All the stats that matter!

Here are are stats for the Round 4 game against UTAS

It is important to note that these stats aren’t 100% accurate.  This is because 1) It was the first time we used the software, and although it’s easy to use, we made a few silly mistakes.  2) We finish the game midpoint.  Nowehere else in the world does an ultimate game finish mid-point, so the software is not designed to handle it.  We have to put in a “fake” goal for UTAS at the end to keep some stats correct, but this ended up compromising the PM and RPM scores (probably the most important ones).

Even if the scores were accurate, it’s important to remember that it’s just one game, and the main reason we’re playing with this “stats” thing is to look for long-term trends and weaknesses in our disc usage.  Drawing conclusions from one single games’ stats is not really very useful.

Column Explanation

G = Goals Caught

A = Assists (Goals Thrown)

Points = Number of points you were on the field for.

PM = Plus/Minus.  You get a plus if we scored while you were on the field, and you get a minus if they scored while you were on the field.  This is used as a “general effectiveness” measure, to acknowledge those players who play well but don’t get the goal or the assist.

Touches = How many times you touched the disc on offense (either caught then thrown, OR dropped a catch)

Turns = How many times you turned it over from a throw (either your throw was a throwaway or you dropped a catch)

Retention = Turns/Touches (as a percentage, how consistent are you)

RPM = Relative Plus/Minus – In Ulti there are people who do great D and are very effective players, but still end up being on the field while they get scored on.  RPM attempts to “Normalise” the Plus/Minus using some complex calculations to give a more accurate figure of how useful the player is on O and D.  This is broken down into the next two columns, oRPM (offensive) and dRPM (defensive) which are added together to give the overall RPM.  See the next section for a description

Ds – how many Ds you got (got the disc when we’re on defense)

For full details on RPM and how (and why) it’s calculated, go here:

http://groups.google.com/group/rec.sport.disc/msg/fb9f0f2e03f06a79?pli=1

Here’s some quick points pasted from that website.

Initially, plus/minus looks like a promising measure, but as noted, it has many flaws.  Mostly, it doesn’t take into account how difficult the opponent, and either rewards or punishes players based on the number of points played.  Example:  If you win a game against an weak opponent 15-0, the players who play more (which may not be your best players) have a higher plus/minus.  If you then lose 15-0 to a strong opponent, those who played more points (perhaps your stronger players), will have a lower plus/minus.

…
Specifically, with RPM, I believe I have found a way to remove the skew caused by the number of points played, the skew caused by the final overall score, and the skew caused by the difference between starting on offense vs. starting on defense.  The result is a number that represents a player’s plus/minus relative to that of the team average, calculated separately for offense and defense before generating the total sum.
Regardless of the final score, you will always have a positive RPM (even if your total plus/minus was negative) if your team’s plus/minus was better with you on the field.  In fact, in any game, because the RPM is relative to the team average, the sum total of all the RPM numbers for the team will always be zero, regardless of the margin of victory or defeat.

…
I do feel that while this stat may have some randomness on a game to game basis, over the course of a season, it might begin to represent some of the intangibles that don’t show up in other places on the scoresheet.  If someone if consistently positive, perhaps they are having a positive influence on the game when they are on the field, to the point that we score more often when they are on the field?

…
When I look at the numbers for a game, I tend to ignore any numbers that are in the -2 to +2 range, assuming that this is down in the noise.  However, when I start seeing larger numbers, I begin to think that perhaps it might be more than just randomness.