Simple NFL Systems # 18 – Early Season Match-Ups and Pythagorean Win
Similar to the Playoffs, early season games need to be handicapped much differently than those played after around Week 3 of the regular season – when it comes to using situational methods.
One of the biggest challenges at this stage of the season comes from the lack of useful data from recently played contests.
Unfortunately, pre-season games have never been a good indicator of what lies in store for a team in the first few weeks of the regular season. The large number of players that see playing time in the pre-season who are ever cut or relegated to 2nd or 3rd team status does not help, nor does the fact that marquee QB's and other important players often only take the field for aful of plays in the earlier games, if at all.
All is not lost; however, as there are key stats from the previous season that can lend real insight into games played in the early part of the following year, and there is also the previous history between the 2 teams involved that one can consider (more on this later) .
One key stat from the past season that works very well as a handicapping tool in the early part of the following one is Pythagorean Win Percentage .
Pythagorean Win Percentage (PWP) was first developed by sabermetrics-pioneer Bill James as a method of removing the effects of 'luck' from a baseball team's won / lost record by focusing solely on runs for and runs against . The formula actually works equally well for the National Football League after a few minor 'tweaks', the most significant of which includes using points for and Against in its calculation as opposed to 'runs'.
By comparing a team's PWP with their actual winning percentage, it becomes easy to ascertained which teams have had an over-abundance of either good or bad misfortune – knowledge which has obvious implications for those of us trying to handicap current games based on past performance.
PWP, as it applies to Major League Baseball, has seen a number of improvements since James first came up with the idea and more advanced formula's now not just runs themselves, but also the ratio of singles, doubles, homeruns etc. that went into producing these runs, along with alternate multipliers depending on the different ball-parks where the scoring occurred.
Some of these improvements do not apply so much to the game of North American football, where the field of play is obviously identical from stadium to stadium, and the original formula that James developed for MLB remains a simple, yet accurate method of calculating a team's winning percentage that is often more reliable than won / lost records alone.
The formula for calculating PWP for NFL teams is as follows:
Points For ^ 2.37 / (Points For ^ 2.37 + Points Against ^ 2.37)
An exponent of 2.37 has been found to provide the most accurate results for the NFL while 1.83 is the most commonly used exponent for MLB teams. This formula even works when applied to NBA teams, where an exponent of between 14 and 16 is prevalent.
In order to best explain exactly how this formula works, it's probably best to look at a couple of examples from the past season.
The New England Patriots are an example of a team who actually 'overachieved' in 2007, when their won / lost record of 16-0 is compared against their PWP.
Anyone who watched the Pats-Ravens game in Week 15 and to a lesser extent, their regular season finale against the NY Giants, would probably agree that New England could have easily ended the season at 15-1 or 14-2 and their PWP shows that either of these records would actually have been more indicative of their level of play in '07.
Based on their Points For of 589 and Points Against of 274 , New England's PWP works out to 0.860 (589 ^ 2.37 / (589 ^ 2.37 + 274 ^ 2.37)).
Given their WP to PWP differential of +0.140 (1.000 – 0.860) it appears that New England was in fact, luckier than most teams in the league last year-an opinion that Ray Lewis and the rest of the Ravens would certainly not argument.
An example of a team that underachieved in 2007 would be the Philadelphia Eagles, who finished the year at 8-8, yet, had a PWP of 0.567 (336 ^ 2.37 / (336 ^ 2.37 + 300 ^ 2.37)), leading us to believe that they were sometimes more deserved of a 9-7 record.
So, how does a team's PWP from the previous season figure into the process of handicapping games early in the next one?
One interesting use for this stat involves teams that had a PWP (as long as this meeting occurred within the past 4 years). Teams in this situation are a dismal 33-60 ATS (35.5%) since 1994 in the first 2 weeks of the regular season immediately following .
As I mentioned near the top of this article, past history between the 2 teams in question is important early on in the season and in this case, teams with a weak PWP from the previous season that are also facing an opponent that may be seeking revenge for a relatively recent defeat, creates a potent combination that has spelled trouble versus the line over the past 14 years.
While a situation with a record of 33-60 ATS is sufficient enough, there is one other Secondary condition concerning the past meeting between these 2 teams that when added, greatly reduces the number of games involved while maintaining a similar level of profit.
This condition concerns teams that not only won in the last meeting, but, did so in convincing fashion (at least offensively anyway).
When we only include teams that scored at least 30 points in this game , the record for this situation drops to a crushing 5-28 ATS (15.2%) for a tidy profit of $ 2,250.00 when wagering $ 110.00 to win back $ 100.00 against the team in question .
The final Secondary condition that I like to add to this trend involves something I touched on earlier, and that is, the comparison of a team's actual winning percentage with their Pythagorean winning percentage.
Teams that meet the criteria discussed so far that also had a SU winning percentage at least 0.100 points higher than their PWP last season have been a perfect 3-0 ATS, so, by eliminating teams that outperformed their PWP in the previous season, we are left with a trend that has been 2-28 ATS since 1994.
Here are all the details.
(Notes: ASMR stands for Average Spread Margin Rating A positive rating indicates a trend that is stronger than average versus the line negative negative than average. at one time or another. WT% is the percentage of teams that are .500 or better and SPR is the average spread for teams in this situation.
System # 18 Summary
Primary Conditions (Building Blocks)
1) Game is being played in Week 1 or 2 of the regular season.
2) Last Seasons Pythagorean Win% Secondary Conditions (Tighteners)
1) Exclude Teams with a SU WP at least 0.100 points higher than their PWP LS.
2) Points For> = 30 in their Last Meeting (LM4).
Top Teams: CIN (4); NO (4); ATL (2); BUF (2)
Overall (Since '94): 2-28 ATS
2007 Season: 0-1 ATS
2006 Season: 0-1 ATS
2005 Season: 0-4 ATS
2004 Season: 0-2 ATS
Last 3 Results. Pick in Brackets.
2007 WK2 – DET 20 MIN 17 (DET-3) P
2007 WK1 – MIN 24 ATL 3 (MIN-3) W
2006 WK2 – NO 34 GB 27 (NO -2) W