HANDICAPPING THE COLLEGE BOWLS - Stats to Use

Pregame Blogs

Pregame Blogs
Videos are just the START of the conversation. Each show has a dedicated blog post with show notes, links, and pics. Plus, the host and guests continue the conversation in the comments section!

HANDICAPPING THE COLLEGE BOWLS - Stats to Use

There are three sets of statistics from which to choose for the purpose of handicapping a Bowl game being played on a truly neutral field:

1. Neutral field statistics when they are available.

2. Away statistics for both teams

3. Overall statistics including Home, Away, and Neutral field games. 

Many handicappers make the mistake of assuming that because both teams are playing on the road, that the most accurate handicapping statistics for a neutral field would be the Away statistics for both teams.  Road statistics for both teams are the worst of the three choices.

The corollary to a home field advantage is a road disadvantage. Just as when you pick a winner you must also pick a loser, the home advantage and the road disadvantage are inseparable from one another. Built into the road statistics for each team is their road disadvantage and each opposing team's home advantage.  As a result, using road statistics for both teams is no more accurate than it would be to use home numbers for both teams

The best statistics to use would be neutral field statistics, but for the fact that there are not enough such games in any season to create a statistically valid set of past performances. In college football, prior season results and statistics should never be a part of your handicapping. Players change, and those players who remain the same have matured and gained experienced. The team from prior years is rarely similar enough to the current team for prior year statistics to be used.

The only choice left are the overall statistics from all games. These are, from a practical standpoint, the most accurate statistics to use when handicapping the bowls. The home and road advantages and disadvantages offset one another within the overall numbers to form a true picture of a team's overall ability.

Keep in mind, however, that use of overall statistics is the best only for truly neutral fields. If one team is playing close enough to their campus to create a partisan crowd and eliminate the need to travel, every factor that creates a home edge will exist except familiarity with the stadium. Unless there is something unusual about the home stadium, familiarity with the physical stadium can be assumed to be the least important factor in the home field advantage.  The best numbers to use when one team did not need to travel are Home statistics for the team playing near their campus, and Road statistics for the team that is required to travel.

Every one of my Bowl selections for the season is now available in one Bowl Season Package, or if you prefer you can purchase the single day package.  Both the Full Season and One-Day Packages are available by CLICKING HERE.

  • By the way,  from a purely anecdotal standpoint, I have used both Away/Away and Overall statistics to handicap the same games, and found my prediction of score differentials are generally more accurate using the overall numbers.  That is not to say that comparisons cannot be validly made using the Away/Away numbers.  When makikng comparisons, so long as the handicapper uses the same statistics for both, a comparison can be made that has some validly.  It is simply a question of comparative accuracy between Away/Away and all games.  What can be done to obtain comparisons of teams for the purpose of a final point differential,however, cannot be used at all for absolute score predictions such as a total.  To predict whether a line will be covered, all i need to do is compare the teams on some equal basis.  If one team is significantly better than the other, I can make a valid prediction using any equivalent statistics for the two.  If the line is -3 and the Away/Away numbers predict a 10 point win by the favorite while the Overall numbers predict a 17 point, I will have correctly predicted a score cover either way.  Still, if the game ends with the favorite winning by 17 or more points, I will consider the overall numbers to have produced the more accurate result.  That becomes important if the line is close to the prediction or if the plays are to be rated.  When predicting a total, however, the handicapper must come up with a single score size prediction rather than a team comparison.  For that prupose, I have found that using Away/Away numbers lacks sufficient accuracy to create a wining percentage.  In fact, the predictions of the total can often be completely opposite using both sets of statistics on the same game, with the Overall having a significantly better win percentage over the two seasons that I used both sets of numbers.  In fact, using the Away stsatistics would have produced two losing seasons, while using the overalll numbers produced one small winning season and one big winning season during the same years.

    It seems that the location of a game can greatly affect the total, and averaging the locations is simply producing a less skewed single score size prediction.  Sciores are not a good way to predict totals in football, but you can do your own research by adding up total yardage offensively and defensively and averging them separately for home games and road games.  Then compare the yardage.  You will see some consistent differences for each team.  This comparison must be done for each separately because some teams will improve more offensively at home, while others will improve more defensively at home.  

  • Thanks.  I agree that a comparison of neutral field numbers to home and away numbers would be very informative.  Unfortuantely, in college football there simply are not enough neutral field games for valid research to be done.  Often, the bowl game is the only neutral field game for the season.  When there is an earlier season neutral field game, it commonly is not a true neutral field, but rather an alternative local field for one of the teams.

  • Very interesting thread, thanks for the response.  While all of your assumptions are purely speculation and based on your own personal intuition, I would be very interested to see some stats on how teams perform in neutral site games.

    For example, I would be interested to see if neutral site games play out more like a road or home game for teams.  Would be very interesting to see those results if anyone has them.

  • To:  Betting Research

    Obviously, any specific won-lost records I could gather to prove that it is better to use Overall statistics than Away/Away statistics would be based on my own statitistical method of handicapping.  They would also be strictly anecdotal.  

    There is, however, mathemaitcal support for my conclusion as to the statistics to use.  The first is the Law of Large Numbers.  Statistically, the larger the number of trials, the more statistically accurate the results.  From that we can conclude that the usual 0 - 2 games on a truly neutral field are less accurate for statistical extrapolations than all games for the season.  

    That which can't necessarily be proven by statistics, can often be proven by logic.  As an extreme example, we can logically deduce that using game statistics for the teams playing in the Super Bowl is a more accurate means of predicting the results of the game than using statistics from two random teams that are not playing in the game.  Likewise, a statistical correlation between the Conference that wins the Super Bowl and the performance of the stock market in that year can be shown, but logical deduction allows us ot conclude that whether the AFC or NFC wins the Super Bowl would be a poor basis for putting our 401-K into cash for the year.   That is true even though the short term random statistical pattern opposes the logic.  

    The home edge can be statistically demonstrated in every season by comparing the total SU home wins to the total  SU road wins for every team.  We can certainly observe visiting QB's being unable to communicate with the team due to crowd noise.  We know that travel time necessarily takes away from practice and preparation time.  There are no lack of news stories about teams having travel delays, hotel problems, and food poisoning.  The Giants and Jets opening and closing the stadium doors to negatively effect the visitors is easily observed.

    If there is a difference between home performance and road performance, then the statistics using road performances and home performances for the same team can be assumed ot be different.  This can be observed for every team.  It is not just the won-lost records that are different.  It is the home and road yardage, turnovers and penalties and filed goal accuracy that differ for virtually every team every year.  That does not mean that for every team the home numbers are better than the road numbers.  Some teams perform better on the road.  It does not matter, because we are merely discussing which set of statistics will provide the most accurate picture, not which set of statistics will be better for every team.

    I don't need to prove statistically in evaluating a pharmaceutical, using cure statistics for humans will be more accurate than using cure statistics for rats.  That is true based on logic, and even though in many instances or even a majority of the instances if something cures a rat it will also cure a human.  Likewsie, if I want to predict how a team will perform at Home, I can deduce logically that extrapolating from their performance at home will be more accurate than extrapolating from their performance on the road.  Indeed, using performance on the road and rejecting performance at home when attempting to predict home performance flies in the face of logic.

    Using fa smaller set of numbers that we know are skewed to only the road portion  of a team's performance would only be logical if the team will experience most of the same disadvantages that are experienced in road games.  On a truly neutral field, most of he road differences don't exist.  The crowd is not creating so much noise that one team can't hear while remaining silent for the other team.  Neither team can control the environment and pracitce times to the detriment of the other.  Thus their performance only when such factors exist cannot be said to be more accurate than using statistics from a larger set of games.  Mathematically, the only excuse for extrapolating from a smaller set rather then a larger set is that the factors involved in the smaller set more closely resemble the conditions the teams will encounter today.  When playing on a truly neutral field, both teams are equally advantaged and disadvantaged.  

    As stated in the article, neutral field performance would be the most accuate measure but for the fact such figures don't exist for many teams and when they do exist the samples are tiny.  If you can find two teams with 3-4 games on a truly neutral field for the season, by all means use those statistics.  

    Using overall statistics can be logically deduced to be a complete picture of  the team's performance under all conditions.  Using smaller samples that are skewed by factors that will not exist in the game to predicted is no more logical than intentionally using only performances in heavy rain and wind to predict performance on a clear sunny day, while rejecting clear sunny day performances.  If we don't know what the weather will be, the most accurate statistics would be all games.  

    The purpose of the article was not to teach the reader logic, but to point out a circumstance which many forget when using Away, Away statistics, which is that Road statistics don't reflect he conditions of a neutral field any more accurately than home statistics.    

    The use of game statistics in one's handicapping

  • Rob,

    Very interesting write up you made here, and I respect the time you took to make this.  But, how do you know that the claims you made in this write up are accurate?  Have you checked results based on using these methods for making selections in years past?  I'm just not sure I am seeing how you can prove these statements to be true?  Would love to know where you derived these claims from.

    Thanks!