Points per Minute (PPM) - Part 1

There is one truism that everybody knows about Supercoach: a player can not score points if he doesn't get on the field! Subsequently, what this means is that a player who is on the field for a longer period is more likely to accumulate points. However this doesn't mean that any 80 minute player will score double that of a 40 minute player because it is all relative to what position they play - for example a winger versus a prop.

The following analysis on PPM per position (FRF, 2RF, 58th, HFB, CTW and FLB) will highlight these differences and provide a platform for comparison amongst each group. The data is for all players listed in those positions in Supercoach 2011.

The concept of PPM is not new and neither is it very complicated. Points per Minute = Total Supercoach Points / Total Minutes played.

On the face of things, this is a very powerful tool that can be used to a) forecast how a player's expected score will change given a change in the average minutes played and b) compare players within each position.

Intuitively, it is hard to argue that there is a direct relationship between minutes and points scored. The scatter gram below illustrates this:


Applying a statistical technique of linear regression on the data from the 2011 season, we get an estimated PPM of 0.6641 for all players which is represented by the linear slope through all the data points. Looking at the graph, however, we can see that the data fans out which means there is something else other than minutes played affecting a players ability to score points. Statistically, the estimated PPM is meaningless because it breaks one of the important rules.

We know what the problem is - you can't compare a CTW versus a FRF and so on! It is necessary that we analyse the data separately and we can do this by separating them per position each player was available for selection. An inherent problem with this is where there are dual positioned players such a centre who is listed as a dual 2RF/CTW. These players will be included in both sets of data and will present as some problems which I will address in a separate post. For now, this level of granularity I think is sufficient.

So lets look at each position separately then:

FRF

The shape is not quite linear and displays a diminishing return but generally looks better than the previous graph. The main problem on the long end are dual positioned FRF/HFB and hookers that are classifed as FRF.

The estimated PPM for FRF is 0.9048. Most props only average around 40 mins so the actual PPM for props is probably alot higher than this which Ill address at a later post.







2RF

The regression here looks a lot better with the exception of the longer end of the graph. Again, the main problem is dual positioned CTW/2RF but the estimated PPM is generally a good fit.

The estimated PPM for 2RF is 0.7278







58th
HFB
CTW
FLB

So why is all this important?

Ultimately a player's value will rise if he scores consistently above his average and since we have proved that minutes directly affect a player's score, we can use PPMs to find those that are likely to increase more than others if we believe that they're average minutes will change during the course of the 2012 season. Our task as supercoaches is like that of investment managers: to make sound investment decisions that will yield a higher return based on our initial outlay and known risk factors.



TO BE CONTINUED.

Is it 2012 yet?

I promised to post PPMs in my previous post and the work is done but I just haven't had the time to tidy it up.

The one thing that I have done however is to do a Bye Planner after the release of the NRL 2012 fixtures.

Check out the page up top and tell us what you think you will do with your initial squad!