Breaking down Atlanta United player ratings so far this season

A request in a comment to the most recent staff player ratings (h/t: PenthouseTiger) gave me the idea to pull together all the community and staff player ratings for every game to date this season and see how we feel the team has been progressing. This was an exercise I undertook just for kicks, but it ended up with some surprising results.

The first thing that drew my attention was that only 3 players have started every game this year. I had expected it to be a few more than that, especially in view of Tata’s evident preference to maintain a consistent first team. In fact, Atlanta United has had a total of 19 different starters this year, and only 13 of those have started 10 games or more. A total of 15 players have made substitute appearances, including 5 who have not had any starts.

So who are those 3 unbenchable players? Well, they are Jeff Larentowicz, Miguel Almiron and Josef Martinez. At the least I had expected Michael Parkhurst to be on that list. He has in fact played in all 22 games, but came on as a substitute in the season opener against Houston after only 39 minutes. He was also the community Man of the Match for that game (no staff MOTM was awarded given the disastrous performance).

Despite not starting that one game, Parky is the team leader in minutes played at 1,940, beating Miggy by 12 minutes. Since that first game, he has missed only 1 minute (he was taken out late against FC Dallas).

Taking that a step further, here is the lineup based on most minutes played:

The formation is mine, but it’s not exactly shocking that the preferred players fit neatly into the 4-2-3-1.

As far as substitutions go, the three most regular subs are Kevin Kratz (18 appearances off the bench), Romario Williams (12) and Tito Villalba (6).

So, we have 19 guys who have started games. How well did we all think they did? The table below shows the average rating for each player (based on his aggregate rating divided by the number of starts):

Note first that there were no staff player ratings for the Portland game as yours truly was on vacation, so some of the difference is accounted for there (the staff ratings are based on one less game each for the 11 players who started that game). Overall, the community and the staff seem largely to be in agreement. The biggest disparity is for the now-injured Sal Zizzo, but since he only had 1 start that’s to be expected. In contrast, on Chris McCann we are in infinite accord: his average rating on both scales is 6.681818 recurring.

These numbers also indicate that we all feel that the squad is pretty consistent in its quality. The standard deviation for the community ratings is 0.45 and for the staff ratings 0.35. Remove the Zizzo ratings and the results are even closer (0.33 and 0.35 respectively). Those are fairly low numbers and given that we have a few superstars on the team, it means that those stars have an excellent supporting cast.

I also took a look at how the 4 most frequent starters did from week to week. Here’s the chart for the staff ratings:

I omitted the Houston game from the chart, as it was a complete outlier for the entire team. Interestingly, each of the four players has had one truly mediocre game. However, the trendlines show that Parky and Miggy are fading slightly, whereas Jeff is Mr. Reliable and Josef is improving. That last is certainly to be expected with Josef on a tear right now. Parky’s decline might be attributable to fatigue, if it weren’t for Jeff having such a flat trend and only 91 fewer minutes played. Of concern is Miggy, whose negative trend is much sharper. Again, this is not unexpected as his performances of late have definitely not been up to his usual high standard.

The same chart for the community ratings is below. It excludes the Portland game to make it comparable to the staff chart (all four players had a moderately off day, as it happens).

Not dissimilar results, especially as it relates to the trends.

So: a nice tidy consensus. Which means either we are all perfect soccer analysts or none of us has a clue what we are talking about. At any rate, let us know what you think of this in the comments, and if there any further data points you’d like us to include in future editions of this, give us a holler on that too.