Janus
Advocatus Diaboli
- Joined
- Sep 9, 2007
- Posts
- 23,430
- Reaction score
- 57,318
- Location
- Kansas City, Missouri
- AFL Club
- Port Adelaide
- Other Teams
- Dallas Cowboys, Chicago Bulls
Yeah it's a reasonable point re Collingwood.
Looking purely at "Best 22" we only have the Pies ranked 7th yet the season projections have them 3rd on average. The reason is two fold. Firstly: we have a random injury generator in the simulations and we rate Collingwood's "23-30" on list higher than any other side leaving them well placed on depth. Secondly they have a friendly draw (Carlton and Melbourne twice is a huge bonus).
It's so volatile though - the fact that we could have Port dropping from 2nd to 7th (albeit effectively equal 4th) on the back of the removal of just 2 players (neither of whom we rated as elite) is a fairly good demonstration of the uncertainty of predictions and impact of season ending injuries (or conversely a player having a break out season that lifts their rating considerably).
So really all these predictions can be taken with a grain of salt (which is a shame for me because the simulation model took weeks to set up!)
I'd probably tweak the injury generator then. You could probably do a sliding scale of probability - starting with 4.5% chance in R1 (when everyone is at peak conditioning) and then gradually ramping it up by that percentage each round once the wear and tear of the season takes its toll on hamstrings etc. It makes logical sense to do it that way - and also increases the possibility of simulating older players being rested for general soreness in later rounds. If you wanted to get really technical you could even do a partial "reset" when a team has the bye.
Don't know how hard that would be though.












