How I Pick Winners
As I previously mentioned, it begins with value handicapping. I determine how many points team A is better than team B, and then adjust my projection by taking homefield advantage into consideration. I then compare it to the posted pointspread. Since the oddsmakers are primarily in the business of gauging public perception (as a means of getting sufficient action on both sides of a game), not predicting the margin of victory, it is certainly more than possible to beat the pointspread on a somewhat consistent basis. It is also worth noting that I do not automatically adjust my projected pointspread by adding 3 points to the home team. This is because while 3 points is the average homefield advantage, it is not the homefield advantage for every NFL team. In fact, a team's homefield advantage can even be affected by the time of year. For example, in December and January my favorite team, the Green Bay Packers, have a homefield advantage of at least 4 points. On the other hand, the Miami Dolphins' homefield advantage in December and January is probably only 2 points.
I then look at the statistics that are most likely to indicate whether a team is overrated or underrated. Obviously, if team that is overrated is playing against a team that is underrated, I'll usually make a play on the underrated team, if I make a play on the game at all. I tend to shun computer generated predictions, which are usually based on multiple regression models. The problem with computer generated predictions is that while they do a great job of explaining past results, they do a mediocre if not horrible job of predicting future pointspread results. This is primarily due to the fact that correlation does not imply causation.
While 98-99% of trends are useless at best, there are a few trends that are worth looking at and I adjust my selections accordingly. The trends that I use are almost always league-wide trends, not team trends. The problem with using team trends is 2-fold. First, the sample size is too small. And second, personnel changes very rapidly in the NFL. What relevance is it what the Minnesota Vikings' record is in a certain situation over the last 10 seasons? Very few of the current Viking players were on the team 5 years ago, let alone 10 years ago. For me to use a trend there has to be both a significant sample size as well as a logical causative relationship that makes it likely that such a trend will continue. So what if the Dallas Cowboys are 9-1 ATS during the 3rd game of the season during the past 10 years? It makes more sense to relegate such a trend to what it really is, a coincidence. Nevertheless, the few trends (you can count them on one hand) that do meet my criteria are useful because they indicate when teams are likely to play over their heads and when teams are likely to play under their heads. I also look at injuries when making my pointspread selections. Specifically, I take into consideration which injuries the public is likely to overreact to and which injuries the public is likely to underreact to.
And last but not least, I use my inside contacts to find out where the public is on games and then usually go the other way (especially if the public is lopsided on a game, i.e. 70/30 or more). This is because, as I previously mentioned, the public tends to do horrible on the NFL, especially after the first 2 weeks of the regular season. The public wins less than 50% of the time, and hence they'd do better if they flipped a coin (which I think, by the way, is what a lot of professional sports touts actually do). Since I am a contrarian by nature, going against the public comes naturally to me. I by no means do this blindly, as it accounts for about 30% of my system. But very rarely do I go with the public on a game that the public is lopsided on, and when I do, it is almost always a small play. On the other hand, it is likely, but by no means certain, that I will go against the public in games, especially games that the public is lopsided on.
And contrary to popular belief, one can't look at line moves to determine where the public is on a game. This is true for several reasons. First, where the public is (i.e. the number of people betting on each side of a game) and where the money is is not always one and the same. Often times the public will be on one side in a game, and the money will be even or on the other side. Second, line moves are often determined by which side the wise guys play, not which side the general public plays. Third, often times the linesmakers adjust the line to take into account injuries that they did not know about earlier in the week. And fourth, the linesmakers often move the lines without concomitant action as a means of fooling the public, i.e. to get them to bet on the wrong side and/or to get more action on a game.