i found some kinda neat stuff(cappin)

[T-Bone]

thought id post some of it home it copies right
Can prior season WL and ATS records predict future season performance?
Theory One on the NFL is that the league changes dramatically from one year to the next, with free agency, a new crop of draft picks, coaching changes, and adjustments to the strategies that worked successfully in the prior season taking their toll on the predictive value of past season performance.
Theory Two is that there are always biases in how people respond to the events of the prior season (note for instance the poor historical record of the Super Bowl losers in the next year), and capitalizing on this should be something possible with enough data and the right tools.

We built a database with the straight-up and spread records for each team during the regular season from 1990 on. Then we ran some correlations and regressions on the data to see how well you can predict the next year by the previous ones.


Simple Correlations between previous records
and the current regular season (1990-2003)
Variable Correlate to Correlation
Last Season Wins This Year Wins .49
Last Season Wins This Year ATS .19
Last Season ATS This Year ATS .36
Last Season ATS This Yr Wins .41

So on the surface it looks like there is a fair amount of correlation. Ah, but the problem, as Bob Dylan has sung, is times have changed. If we change the data set to only reflect the 1999 to 2003 timeframe (using 1998 history for '99 projections) we get some notably different results:


Simple Correlations between previous records
and the current regular season (1999-2003)
Variable Correlate to Correlation
Last Season Wins This Year Wins .26
Last Season Wins This Year ATS -.20
Last Season ATS This Year ATS -.15
Last Season ATS This Yr Wins .14

All the correlations drop, and indeed two of them switch signs! A good prior season WL record or ATS record has been a negative indicator for the ensuing season's ATS performance. Some would say the year the Rams came out of nowhere to win it all in 1999 was the line in the sand between the old NFL and the new game...

All right, you may say, last year is not going to tell us very much in itself about this season, but what about using some kind of rolling number of years?

Well, we're glad you asked and we'll move next to regressions:


Regression to predict current season regular season Wins
Factor Coefficient Std Error
Intercept 4.791 .560
Wins, 4 yrs Ago .078 .055
Wins, 3 yrs Ago .027 .060
Wins, 2 yrs Ago .021 .060
Wins, Last Yr .287 .059
R^2 = 0.12
Ugh. Even with a fair amount of data points (n=305), the R squared is very small, and the standard errors are bigger than the coefficients in some cases! This suggests that you're not going to fare very well trying to predict the current season Wins using the last four WL records.

The next track then is to up the ante, let's use both the past four seasons Win records and Spread records to see if we might not find some tidbits of value -- for instance a team that posted a nice number of wins but a lousy spread record might be deemed an underachiever which could have a trend one way or the other in the following year.


Regression to predict current season
Factor Coefficient for WINS Coefficient for ATS
Intercept 4.658 8.531
Wins, 4 yrs Ago .136 .120
Wins, 3 yrs Ago .031 -.030
Wins, 2 yrs Ago -.071 -.060
Wins, Last Yr .325 -.079
Spread Wins, 4 yrs Ago -.110 -.130
Spread Wins, 3 yrs Ago .044 .043
Spread Wins, 2 yrs Ago .160 .119
Spread Wins, 1 yr Ago -.087 -.049
R^2 = 0.13 / 0.05
Even with eight factors in the regression, the R squared values are terrible. Basically we are being told that we should look elsewhere if we're intent on trying to predict next year performance.

Given this, the following projected win tallies should be taken very lightly as they will likely have little predictive value. Not to worry though, we have better regression tools in store!


Wins Spread Wins Projection
Team '00 '01 '02 '03 '00 '01 '02 '03 Wins ATS
TEN 13 7 11 12 7 5.5 10 8.5 10.1 8.4
KAN 7 6 8 13 7 7.5 9.5 10 9.7 7.7
PHI 11 11 12 12 10 10.5 10 11 9.5 7.7
BAL 12 10 7 10 10.5 8 10 9.5 9.3 8.2
STL 10 14 7 12 7 9.5 4 9.5 9.3 7.5
CAR 7 1 7 11 8 7.5 10 6 9.2 8.2
MIA 11 11 9 10 10.5 10.5 9 7 9.2 8.0
NWE 5 11 9 14 7 11 6 13 9.1 6.8
IND 10 6 10 12 9 6 6 9.5 8.8 7.3
DEN 11 8 9 10 9.5 7 7.5 8 8.8 7.8
SEA 6 9 7 10 7 6.5 9 8 8.8 7.8
MIN 11 5 6 9 8 5.5 8 8 8.8 8.4
GNB 9 12 12 10 8 9.5 8 10 8.6 7.6
DAL 5 5 5 10 7.5 8 8 9.5 8.4 7.7
NOR 10 7 9 8 10 6 9 8.5 8.1 8.0
NYJ 9 10 9 6 8 8 10 6 8.1 8.5
TAM 10 9 12 7 9 6.5 10 7 8.0 8.1
OAK 12 10 11 4 10.5 7.5 10 3.5 7.6 8.7
CHI 5 13 4 7 7.5 11.5 5 8 7.5 7.7
BUF 8 3 8 6 7 6 8.5 7 7.5 8.5
NYG 12 7 10 4 9 5.5 8.5 4 7.4 8.7
CLE 3 7 9 5 5 8.5 10.5 6 7.3 8.4
SFO 6 12 10 7 7 9 5.5 8 7.2 7.5
WAS 8 8 7 5 6.5 8.5 7 8.5 7.2 8.4
JAC 7 6 6 5 7.5 8 8 8 7.1 8.4
CIN 4 6 2 8 8 8 4 10 7.1 7.4
DET 9 2 3 5 9 9 6.5 8.5 7.1 8.6
ATL 4 7 9.5 5 6 7 10 6.5 7.1 8.2
PIT 9 13 10.5 6 11 11.5 5.5 8 7.0 7.4
ARI 3 7 5 4 4 9.5 6 6 6.6 8.4
HOU 4 5 8 9 6.5 8.4
SDG 1 5 8 4 7 5.5 9 6 6.1 7.8

So according to this admittedly dubious formula we've concocted from the regression outputs, the top three spread value teams for the NFL 2004 season are projected to be the New York Giants, Oakland Raiders, and Detroit Lions, whereas by far the worst spread value for the coming season is predicted to be the New England Patriots. Hmmm, there may be some hint of truth to that after all...



--------------------------------------------------------------------------------

 

[T-Bone]

"Contrarian Bridgejumping" Revisited
To refresh everyone's memory, the "C/B Jumping" methodology involves looking at how a team has done against the line set on its games. Many people look at a team's "net points" on the year (points for minus points against), but few people actually track how a team has done on average against the spread of its games. For instance a team that wins by ten points when they were favored by seven would have a +10 for "net points" but only a +3 for their performance against the line in that event.

The C/B Jumping theory said to play a team in a matchup whose "net against the line" was more than SIX points WORSE than its opponent. E.G. Team A is +2 against the line on average, Team B is -5, so you would play Team B since it is 7 points worse. (For the full text of the original article read "Contrarian Bridgejumping").

We'll take a look at this approach using a database that includes all the games played from week 5 on from 1983-2000:


Contrarian Bridgejumping 1983-2000 (Wks 5-12) C/B Net 0 - 1.9 2 - 3.9 4 - 5.9 6 - 7.9 8+ Pts ALL
HomeFavs 147 - 164 100 - 125 57 - 60 58 - 30 57 - 37 419 - 416
HomeDogs 81 - 67 67 - 66 74 - 64 64 - 48 126 - 119 412 - 364
AwayFavs 48 - 59 30 - 51 19 - 27 13 - 12 14 - 15 124 - 164
AwayDogs 155 - 164 138 - 148 124 - 119 101 - 91 156 - 127 674 - 649

Favorites 195 - 223 130 - 176 76 - 87 71 - 42 71 - 52 543 - 580
Underdogs 236 - 231 205 - 214 198 - 183 165 - 139 282 - 246 1086 - 1013
Home Teams 228 - 231 167 - 191 131 - 124 122 - 78 183 - 156 831 - 780
Away Teams 203 - 223 168 - 199 143 - 146 114 - 103 170 - 142 798 - 813

ALL PICKS 431 - 454 335 - 390 274 - 270 236 - 181 353 - 298 1629 - 1593
WIN % 49 % 46 % 50 % 57 % 54 % 51 %

Spread Range
in 6+ games 10+ points 5 to 9.5 0 to 4.5
Favorites 9 - 3 41 - 19 92 - 72
Underdogs 106 - 102 193 - 158 148 - 125


ANALYSIS: The original theory only involved looking at the big difference (6+) matchups, and in those two columns above (the 6-7.9 range and 8+ range) the total adds up to a 589 - 479 record, which is 55.1% overall...not bad, but not the heights of the 1992-1996 period (there are admittedly some likely discrepancies between the lines we are using now and the lines in the database used for the prior assessment).

What is particularly interesting is that favorites have been astonishly good plays under these circumstances -- 142-94 overall (60%) and a scintillating 50-22 (69%) in cases where the favorite is favored by 5+ points. So it would seem the theory has considerable merit over the lengthy time period of 18 years.

The next concern is "what have you done for me lately?"


Contrarian Bridgejumping 2001-2003 (Wks 5-12) C/B Net 0 - 1.9 2 - 3.9 4 - 5.9 6 - 7.9 8+ Pts ALL
HomeFavs 8 - 9 8 - 6 8 - 9 7 - 7 4 - 7 35 - 38
HomeDogs 7 - 2 3 - 10 12 - 10 7 - 6 20 - 9 49 - 37
AwayFavs 4 - 6 1 - 4 2 - 2 1 - 1 2 - 2 10 - 15
AwayDogs 16 - 12 11 - 17 17 - 9 14 - 10 26 - 14 84 - 62
Favorites 12 - 15 9 - 10 10 - 11 8 - 8 6 - 9 45 - 53
Underdogs 23 - 14 14 - 27 29 - 19 21 - 16 46 - 23 133 - 99
Home Teams 15 - 11 11 - 16 20 - 19 14 - 13 24 - 16 84 - 75
Away Teams 20 - 18 12 - 21 19 - 11 15 - 11 28 - 16 94 - 77
ALL PICKS 35 - 29 23 - 37 39 - 30 29 - 24 52 - 32 178 - 152
WIN % 55 % 38 % 57 % 55 % 62 % 54 %

Spread Range
in 6+ games 7.5+ points 3.5 to 7 0 to 3
Favorites 1 - 2 4 - 3 9 - 12
Underdogs 23 - 7 33 - 23 11 - 9


ANALYSIS: An 81-56 record in the key 6+ difference games, good for 59% against the spread...uh, yeah I'll take that! On the other hand, favorites with the C/B call have only been 14-17 against the line, so it's a dog driven method of late, turning the tables somewhat on previous patterns.

Conclusion: the Contrarian Bridgejumping handicapping approach is still a force in the NFL during the critical mid-season stretch. Heed the numbers!

 

[T-Bone]

NFL Preseason Research: "GAP" Comparisons
Wagering on preseason games is often deemed to be a risky proposition by many. Teams are not playing with the same emphasis on winning, thus leading to questionable/variable motivation. The focus is on evaluating specific players trying to make the final roster, getting the regulars ready for the season, trying out a few new schemes (but without giving anything away)...and most importantly, avoiding key injuries!
To paraphrase Charles Dickens, "There was everything at stake, there was nothing at stake."

However, maybe by scrutinizing the previous year performance of teams we can find some spread trends of note. For instance, are teams coming from poor seasons more interested in posting some wins in pre-season than teams coming off strong seasons? Perhaps the line over-adjusts for this common theory and the reverse is true! Let's look to the data to tell us the answer...

We elected to categorize teams as follows:

Good -- teams with a 10 or more wins in the prior regular season
Average -- teams with 7 to 9 wins
Poor -- teams with less than 7 wins
Running these guidelines through all the pre-season match-ups for the period of 1997-2003 (eg a seven year span) produced the following results:

Home Team Away Team Won
(vs Spr) Lost
(vs Spr) Home
W%
Good Good 19 21 47 %
Good Average 24 22 52 %
Good Poor 20 25 44 %
Average Good 25 26 49 %
Average Average 34 23 59 %
Average Poor 12 29 29 %
Poor Good 16 17 48 %
Poor Average 21 30 41 %
Poor Poor 24 28 46 %
Home 195 221 47%


Away teams have held a slight edge overall, but were only 32-34 in the 2003 preseason. Clearly the most remarkable results have been the "Average home team vs Poor away team" match-up, where the poor team has covered 71% of the time, however the '03 season again did not follow this pattern. Thus the underlying truth to some of this research comes out quickly: the sample sizes are too small in most cases to be meaningful.

The next thing you might say is, well what about breaking it out by the specific weeks (conventionally labeled week "zero" to week four)? The drawback to this is our already small sample size gets even smaller, but we went ahead and ran it anyway and rather than give you the full tables, we'll summarize the interesting findings:

Week 1 - Poor prior season teams were 2-14 against the spread at home against poor or average away teams from '97-02 and 1-2 in 2003!
Week 2 - Poor prior season teams are 26-9 (74%) versus the line against good or average teams (just 2-2 in 2003). Interestingly the poor sides are especially strong as road warriors, 15-2 versus good/average home teams.

Week 3 - In what some people consider the most serious week of play, poor teams have struggled at home against average/good teams, mustering a 5-16 record (1-2 in 2003) as the superior teams "come to play."

Week 4 - nothing stands out as too strong.


--------------------------------------------------------------------------------

Another logical area to explore is breakouts of favorites and underdogs.


Home Favorites/Away Underdogs, 1997-2003 Preseason Home Team Away Team Won
(vs Spr) Lost
(vs Spr) Home
W%
Good Good 15 19 44 %
Good Average 22 21 51 %
Good Poor 18 25 41 %
Average Good 16 20 44 %
Average Average 30 22 57 %
Average Poor 12 28 30 %
Poor Good 4 7 36 %
Poor Average 13 22 37 %
Poor Poor 18 25 41 %
Totals 148 189 44%


So playing the underdog in the preseason has been a solidly profitable play over the last seven years, good for a 189-148 and 56.0% record. During the '03 exhibition games the dogs were 29-25 for a lower than usual 53.7% number.

As to whether poor previous year teams 'try extra hard' at home to win back the fans, the conclusion we would draw is no. Poor teams are only 35-54 (39%) as home favorites, and 3-5 to go with the trend in 2003. Ah, but 'poor' teams as Road dogs are a stellar 78-48 for 62% (although just 6-7 in '03).

Upping the line restriction to say being a field goal favorite or more, doesn't change the overall percentages much at all, and with the exception of an oddity in 'Average' teams at home versus 'Average' away teams, where you find a 17-6 record as favorites of 3.5 or more, there isn't much to comment on.


Home Underdogs/Away Favorites/Pick'Em games Home Team Away Team Won
(vs Spr) Lost
(vs Spr) Home
W%
Good Good 4 2 66 %
Good Average 2 1 66 %
Good Poor 2 0 100 %
Average Good 9 6 60 %
Average Average 4 1 80 %
Average Poor 0 1 0 %
Poor Good 12 10 54 %
Poor Average 8 8 50 %
Poor Poor 6 3 66 %
Totals 47 32 59 %


Once more the call to action is clear -- play the home dogs over away favorites.



--------------------------------------------------------------------------------

We believe some of the GAP data conclusions are worth looking for in the 2004 action --

Look to play underdogs where possible.
Play AGAINST Poor teams as home favorites
Play on Poor teams as away underdogs
Play on home underdogs
Week 1: Lean AGAINST Poor home teams versus AVERAGE/POOR away teams
Week 2: Lean to POOR teams against GOOD/AVERAGE opponents, particularly on the road
Week 3: Lean to AVERAGE/GOOD home teams on the road against prior year POOR teams
Lean to POOR away teams against AVERAGE home teams
Lean to AVERAGE home teams when favored by 3.5+ points against AVERAGE away teams
Lean AGAINST Poor home teams when they are favored by 3+ points
Don't worry, if you are a subscriber we will point out all the games where the above 'rules' are at work in the weekly preseason coverage.
Some of the more interesting preseason scenarios involve looking at a team's W-L record in the preseason to that point (the famous "is a team without a preseason win so far, more or less likely to cover the next time out)...the next article in this series!