Online Rankings System Update
Hello, my name is Michael Weisbecker and I am one of the designers on NCAA Football 13. Recently, there have been some questions and concerns with online rankings and the points earned from head-to-head online games. I am here to give some insight into what we did, how it works, and to also outline a few post launch changes made.
This year NCAA Football 13 switched our ranking system to be an Elo type ranking system. The Elo system calculates the relative skill of players and help match up players with a similar skill level. The original use for Elo was for chess rankings. Today, Elo is used for many different types of games, including other video games. In our previous system, a player could be very highly ranked on the boards with a losing or close to losing record if they played a large amount of games. The Elo system remedies that. In Elo, you can only rank up by winning.
Here is a brief overview on how Elo works. Users have a number associated with their rating. Obviously, a higher number represents a better player based on the results of games played against other users. When two players match up, the winner will gain and the loser will lose a certain amount of points. The difference between the two player’s ratings determines how many points are won or lost by each player. The maximum number of points a player can win or lose in a Match is the “K factor.”
Therefore, as an example, we will assume the K factor is 50 and we have two users Player A and Player B. All the numbers and results used are for the purpose of example and do not necessarily represent actual results from the game.
If Player A and Player B have the exact same rank and are matched up, the winner will gain 25 points and the loser will lose 25 points.
In the next examples, we assume Player A is ranked level 30 and Player B is ranked level 10.
If Player A wins, they would only gain 10 points, Player B would only lose 10 points.
If Player B wins, they would gain 40 points and Player A will lose 40 points.
Since Player A is ranked higher than Player B, they are expected to win in a matchup between the two. Therefore, if Player A wins, they earn fewer points for the win. On the other side, if they lose, they lose more points than normal since they are going against expectations.
Elo’s design is that over time, if users win more than they are expected, their rating will go up. Early on with an Elo type system, the rankings can fluctuate and be more volatile when a large number of players with varying skills are near the starting rank. Over time, players will reach their optimal rank as they play more games. The end goal is when two players with the same ranking are matched up, their skills are relatively equal and each player has a 50% chance of winning the game.
Now that the game has been out for a little while, we looked at the feedback. Based on that, we made a few changes to the ranking system. First, we are updating the skill level thresholds. They will be more granular at the lower levels than they are now. Many players will see a significant jump in their level after playing their next online game. The second change is adding a slight point bonus for playing and completing a game. This means players will get slightly more points for a win slightly fewer points for a loss than they received previously. With these changes, lower ranked users might see themselves go up or down three, four, or even five levels after winning a head-to-head online game. This is normal for many of the lower levels and will stabilize over time as more games are played and they level up. The final change we did was to institute a point floor of 300. With the point floor, a user will not fall below 300 points if they lose a game. For any user that had less than 300 points now, it means you will not lose any points for a loss and still gain normal points for a win until you get above the 300 point floor.