"Handicapping" is an attempt to equalize the relative ability of two teams by giving the weaker team extra points (the point spread). A perfectly handicapped game gives either team a 50% chance to win after the point spread is considered. There are a number of different handicapping approaches that can be used when trying to break down a game. Although the definitions can vary by author, I break handicapping down into 3 primary forms:
* fundamental handicapping, in which the teams' relative strengths and weaknesses are measured (using game statistics and "power ratings") in order to arrive at a fair pointspread;
* situational handicapping, which identifies situations in which a team over- or underperforms relative to the point spread, most often implemented through a data-mining approach;
* and technical handicapping, in which betting patterns and line movement are used to gauge the relative strength of a wager.
This article will provide an introduction to linemaking from a fundamental (statistical) standpoint. One major issue that bettors have in using statistics is that they tend to use "game statistics" as a starting point, which can spell trouble. There are two major problems that fundamental handicappers often fail to consider in relation to past performance numbers:
* The strength of a team's prior opponents, and
* Statistics in terms in the number of plays used to compile them.
As an example of the first point, let's say it is the fifth week of college football, and we are handicapping a game between team A and Team B. Team A rushed for an average of 120 yards per game over the first 4 weeks, while Team B rushed for 90 yards per game. Team A is the better rushing team, right?
Not necessarily. Team A might have racked up those 120 yards per game against Temple, Buffalo, Kentucky, and Purdue, four weak defenses. Meanwhile, Team B may have run for 90 yards per game against Michigan, Virginia Tech, Miami, and LSU, four very strong defenses. Against an "average" defense, Team B may very well have a better rushing game than Team A. When using statistics to make predictions about upcoming games, it is important to adjust past game statistics according to the strengths of the teams those statistics were compiled against. The raw statistics themselves can often be misleading.
In regards to the second point, the Carolina Panthers passed for an impressive 423 yards in Week 14 versus the Giants. However, this seems much less impressive when you consider that Carolina had 61 pass attempts in that game, for only 6.9 yards per pass. By comparison, Philip Rivers has passed for 7.4 yards per pass the entire season, even though he has been largely overshadowed by the Chargers' running game. When the number of attempts is considered for handicapping purposes, it can make a big difference in assessing how a team actually performed. Teams that go out to an early lead may "pad" their rushing statistics by trying to run out the clock, while teams that regularly fall behind may pass a disproportionate amount of times to try and catch up on the scoreboard. Ideally, you should look at statistics on a "per" basis: yards per rush, points per play, yards per point, etc.
As an example of some basic fundamental handicapping, let's take a look at the upcoming BCS championship game.
Florida averaged 26 points per game (ppg) this season. Their opponents gave up an average of 25 ppg. So their offense is only slightly (4%) above average.
Ohio State gave up an average of 10 ppg against teams averaging 23 ppg. So their defense is far above average (57%).
So one way to predict "average" scores in this matchup is to take Florida's average offense (26 points) and adjust based on Ohio State's defense (10 divided by 23 = 0.435). 26 X 0.435 gives us 11.3 points for the expected Florida offensive production.
We can also look at the flip side of the coin and use as a baseline Ohio State's average defense (10 points) and adjust for the strength of Florida's offense (26 divided by 25 = 1.04). 10 X 1.04 gives us 10.4 points of Florida offense. Let's average the two to give us a prediction of 10.9 points for Florida.
Likewise, we can apply the same kind of analysis to Ohio State's offense. Ohio State averaged 36 ppg against teams that gave up an average of 24 ppg, a very strong offensive performance. Florida's defense is also very strong, allowing only 15 ppg to teams scoring 20 ppg. Adjusting as described above gives us a projected Ohio State score of 22.5 or 27 points. Averaging the two gives us 24.75 points for Ohio State.
So a simple predicted score stands at
Ohio State 25 Florida 11
giving us a projected line of Ohio State (-14) with a total of 36.
Clearly making a line can't be this easy, or we would all be picking 75% winners against the spread. There are numerous other factors that must be considered. But this is a good starting point for beginning handicappers so they can understand the fundamental matchup between two teams. Of course, this can be taken further, looking at yards per rush, yards per pass, etc. Let's take a look at a few of these additional factors.
Strength of schedule should be an important consideration, particularly in college sports. A team in the MAC may have demolished their opponents, but shouldn't be expected to have the same level of success when stepping up in class to play a Big 10 team. One easily accessible tool for evaluating strength of schedule is available in the Sagarin ratings published by USA Today. In our BCS championship matchup, this is not too much of a consideration, as the teams are fairly close in terms of strength of schedule (Ohio State - 72.08; Florida - 74.48). We can also take note that the Sagarin "predictor" score gives Ohio State an 11-point advantage in this contest.
There are no particular injuries of note in this game, so that is one less thing to worry about. However, we may want to consider Ohio State's extended layoff (51 days) coming into this game.
There is one other lopsided factor to consider: turnover margin: Ohio State was +11 on turnovers in 12 games, while Florida was only +3 in turnovers in 13 games.
There are two schools of thought regarding turnovers: (1) they are random and cannot be predicted, or (2) they are reflective of a teams strengths and weaknesses - better QBs will throw fewer interceptions, better pass coverage will result in more interceptions made, etc. How you interpret turnovers has an impact on your use of them in a handicapping perspective.
If we take the second viewpoint that turnover margin is based on team ability, then turnover margin need not be factored further into our lines; it is already reflected in other statistics like points per game or yards per play. However, if we feel that turnovers are due largely to "luck", then clearly Ohio State has benefited from luck more than Florida during this season. Therefore, Ohio State's statistical numbers are "inflated" relative to Florida's based on sheer luck, and we should adjust our numbers accordingly.
I am of the opinion that turnovers are a combination of both skill and luck. As turnovers can have a large effect on scoring, I will adjust Ohio State's strength slightly downward in my handicapping to reflect the fact that Ohio State has benefited at least slightly from "luck" this season.
This is an appropriate time to make any other adjustments to your projected line, including any "match up" insights that you think are significant, for instance the Florida speed and spread offense which is unlike anything Ohio state has faced this season.
So our original line of Ohio State -14 can now be adjusted slightly downward based on (1) Florida's slightly stronger strength of schedule, and (2) Ohio State benefiting from the "luck factor". Since there are no significant injuries to consider in this fundamental handicapping analysis, I would make the final line about Ohio State -11.5. Note that as a bettor it is not essential to make a "perfect line", especially in football. All we need to know is that the line should be somewhere between 10 and 14, the "key" numbers. With most sportsbooks offering Ohio State at -7, you should make Ohio State a play regardless, at least according to this simple analysis.
While this method of fundamental handicapping was described with an eye towards football, it can be used for almost any other sport with some minor modifications. One caveat in using statistics to handicap is small sample sizes. The more data we have, the more likely we are to determine a team's "true" tendencies. For this reason, using statistical analysis early in the season can be problematic, and it is almost always an issue in football, where, even in the best case, we have only 13 or 19 games to look at in college and pros, respectively.
In the next installment, we will take an in-depth look at situational handicapping, including evaluating "systems" for significance and deciding when once-favorable situations have already been factored into the betting line.