How We Create And Use Power Ratings That Beat The Point Spread
Power Ratings are the crutch that we lean on. We do not watch every game. Life gets in the way and there just isn't enough time in the day for that. We do, however, want to make the most educated recommendations possible and power ratings provide the clarity that we are seeking.
Computer rating systems tend more toward objectivity, without specific player, team, regional, or style bias. A huge advantage of computer rating systems is that they are comprehensive, requiring assessment of all selected criteria and one can objectively track all of the teams in a given league by relying on the numbers.
Yes, power ratings are a lot of work to maintain. Power ratings require constant updating and equal work spread across each team. Further, they are difficult to adjust. It is tough to make accurate movements on a daily basis and avoid making emotional, irrational decisions.
I have taken my years of relying on power ratings and developed a simple method that generates predictive point spreads daily and that we, as a team, compare to market spreads to identify edges of 3 points or more.
Creating the Ratings
Our method is founded on point differential; the numerical gap between a teams points scored and points allowed. To generate predictive point spreads, we compare two teams playing each other and determine the difference in their Power Ratings. We will use the 2018 National Championship Game between Alabama and Georgia as our example.
Alabama averages 37.9 ppg and gives up 11.1 ppg. Georgia scores 36.3 ppg and gives up 15.7 ppg. For the ratings, Alabama is a +26.8 and Georgia is a +20.6.
Next we subtract the difference in the two teams ratings (6.2) and divide this by 2. This gives us a fundamental point spread of Alabama being favored by 3 points in this game. The Power Ratings Spread is Alabama -3. This is the number that we make adjustments to and is a direct prediction of how we think the game would end on a neutral playing field.
Adjusting the Ratings
Travel time, and venue change have a big impact on the outcome of games and markets. To create an even more accurate spread, one must account for home field advantage.
When two teams of equal quality play, the team at home tends to win more often. Across all conditions, simply playing at home increases the chances of winning. The size of the home field advantage changes based on the sport, season length, the number of time zones crossed, and even the qualities of the individual stadium and crowd size.
The general rule among bookmakers is that home field is worth three points. We however, have a range from 1.5 to 7.5 points across all of the sports (.15 to .75 in baseball and hockey); with the advantage being higher in basketball.
Beyond points, we choose to include more granular information about the game. Examples include time of the expected pace of the game, possession of the ball, individual statistics, and lead changes. Data about weather, injuries, or "throw-away" games either in season or near season's end may affect game outcomes but are difficult to model. "Throw-away games" are games where teams have already earned playoff slots and have secured their playoff seeding before the end of the regular season, and want to rest/protect their starting players by benching them for remaining regular season games. This usually results in unpredictable outcomes and may skew the outcome of rating systems.
Moreover, teams often shift their composition between and within games, and players routinely get injured. Rating a team is often about rating a specific collection of players. Here is were we have to be very careful. Injuries spark emotion; and we avoid instant adjustment on injuries at all costs. Unless it is a quarterback, Lebron James, or a handful of elite skill position players, the chances of the spread being affected by more than a point or two is highly unlikely.
We are more interested in keeping an eye out for “cluster injuries”. A cluster injury is when a team suffers multiple injuries at correlating positions. For example, multiple defensive backs, linemen or receivers. These are injuries which can devalue teams and affect the outcome of a game a great deal.