College sports ratings use a somewhat less sophisticated version of the NFL model that does not use individual play data. Final scores and game location are the only data used. Unlike BCS computer ratings, point difference is factored rather than simply the outcome of the game. Instead of adjusting ratings for repeatability, both teams in each game are given equal responsibility for the result. Each team is assigned a normal probability distribution for quality such that the most likely scenario considering all games at once is optimized. In doing so, home field advantage is factored where applicable and strength of schedule is automatically accounted for. After 1 game, a team's rating will simply be their net points divided by 2. Past the first game, relative strength of opponents and how far a result is from the expected result begin to factor in. Like the NFL ratings, the first few weeks' ratings are unreliable but quickly become increasingly accurate thereafter. Subtracting one team's rating from another's will give an expected point spread should they play at a neutral site.