Saints are the 3rd best team in the NFC, at least according to one statistical, points-based way of looking at it.  I’m talking about the SRS (Simple Ratings System), as seen at

Overview of SRS

So, what exactly is SRS?  Good question.  The SRS rankings are based entirely on points scored and points allowed.  It ignores win/loss records, yardage stats, turnover stats, etc.  It begins with MoV (margin of victory).  MoV is simply (points scored – points allowed) / games played.  Let’s say a team scored 100 points and allowed 91 points in 3 games played.  The team would then be +9 in point differential (100 – 91 = 9).  If we divide the point differential (9) by the number of games played (3) we get an MoV of +3.  The team has outscored its opponents by an average of 3 points per game, another way of putting it.  (100-91)/3 = 3, simple enough to figure out MoV for any team.  Saints currently have MoV = 3.8, which is points scored (93) – points allowed (78) divided by games played (4): (93-78)/4 = (15)/4 = 3.75, rounds to 3.8.

While MoV is very straightforward and simple to understand, SRS is more complicated (despite the “simple” in the name).  SRS = MoV + SoS.  In other words, a team’s rating is based on its margin of victory combined with its strength of schedule.  If we only used MoV, we’d be ignoring the quality of the opponents.  So, in order to build up the complete SRS for each team we need to know not only the MoV for each team, but also the SoS for each.  SoS would be the average SRS rating of a team’s opponents.  Problem is, we don’t know the SRS before we know the SoS because SRS = MoV + SoS, and we don’t know SoS until we know the SRS for each team because SoS is the average SRS of each opponent.  We have this weird strange loop self-referencing thing going on with this.   SRS depends on SoS, which depends on SRS.

More than you really want to know about SRS

So, how do you get around this self-reference problem?  One way is to do it in iterations, beginning with some assumed value (example: begin the first iteration with the assumption MoV = SRS).  So, the first pass through you calculate each team’s SRS as it’s MoV (known) + its SoS (assumed to be opponents’ average MoV).  After that first pass all of the teams’ SRS’s will have changed (except for any that might have played a neutral strength of schedule where all its opponents MoV’s averaged to 0).  So, we do another pass, this time using MoV + (average SRS of each opponent — as opposed to average MoV for each opponent).  After this pass, the numbers have all changed again, but by a lesser amount.  We keep going through these iterations until the SRS numbers stabilize and stop changing (at least stop changing to some digit somewhere off the right of the decimal).  Typically, this might take a few hundred iterations to reach a desired level of stability (example, no SRS value changed more than 0.000001), but it’s child’s play for a modern computer using a good spreadsheet program, such as Excel or with a computer program written specifically for this purpose.

I once wrote and maintained a Java Applet (NFLPicker, now defunct) that attempted to handicap NFL games against the point spread.  I used something very similar to SRS in that applet.  I took average MoV for each team, then added average Mov of each of its opponents, and then added to that the average MoV of all the opponent’ opponents.  I don’t remember the exact weighting I gave to each, but I know MoV was most important, followed by opponents’ average MoV, followed by opponents’ opponents’ average MoV.  I reasoned if you keep going far enough eventually everybody played everybody and all the average opponents’ opponents’ opponents’ opponents’ MoV’s would average to 0, so I stopped at that 3rd level deep.  (By the way, NFLPicker was not very good at making predictions.)

Save yourself a headache and skip right down to here

Okay, so, in a nutshell, SRS = MoV + SoS.  Simple enough, it’s the (MoV) margin of victory adjusted by the (SoS) strength of schedule, which said SoS is merely the average SRS of each of the team’s opponents.  Pfref gives us OSRS and DSRS in addition to SRS, MoV, and SoS numbers.  Here, OSRS is simply offensive SRS and DSRS is simply DSRS.  So, these values focus on a team’s offensive output (points scored for OSRS) and defensive output (points allowed with DSRS).  One final formula before we look at the rankings: SRS = MoV + SoS = OSRS + DSRS.

Problems with SRS

No ranking system is perfect, and SRS certainly has its own flaws.  First of all, it doesn’t take into consideration things that are really important.  For example, it doesn’t look at win/loss records.  It treats a 30 point win and a 2 point loss (+28 net total) the same as 2 14-point wins (+28 net total).  I don’t know about you, but I’d rather win 2 games by 14 points each than win 1 by 30 and lose 1 by 2.  It doesn’t take into account garbage time scores.  In week 1 the Vikings had a comfortable lead late in the game, so they basically let the Saints march down the field and get a meaningless touchdown.  But that meaningless touchdown counts as part of the Saints’ SRS all the same.  It doesn’t take into account some fluke scores.  Take for example, the Monday night game where Washington has to try a very risky multiple lateral play, the ball is fumbled (almost inevitably on those type plays), and is picked up and run in for a meaningless (for the purposes of who won the game, not so meaningless for gamblers) score.  If it’s a tie game before that play, no way Washington tries it.  They probably take a knee or just do a standard hand off and play for OT.

Saints SRS ranking

Okay, I know I haven’t been very succinct and most of you reading this are actually no longer reading this, so here is how the Saints rank using this system.

Saints are 3rd in the NFC with SRS = +7.5
Saints are 5th in the NFC in MoV = +3.8
Saints are 4th in the NFC with SoS = +3.8
Saints are 4th in the NFC with OSRS = +5.2
Saints are 5th in the NFC with DSRS = +2.3


Does this mean the Saints really are the 3rd best team in the NFC?  Probably not.  We’ll have to wait and see whether the Dr. Jeckyll defense of weeks 1 and 2 shows up or the Mr. Hyde defense of weeks 3 and 4 shows up the rest of the way.  Still, even though the SRS system is not without its problems, it does give us a fairly simple (in a complex form of simplicity) to come up with an *objective* method of ranking teams.  Some of the issues with SRS can be addressed, for example, by awarding +3 points bonus for each win and -3 penalty for each loss, which would factor in the win/loss records.  You could also cap the blowout wins such that even if a team won by 30, it only gets credit for a +10 win (and -30 blowout losses only count as -10 losses).  Lots of things you can do to address deficiencies in the system, but each one adds subjectivity (how many points should a win be worth?) and certainly complicates the process even more.  I guess that’s why they call it simple.