November 30, 2008

Tie Breakers - A Critique

The conference tiebreakers can be found here. That tread was intended to be informative, presenting the rules as they are without commentary. With the discussion about the Big 12 conference championship tiebreakers one could ask, what are good tiebreakers?

Here are my thoughts.

Conference tie breakers used:
Settle it on the field
- Head to head among tied teams
- Results vs opponents within the same division with the best record, repeated through the team with the worse record.

These two are the most consistently used in college football. Head to head chooses to let the dispute be settled on the field at the price of selecting the team that lost to the worst them. These tiebreakers favor an inconsistent team that plays up to a good team and down to a bad team over a team that plays at a consistent level throughout the season. It leans towards the upper end of a teams performance level.

At least with head to head you are using games between the teams. Head to head is also very effective for divisions where every team plays each other, never failing with a tie of an even number of teams and often succeeding when three teams are tied. If head to head fails the decent through the standings reduces to eliminating the team who lost to the best team, leaving the teams that lost to the worst teams. In cases where a group of teams beat all outsiders this method will always fail.

While head to head is here to stay, the decent down the standings should go.

Common Opponents
- Record against teams within the same division
- All common conference opponents

Comparing results vs common opponents is one of the most sound tiebreakers and should not be used more heavily.

A great, seldom used, tiebreaker
- Combined Record of non-division opponents

If head to head and division records are don't break the tie, the difference in cross over games can be used to break a tie. The team that beat the teams with the highest combined conference win loss record is a meaningful tiebreaker that would have a high degree of success.

Ineffective tiebreakers
- Eliminate teams not eligible for post season play
- All non-division opponents
- Results vs all conference opponents with the best record, repeated through the team with the worse record

Teams not eligible for post season play are usually eliminated a priori.

If a tie exists and is not broken by the record in the division, how will the non division opponents not be tied? Here a descent down the standings might make sense, but is inferior to the combined record of non-division opponents.

Interest of the conference
- Best overall W-L Record
- Least number of FCS teams played

These tiebreakers favor teams that benefit the perceived strength of the conference. Also teams have a direct say in controlling the number of FCS teams played.

Parity
- Eliminate the team with the most recent championship appearance

I like this tiebreaker as it gives a slight advantage to the underdog. It is also supported by the rich tradition of the Rose Bowl.

NFL Tiebreakers
- Total Point Head to head
- Total TD's in all games played

College football has a strong allergy to factors that make margin of victory a part of any procedure, as it is perceived as anti-sportsmanship.

Coin Flip
- BCS Standings
- BCS standings, unless top two are within 5 rankings, then head to head of top two in the BCS standings.
- Draw

The BCS standings are slightly better than luck of the draw, and almost guarantee the tie will be broken. The SEC rule is viewed as superior, but it would be even better to use the BCS to eliminate the lowest ranked team and reapply the rules.

Ideal Tie Breaker Procedures
When a group of tied teams is reach the procedures are followed until the tie is broken. The procedure is then repeated from the top.

1) Head to Head record of tied teams.
2) Division record of tied teams.
3) Record of tied teams vs all common conference opponents.
4) Combined conference record of non-division conference opponents.
5) Eliminate the team with the lowest BCS standings.
6) Eliminate team with most recent championship game appearance.
7) Draw lots publicly.

Cases in point
For the Big 12 south this would fall to #5, as the three teams would be tied at #4 at 11-13. #5 would eliminate Texas Tech, then Texas would win the head to head.

For the PAC 10, if USC loses to UCLA, this would fall to #5. #2, #3 and #4 are meaningless in this case. Oregon State would be eliminated, leaving USC as the PAC 10 champion. Currently the tie would be broken by California, the highest team in the PAC 10 standings they did not all beat. Oregon would be eliminated, crowning Oregon State.

These are the only possible ties this year that can't be settled by head to head.

Comments:

Tie Breakers - A Critique
The conference tiebreakers can be found here. That tread was intended to be informative, presenting the rules as they are without commentary. With the discussion about the Big 12 conference championship tiebreakers one could ask, what are good tiebreakers? Here are my thoughts. Conference tie breakers used: Settle it on the field - Head to head among tied teams - Results vs opponents within the same division with the best record, repeated through the team with the worse record. These two are the most consistently used in college football. Head to head chooses to let the dispute be settled on the field at the price of selecting the team that lost to the worst them. These tiebreakers favor an inconsistent team that plays up to a good team and down to a bad team over a team that plays at a consistent level throughout the season. It leans towards the upper end of a teams performance level. At least with head to head you are using games between the teams. Head to head is also very effective for divisions where every team plays each other, never failing with a tie of an even number of teams and often succeeding when three teams are tied. If head to head fails the decent through the standings reduces to eliminating the team who lost to the best team, leaving the teams that lost to the worst teams. In cases where a group of teams beat all outsiders this method will always fail. While head to head is here to stay, the decent down the standings should go. Common Opponents - Record against teams within the same division - All common conference opponents Comparing results vs common opponents is one of the most sound tiebreakers and should not be used more heavily. A great, seldom used, tiebreaker - Combined Record of non-division opponents If head to head and division records are don't break the tie, the difference in cross over games can be used to break a tie. The team that beat the teams with the highest combined conference win loss record is a meaningful tiebreaker that would have a high degree of success. Ineffective tiebreakers - Eliminate teams not eligible for post season play - All non-division opponents - Results vs all conference opponents with the best record, repeated through the team with the worse record Teams not eligible for post season play are usually eliminated a priori. If a tie exists and is not broken by the record in the division, how will the non division opponents not be tied? Here a descent down the standings might make sense, but is inferior to the combined record of non-division opponents. Interest of the conference - Best overall W-L Record - Least number of FCS teams played These tiebreakers favor teams that benefit the perceived strength of the conference. Also teams have a direct say in controlling the number of FCS teams played. Parity - Eliminate the team with the most recent championship appearance I like this tiebreaker as it gives a slight advantage to the underdog. It is also supported by the rich tradition of the Rose Bowl. NFL Tiebreakers - Total Point Head to head - Total TD's in all games played College football has a strong allergy to factors that make margin of victory a part of any procedure, as it is perceived as anti-sportsmanship. Coin Flip - BCS Standings - BCS standings, unless top two are within 5 rankings, then head to head of top two in the BCS standings. - Draw The BCS standings are slightly better than luck of the draw, and almost guarantee the tie will be broken. The SEC rule is viewed as superior, but it would be even better to use the BCS to eliminate the lowest ranked team and reapply the rules. Ideal Tie Breaker Procedures When a group of tied teams is reach the procedures are followed until the tie is broken. The procedure is then repeated from the top. 1) Head to Head record of tied teams. 2) Division record of tied teams. 3) Record of tied teams vs all common conference opponents. 4) Combined conference record of non-division conference opponents. 5) Eliminate the team with the lowest BCS standings. 6) Eliminate team with most recent championship game appearance. 7) Draw lots publicly. Cases in point For the Big 12 south this would fall to #5, as the three teams would be tied at #4 at 11-13. #5 would eliminate Texas Tech, then Texas would win the head to head. For the PAC 10, if USC loses to UCLA, this would fall to #5. #2, #3 and #4 are meaningless in this case. Oregon State would be eliminated, leaving USC as the PAC 10 champion. Currently the tie would be broken by California, the highest team in the PAC 10 standings they did not all beat. Oregon would be eliminated, crowning Oregon State. These are the only possible ties this year that can't be settled by head to head.