Strategy III - Team roster management

This is the third of a series on long term strategies. The first two are listed here

The first article discussed filling the roster with mid-range mid-dollar role model FA to fill holes so one can use BPA in the draft. The second focused on how to use inefficiencies in the draft chart to minimize risk and maximize value. In this article, I want to expand the discussion to the existing roster.


OCC has published several articles last year and again this year based on how our players match up at their position compared to the other players at the same position. He used the PFF rankings for each player and converted them to percentiles. He then grouped the players into five groups color coded to make it easier to see.

This is similar to what I suggested last year but with six groups [or sextiles – fancy word for six groups]. I chose six based on the range from best to worse is equal to 6 groups with three below the average and three above.

In statistics there are several rules of thumb: Recall the properties of the bell curve and the probabilities from a standard normal distribution. One feature has to do with the amount of data that falls within a certain number of standard deviations:

Approximately 68% of the data is within one standard deviation (higher or lower) from the mean.

Approximately 95% of the data is within two standard deviations (higher or lower) from the mean.

Approximately 99% is within three standard deviations (higher or lower) from the mean.

We get the same point basically using five groups of 20% intervals and it is easier to see and use.

Key points

The key takeaway is that we have a range of talent on this and every team. We have stars and folks in each quintile [fancy word for 5 groups]. Everybody wants the stars and nobody wants the last players in the last group. We are going to focus on the latter group today.

Every team gets about the same amount of money to spend each year. What a team does with that money is up to them. This can be an advantage to a team with a better front office strategy

The first thing to do each year is evaluate the talent and estimate what the players will do in the future. Note the percentiles are used as the starting point.

Teams should rid themselves of the lowest scores. This is not rocket science, and I think everyone already wanted to do that.

Yet ridding the team of the lowest scores is not the same as same as the ridding the team of the lowest grouped players. Costa was not ranked well last year, but the team had higher expectations for the future. He was bad early in the season last year. One vividly memorable game last year against NE singlehandedly moved him down 6 ranks by itself.

Divide his season into three parts and we see

Early – terrible

Mid-season with Holland – average

Late without Holland but the work with Woicik kicking in – average.

Note how much he improved as the season went along. Further, the coaches saw his determination to improve in the off-season. That and how well Kowalksi played at center as a UDFA last year showed why the team did not invest in a center in the off-season. When both were injured and basically unavailable during training camp, the team traded a 7th round pick to get Cook as insurance.

Cook played well individually, but did not know the playbook or even the terminology. The OL was not used to him and suffered initially, but got much better over time. I shudder to think what the season would have looked like without Cook.

Yet, in the future, I envision Cook more of the Proctor utility game day backup. That has value as he has started at tackle, guard, and center. Next year he will be 30. Right now that is the 3rd oldest of the OL, behind Dockery, who will not be back, and Livings. Otherwise the OL is still very young and could use some veteran presence.

Costa only played three games this year or about 10% of the total snaps. In the first game, he had three snaps, before we brought in Cook. I suspect that Costa rushed back too soon as he really was not healthy in TC.

After being out of the first game, I suspect that he rushed back during the season too after we saw how poorly the entire OL played with timing off with Cook initially. Costa played one and half more games.

In the first game, he was more dominant than he was as bad against NE last year. His rating in that game alone would have been good enough for a season ranking of about 19th for all centers. PFF wanst 25% of the snaps to avoid small sample size to be ranked. Yet, Costa surprised even detractors and demonstrated the faith of the coaches.


Backups will have fewer games to be rated by PFF or have played fewer snaps in each game. In each case, their game scores and cumulative scores should be less than full time players who played at a high level.

Yet the backup’s scores hint at more playing time the next year. Last year we noted that Church and Lissemore would be key players this year. They were until injuries changed how we used them. Developing young new players is a key strategy for a good team. Teams always have to be getting younger, cheaper, and healthier.

A team with lots of quality backups can rest their starters, which allow them to play at a higher level. Bill Parcells always wanted to rest Ellis, but who did he have that was a quality backup.


Some players would be ranked higher but the PFF scores were limited by their injuries or other reasons to miss games. PFF adds each game score together to get a cumulative score so miss a few games and your score can be artificially lower and thus the rankings will be affected negatively. One good example is Bernadeau played two games at center, so he was not evaluated at guard. He had only 14 games at guard to be scored and was lower in the cumulative score compared to his peers.

Further, the scores and rankings themselves have to be assessed to what the coaches expect in the future. Again Bernadeau is a good example. His scores started slow, improved, dropped significantly and then improved again. The good scores are off-set by the bad and overall they were not good. Yet an examination of the season should note the following that correspond with the trends and inflection points.

He had off-season surgery and did not practice until late in training camp. His first few games were disappointing. Yet as he recovered his scores improved significantly. He was playing at a high level until he was moved to be the emergency fourth string center. He moved back to guard, but just like reading a book and getting interrupted, one does not start off exactly where one left off. He improved consistently until the end of the season where again he was playing at a high level.

I suspect that the team expects Bernadeau to start next season at a high level. He was the first FA signed last year as the scouts thought he was ready to break out. We saw glimpses of that and the trend was in the right direction.

Bernadeau is just like another player Costa, who had improved over the last season. Yet their image in early games seared an image into many folks’ minds. Initial impressions are hard to overcome in fans minds.


The first step of roster management is to evaluate the play of each player. The natural step is to remove the lowest scoring and upgrade them. Yet one way of upgrading is to develop players. Backups get more snaps and their rankings improve. All players rankings improve they play more games. Teams must be careful to evaluate the entire season and look for trends. Inflection points are indicators to look for deeper causes of action.

In the next part we will discuss other ways to assess the roster using this method.

Another user-created commentary provided by a BTB reader.