FBR09 Rankings September 18th, 2017

How does FBR account for games played out of state?

ProphAC wrote:How does FBR account for games played out of state?

It's up to each individual person who is doing the rankings, but generally out of state games won't be counted unless they are played against a team that is already established in the database with 3 or more games against other NTX teams.

For example, if 2 NTX teams go up and play in a tournament in OK and end up playing each other at some point, I would include that game, but wouldn't include the games against the OK teams.

The problem with out-of-state games like that, is establishing an accurate SoS for the out-of-state teams, since they won't have multiple games against NTX to establish an accurate baseline.

The opposite is true when an out-of-area team comes and plays in an NTX tournament. In that case, the out-of-area team will generally get at least 3 games against NTX teams in the database, so an accurate SoS can be established for that team, relative to the other NTX team in the database. So those games will generally all be included in the rankings.

Does the system adjust if an opponent drops out of a tier? For example, if you beat a tier 1 team early and that team has a bad year and drops out of tier 1, is that win retroactively adjusted for where the opponent is now, even though the game was months ago?

smugrr wrote:Does the system adjust if an opponent drops out of a tier? For example, if you beat a tier 1 team early and that team has a bad year and drops out of tier 1, is that win retroactively adjusted for where the opponent is now, even though the game was months ago?

Yes.  It's a dynamic system that adjusts over time.  It's one of the factors that was considered in determining how long of a history we keep in the database.  In general, we keep the most recent 16 "playing weeks" (weeks in which games were actually played, as opposed to 16 calendar weeks - this helps keeps some level of continuity from fall->spring and spring->fall season).

If I could change 1 thing in the FBR SW package, it would be to add in the ability to give a sliding weight scale based on age of result, that would weigh more recent results heavier than older results.  Unfortunately, we were never able to compel the software developers to add that into the package.  So, as it stands, a game played last weekend carries the same weight in the rankings as a game played 16 weeks ago.

FBR is not a perfect system (actually, no ranking system is perfect), but it has a long track record of working pretty darn well.  My philosophy was always to try and be as above the board as possible with the methodology, so that people could understand it's points of emphasis, strengths, and also limitations.  

People tend to get really hung up on whether a team is ranked a couple of spots too high or too low, and my response to that has always been that we try to make FBR as "accurate" as possible, but if you are focusing on FBR at that level, then you are ultimately missing the point of why we publish the rankings. (or at least why I published the rankings for several years)

This has been a very helpful thread... thanks!

bwgophers wrote:

smugrr wrote:Does the system adjust if an opponent drops out of a tier? For example, if you beat a tier 1 team early and that team has a bad year and drops out of tier 1, is that win retroactively adjusted for where the opponent is now, even though the game was months ago?

Yes.  It's a dynamic system that adjusts over time.  It's one of the factors that was considered in determining how long of a history we keep in the database.  In general, we keep the most recent 16 "playing weeks" (weeks in which games were actually played, as opposed to 16 calendar weeks - this helps keeps some level of continuity from fall->spring and spring->fall season).

If I could change 1 thing in the FBR SW package, it would be to add in the ability to give a sliding weight scale based on age of result, that would weigh more recent results heavier than older results.  Unfortunately, we were never able to compel the software developers to add that into the package.  So, as it stands, a game played last weekend carries the same weight in the rankings as a game played 16 weeks ago.

FBR is not a perfect system (actually, no ranking system is perfect), but it has a long track record of working pretty darn well.  My philosophy was always to try and be as above the board as possible with the methodology, so that people could understand it's points of emphasis, strengths, and also limitations.  

People tend to get really hung up on whether a team is ranked a couple of spots too high or too low, and my response to that has always been that we try to make FBR as "accurate" as possible, but if you are focusing on FBR at that level, then you are ultimately missing the point of why we publish the rankings. (or at least why I published the rankings for several years)

Software developers?

soccerjack wrote:

Software developers?

Yup. Software developers. FBR stands for "FootBall Rankulator", which is a commercially available software package that was created back in the days of the College Football BCS system, and was designed to allow people to create their own "BCS" style rankings for College Football.

mslater (still lurking on the forum on about the 7th iteration of screen name) stumbled upon the FBR software and started using it to "build a better mousetrap" because he didn't like someone else's methods for ranking the '01's. I got involved a few months later, and the two of us worked to refine the system to it's current methodology, which included getting the FBR developers to add/modify some features to suit our purpose. However, I couldn't convince them to add a sliding weight factor.

Anyway, long story short... anyone who publishes "FBR" rankings on here is purchasing and using the FBR software package to do this.

One more question since you are being so kind to give us the back story. Any reason why a boy parent can't buy the software and use it to create FBR for the boys side b/c they have wished for their own FBR ranking for years?

ProphAC wrote:One more question since you are being so kind to give us the back story.  Any reason why a boy parent can't buy the software and use it to create FBR for the boys side b/c they have wished for their own FBR ranking for years?

None whatsoever. The website is footballrankulator.com. Last I knew, they were charging $9.99 for the software. One-time fee, no yearly license or anything. Shoot me a PM. I've got setup files and instructions to get a newbie started up.

bwgophers wrote:

ProphAC wrote:One more question since you are being so kind to give us the back story.  Any reason why a boy parent can't buy the software and use it to create FBR for the boys side b/c they have wished for their own FBR ranking for years?

None whatsoever.  The website is footballrankulator.com.  Last I knew, they were charging $9.99 for the software.  One-time fee, no yearly license or anything.  Shoot me a PM.  I've got setup files and instructions to get a newbie started up.

The semi retirement lives on.

Very cool. I'm on the road driving... I'll PM you tonight. Thanks!

Any update?