles yards per attempt, completion percentage and interception rate into a handy package; it may not be the most scientifically valid package, but fans recognize it, and it is sitting on Pro Football Reference for the taking.

The Cardinals led the NFL in quarterback rating allowed at 71.2 last season, and since we have been picking on them all through this article it is time to praise them. They had a nasty pass defense in 2012, though it started to come unglued after one too many 19-16 and 17-3 losses. The Bears, Seahawks, Packers, Falcons and Niners came next, a mix of interception-happy defenses and stingy ones, all of which reached the playoffs but the Bears.

The Chiefs, Eagles, and Raiders had the worst quarterback ratings allowed. Those teams finished 24th, 21st and 13th in yards allowed, thanks to the "sit on the ball" phenomenon, so the rating exposed three brutally bad defenses that looked roughly average according to raw stats. All of the coaches and defensive coordinators at the bottom of the list lost their jobs; most of the coordinators at the top either kept theirs or became hot commodities. The Cardinals must replace Ray Horton, now with the Browns, with Todd Bowles, one of the people who coached the Eagles to the bottom. Thus concludes the portion of this essay in which we praise the Cardinals.

If you have to grab one statistic to evaluate a pass defense, rating allowed is a pretty good one. Just remember that it suffers from the same problem we encountered in Rule Two: no sacks.

Rule Seven: Tackle Totals are a Minefield of Meh. Tackle totals are now official, but they are viewed with suspicion, for good reason. Did you know that three rookies finished in the top 10 in tackles: leader Luke Kuechly of the Panthers, Bobby Wagner of the Seahawks and Lavonte David of the Bucs? Probably not, because no one takes tackle totals seriously. Did you know that two Jaguars linebackers are among the top 15 tacklers in the NFL: Paul Posluszny and Russell Allen? Again, you probably did not.

Tackle totals contain a ton of noise and a little signal. The league leaders are usually pretty good linebackers on pretty bad teams, like Kuechly and Poz. The leaders gobble up lots of routine tackles while opponents control the clock. Think about it: A good defense forcing a three-and-out probably creates two tackle situations, assuming the third down play is an incomplete pass. A bad defense allowing 10-play touchdown drives probably creates seven or eight tackle situations. Extended across a 16-game season, you get a massive distortion in which the capable starters on terrible teams make far more tackles than the All-Pro defenders on great defenses.

That said, good information can be extracted from tackle totals, once you know what you are looking for. Defensive linemen register far fewer "opportunity" tackles than linebackers and safeties. J.J. Watt led all linemen with 81 total tackles last year, with Lamarr Houston and Muhammad Wilkerson next on the list with 69 each. Houston and Wilkerson were on the field a lot, but even so, their tackles came close to the line of scrimmage and generally represent "fine plays," not "I happened to be on the field for the fourth quarter of a 38-14 loss and tackled the running back five times as he plopped on the clock."

For cornerbacks, tackle totals are basically a booby trap. You can probably guess why: Many cornerback tackles represent completions that the cornerback allowed, making tackle totals a tally of failures, not success. It's not always that cut and dry. Antoine Winfield tied Cortland Finnegan for the league lead among cornerbacks with 101 tackles. Winfield has a well-established rep as a great run supporter, and he registered 42 tackles on running plays, meaning that his high tackle total is at least partially a reflection of his talent (Finnegan made 30 run tackles). On the other hand, only 12 of Cary Williams' 75 tackles came on running plays. He got picked on a lot, gave up a lot of receptions and earned a fairly high tackle total as a result. The best cornerbacks produce tiny tackle totals: Deion Sanders' hovered in the 30s in his signature seasons, while Darrelle Revis hovers in the low 50s when healthy. But evaluating players on the plays they don't make comes with its own problems.

As for the linebackers, all of the players listed in the first paragraph of this segment are solid players. The rookies are great prospects. A player like Wagner, on a team where tackles are hard to come by, really stands out when you know how to look at the data. (So does NaVorro Bowman, who finished second in the NFL in tackles for a 49ers defense that rarely kept opponents on the field very long.) For a player like London Fletcher, whose numbers included 139 tackles, three sacks, five interceptions and 16 passes defensed, the tackles provide a good indicator of the day-in, day-out heft of his contribution.

Rule Eight: Kickoff Data is the Most Important Data You Never See. Want to make 500+ yards of football production disappear immediately? Ignore kickoff data.

Justin Tucker of the Ravens was one of the best kickoff specialists in the NFL last year. His 87 kickoffs traveled a league leading 5,938 yards, producing 49 touchbacks. Jason Hanson, beloved veteran sharpshooter on short field goals, was one of the worst kickoff specialists in the NFL last year. His 87 kickoffs yielded 5,414 yards and 30 touchbacks. Both teams allowed about 23 yards per return, but the Ravens had to stop one less kick return per game, and with the ball about 5.3 yards deeper at the end of a typical Tucker kick than a typical Hanson kick, the Ravens forced opponents to start a lot more drives around the 20-yard line than the Lions did. Five hundred yards is the production a team gets from a slot receiver or a change-up back, but because in this case the yardage resulted from kickoffs, it is often completely ignored during team analysis.

Everyone focuses on field goals, of course. We are probably looking in the wrong place when it comes to evaluating both kickers and a team's ability to win games with special teams. The NFL field goal percentage is now 83.9 percent; for attempts shorter than 50 yards, it climbs to 87.9 percent. Unless you are evaluating a rookie kicker, a 40-year-old who may have just lost his leg or some emergency replacement off the waiver wire, you are probably just looking at random fluctuations when an established kicker has a field goal slump or streak. (Think of an 88-percent free throw shooter in the NBA suddenly missing a few or going 15-for-15, except that a kicker's streak extends across two months, not two games.) The kickoff data does not fluctuate nearly as much, so someone like Tucker can produce more touchbacks than other kickers year after year and have a significant impact on his team's defense, and never get noticed.

Rule Nine: When in Doubt, Leave it Out. If a statistic has been broken down too far, it is probably useless (fourth quarter first down rushing data on the road, from the I-formation, during a full moon). If a number looks like a one-time fluke, it probably is. Week 1 stats are too tiny and distorted a sample to use. "Clutch" stats (two-minute drills, fourth-quarter-of-close-game-breakdowns) are really a cobbling of a handful of plays taken from radically different game situations: There is a big difference between leading the Jets by four and trailing the 49ers by seven, so why would you clump a quarterback's numbers from those situations together?

Sometimes, conventional wisdom is actually wise. It's not hard to push, pull and prod numbers to reach oddball conclusions like "Peyton Manning is average." The reason for any statistical study should be to find the truth, not to prove a point. If you have developed a statistical method that "proves" the Cardinals had a great offense or Jake Locker was an efficient passer or something, you have proven only that your methods are flawed or biased. If you publish your results, then defend them with nasty online stat rap battles, you will contribute to the attitude among some old-timers that NFL stat analysis is a bunch of malarkey. A cigar really is a cigar, 2,097 rushing yards really is 1.18 miles and great teams and players usually have great stats and vice versa. If your analysis completely bucks the story told by both the regular stats and the opinion of the masses, quintuple check your analysis first.

Rule 10, Which Should Probably Be Rule One: WATCH THE GAME TAPE. We finish with a true story from 2011. The Buffalo Bills defense began the season with 12 interceptions in their first five games. A non-stat-y colleague contacted me asking for a statistical interpretation of what was going on: How were the Bills doing it, and what impact would their turnover rate have on the balance of power in the AFC East?

Before hitting the spreadsheets, I watched all 12 interceptions. With NFL Game Rewind and the play-by-play data available at NFL.com and elsewhere, finding a dozen important plays and watching each one a few times amounts to a $70 investment (for the whole season) and about an hour's work. It turned out that many of the Bills interceptions were tipped passes, a few of them by offensive players. The Bills defense made some great plays, but they were also getting especially lucky. Unless the Bills had developed a new strategy for catching balls that bounce off an opponent's fingers, there was no way that their interception rate was sustainable or had anything to do with a newfound ability to confuse quarterbacks.

Sure enough, the Bills intercepted just eight more passes in their final 13 games. Statistical analysis might have yielded wacky conclusions, but an hour of game tape saved my colleague and I from some bonkers interpretation of what really happened.

There's a misconception that stat analysts don't watch games. In fact, stat analysts watch far more game tape than most conventional football writers, who have a very different job to do, and much, much more than most fans. Every number on a spreadsheet begins as an event on the field, and the best analysis marries statistics with lots of tape grinding, rather than citing irreconcilable differences.

So watch the games. But since there are no games for several weeks, check out Football Outsiders Almanac. And of course, keep checking back here, because Sports on Earth is about to ramp up our NFL training camp coverage. I have got my paws in the dirt, scratching. Get ready for an attack with an enthusiasm unknown to mankind.