Defending Reality

Photo: Brace Hemmelgarn, USA Today Sports

Join the Footballguys Daily Update

Start your morning with our roundup of the most important stories in football - with the fantasy insight you need to make league-winning decisions. Delivered straight to your inbox, 100% free.

Subscribe Now

Nearly every Fantasy League has different rules for how to score Team Defense. Many use the tried and true method of 1 point for a sack and 2 for a turnover. To add a little variety, many also add 6 points for the rare touchdown and yet another two for the rarest of scores - a safety. Other leagues try to tweak this scoring method by incorporating points against, yards against, or both.

The basis for fantasy scoring comes from an attempt to quantify an individual player's performance numerically and assign that a value proportional to his team's performance. Touchdowns and yardage usually translate to team success, and the offensive player is rewarded for such productivity.

The second iteration of fantasy scoring came about as an attempt to normalize different positions to a similar scoring system. Quarterbacks produce more touchdowns and yardage than running backs, who produce even more than wide receivers. The result for many leagues was to reduce the points for passing touchdowns and also passing yardage so as to make running backs more valuable. Other leagues go one step further by awarding an extra point per catch to each player to increase wide receiver and tight end values closer to running backs.

What does any of this have to do with Team Defense? I am glad that you asked that question. This article is an attempt to determine how to quantify Team Defense scoring in a manner that reflects the impact a defense has on the outcome of a game, and also to provide a normalized score for a Team Defense that puts the value of a Defense at or near par for other fantasy football positions.

First, we have to determine what constitutes a good Team Defense. Is it one that gives up the fewest yards, or the fewest points? An argument can be made for either being the case, so let's take a look at the rankings from last year to see which method more accurately reflects a successful season. The results for the 2012 Season are in Table 1:

Table 1: Baseline Defense Rankings

Upon further review of Table 1, for of the Top 5 teams in Points Against were in the 2012 playoffs. Even both Super Bowl teams (Baltimore, San Francisco) and the rest of the Final Four teams (New England, Atlanta) were better in points against than in yardage.  The Ravens, Patriots and Falcons were in the Top 12 in points against but at or below average in yardage as well.  Just like in the past, it appears that Points Against is a better indicator of a good defensive team than just looking at the yardage. We shall adopt Points Against as the baseline for Team Defense.

The most common scoring system for Team Defense awards a point for every sack. This seems like a good idea, since it is an accomplishment by the defense to stop the offense from moving downfield, and it is an easy statistic to track. However, how realistic is this as a measure of Team Defense? Do sacks truly translate to team victories?

We obviously need some way to test this idea. Turning to statistics, we find that correlation is a measure of how two groups of statistics relate to one another. The formula used for correlation gives an answer between 0 and 1, with 1 representing a perfect match - 100% correlation. We can use this to see if sacks line up with our baseline, the Baseline Ranking (Points Against) from Table 1.

Table 2 lists the Team Defense rankings for sacks and the Baseline Ranking from Table 1. The correlation factor is given at the bottom of the table.

Table 2: Sacks vs. Baseline Defense Rankings

Correlation: 0.542

Based upon the results, it would appear that we are on the right track with some good correlation, right? Well, looking back over the years we have for data, the numbers are really all over the place. Table 3 shows the results going back to 2004:

Table 3: Sacks vs. Baseline Defense Rankings 2004-2012

This translates to a wide range of answers, anywhere from a weak correlation (17.7%) between Sacks and Baseline Defensive Ranking to a strong correlation (62.6%). Why would that be the case? The quick answer is that the sack numbers are tightly grouped together. The range of sacks for 25 of the 32 teams is just 28 to 44 sacks, or just one per game. Results this close together will lead to correlations to other data that are unpredictable and all over the map.

So what is wrong with using sacks anyway? The common misconception is that a sack translates readily to the defense getting the ball back for the offense. The reality is that not all sacks are created equal. While a 15-yard sack on 3rd-and-10 would likely result in a Team Defense getting the ball for the offense, a 1-yard sack on first down is not nearly as valuable. However, in the 1 point for a sack scoring system, there is no differentiation between the two different sacks. It would appear that the sack statistic is misleading, and should be replaced by another one.

Perhaps we should look harder at the other common statistic used in Team Defense scoring, the turnover. Surely one cannot dispute that turnovers relate strongly to team victories. That has to correlate well with Team Defense rankings, right?

Table 4: Turnovers vs. Baseline Defense Rankings

                                                                                       Correlation: 0.442

OK, that got worse than the previous method (sacks vs. points against).   How can turnovers not significantly contribute to or improve the correlation for a good defensive ranking?

The answer really comes from the mathematics - correlation does not work well with numbers that are close together. From Table 4, you can see most Team Defenses (25 of 32) have from 16 and 31 turnovers, so such a tight grouping will compromise the calculations.

A sanity check is in order - we need to use the common scoring method in its entirety (1 point per sack + 2 points for a turnover) and correlate that result against Average Points (See Table 5):

Table 5: Turnovers and Sacks vs. Baseline Defense Rankings

Correlation: 0.517

Now we're seeing some improvement. The correlation between the basic scoring system and Points Against is certainly better than the turnover ranking correlation alone and is closer to the first score with sacks. The combination of both statistics helped in achieving point separation (ranging from 53 to 129 points) and agreement with the baseline, but there is definitely room for improvement.

The basic argument against using the sack as a standard measure remains - there is little direct correlation between the sack and elite defenses. Another measure of defense should be considered in place of sacks. However, if sacks are going to be eliminated from the Team Defense scoring system, what will be inserted in its place?

Let us reconsider the basic premise of Team Defense. Defenses have two main objectives - keeping the opposition from scoring, and getting the ball back for their offense. We have already seen that the first objective is the baseline measure of Team Defense, so we need to quantify the second criteria to the best of our ability to see if this should be the new fantasy football measure of the performance of a Team Defense.

There are two defensive categories of getting the ball back for the offense that are overlooked in the "turnover" category. Defenses succeed in stopping the opponent by forcing the traditional turnover (fumbles and interceptions) and also by forcing punts and stopping the opponent on fourth down. Our new formula for Team Defense needs to have basis in reality to weigh the value of forced punts and turnovers on downs.

Punts happen numerous times during the game, but they rarely result in the defense giving the offense a short field (under 50 yards from a touchdown). As such, forcing a punt has to be viewed as less opportunistic and less valuable than a fumble or interception, which results in a short field about half of the time. Therefore, our formula begins to look like this:

New Team Defense Score = Turnovers (INTs and Fumbles) x 2 + Forced Punts

Now for the second portion of the new formula - incorporation of turnovers on downs. While this can happen anywhere on the field, it is more likely to occur at both a crucial point in the game and also in a position where the opposing team is in scoring territory. While the resulting field position may not be as good as with a traditional turnover, the impact of both getting the ball for the offense and the likelihood that the 4th down turnover kept the opposition off the scoreboard gives this type of turnover approximately equal value to a fumble or interception. Therefore, we modify the formula to be:

New Team Defense Score = Turnovers (INTs + Fumbles + 4th Down Stops) x 2 + Forced Punts

We label this new equation the "Realistic Team Defense" scoring system. Now all that is left is to test our new formula.

Let us revisit the 2012 season for data once again. Obtaining the two new statistics (4th down turnovers and forced punts), we get the following results:

Table 6: All Turnovers and Forced Punts vs. Baseline Defense Rankings

Correlation: 0.635

Definite progress. The correlation between the new and improved scoring system and Points Against is over 63% (63.5%), a sign of a strong correlation and the best correlation so far. This also begins to align well with prior years as shown in Table 7:

Table 7: Reality Defense vs. Baseline Defense Rankings 2004-2012

The addition of all turnovers and punts forced has increased the point separation once again, ranging now from 134 to 245 points (a range of 111 points instead of 76 as before in Table 5). For the past nine seasons, only one season was below 60% correlation – 2011.  The smaller correlation in 2011 than in previous years is a reflection of a smaller range from the top to the bottom teams that year. In 2010, the range in Reality Defense was 137 to 226 points, a difference of 89 points.  In 2012, the range is also big – 134 to 245 – comparable to the 2010 numbers.   It appears that there were fewer turnovers in general in 2011, which compresses the range – yet Reality Defense correlation is the best of all the measures of defense year after year.

There are additional benefits to this new equation. First, the scores are higher (average score of 11.1 / game) than under the original system (5.4 / game), which goes more towards a better normalization of the Team Defense position on the fantasy roster. By increasing the average score, the net effect is that a Team Defense is now representative of another normalized position player. Table 8 represents the 2012 average score by position of the Top 12 players based upon the default Footballguys.com scoring system.

*Based upon the new Realistic Team Defense formula.

Table 8: Average 2012 Fantasy Points / Game for Top 12 Players by Position

Now the Team Defense is comparable to the value of close to a tight end or kicker and just below a Top 12 wide receiver, Top 12 RB or QB. With PPR, the Team Defense declines to a good RB2 or WR2 or an above-average TE2 in value. That would seem to be an appropriate position of value for a strong Team Defense, below a top skill position (QB, RB, WR, TE) but above the lesser valued RB2, WR2, TE2 or kicker position.

One last additional benefit (and variance to the Realistic Team Defense formula) is that the addition of scoring points for a Team Defense scoring play (Touchdown or Safety) allows for the added point value, and also reduces the impact of that event to a lower percentage of the total Team Defense score. Previously, under the original scoring method, teams scored between 53 and 129 points for the season without defensive or special teams touchdowns added (see Table 5). Adding a single touchdown (6 points) varied the overall season score by 5-10%, a large impact. Now with the Realistic Team Defense formula, the range is much larger (134 to 245, see Table 6) so adding a touchdown reduces the impact to 3-5%. This smaller valuation relative to the seasonal performance does seem to be more appropriate.

Adoption of the Realistic Team Defense formula for defensive scoring for fantasy leagues going forward would result in a more accurate representation of the value of a Team Defense and better reflect how the actual defensive unit for each team performs in that particular season. The formula incorporates the significant statistics to quantify how well an actual defense performs, and results in a normalized score relative to the skill position players. The variation of adding back in the relatively rare event of a defensive score reduces the impact of the additional points to less of an overall change to the season Team Defense total, and increases the relative value of a Team Defense closer to that of an upper echelon wide receiver.

One last comment – I get asked this question often over the years.  There are several league hosting sites that can support this scoring format, and I suggest that you ask each site whether they can implement these statistics (forced punts, 4^th down stops).  I am positive that MyFantasyLeague.com can score defenses this way, for example, but I do not want to steer anyone towards any particular hosting site.

As always, questions, suggestions and comments are always welcome to pasquino@footballguys.com.

www.nfl.com

www.footballguys.com