09-27-2012, 10:00 PM
|
#8
|
|
Hall Of Famer
Join Date: Apr 2010
Posts: 2,877
|
Quote:
Originally Posted by JeffR
We're just getting to the point of figuring out roster management AI, so I can't really say anything concrete on the issue because I don't know what the guys have in mind. But here's how I'd frame the problem; it's much more complex than it seems:
First, there are two different situations here: under-20 and 20+ players. The under-20's are a distinct case because, although they're signed to an NHL contract, the contract "slides" if they don't play professionally in North America, so their three years doesn't start ticking until they join the NHL team (or one of their farm teams.) So there's a decision process for 18-19 year-olds:
1. Is the player good enough for the NHL? Easiest solution if this is true, he stays with the NHL team.
2. If that doesn't apply, can he play in the AHL? If he played major junior, the answer is no, so the choices are either back to junior or going to Europe. The latter is rarer, but happens in a case like Nicklas Jensen where the team thinks the player is at a point where he won't develop as much in junior.
3. If he can play in the AHL, is there room for him? If he's going to get stuck on the fourth line or in the ECHL, it probably makes more sense to place him elsewhere. EHM didn't make good decisions in this regard.
4. If the AHL is out, Where does he want to go? This where EHM really broke down: if a major junior team held his rights, he'd always get sent there, even if he hadn't played for them before and had a different preference (Europe/college.) What should've happened is a check on the player's wishes re: major junior or Europe.
For 20+ players, it's a bit different, and harder for an AI to handle. The central issue is burning a contract year to send him back to Europe, where he's out of the team's developmental control. The real-life reasons are a bit more varied, and it's not a simple checklist as above. Maybe the NHL team have a lot of AHL depth at the player's position and they think he'll get more playing time and/or better competition overseas. Maybe the player preferes remaining in Europe (that one's tough to model - how do you make a guy willing to sign an NHL contract but reluctant to play in North America?) Maybe he's no longer in the NHL team's plans.
So, the problem is a lot muddier. I don't have an instant answer on how to resolve it: hopefully building an AI that manages its farm resources sensibly will be enough to produce realistic results for the 20+ guys. If not, then it becomes time to start making specific code to deal with this situation (and at that point, it's a decision whether spending time working on that takes priority over some other area.)
|
Speaking of roster management, another this EHM was terrible at was controlling it's minor league rosters. Will you guys be trying to ensure the AI in FHM will be able to understand the veteran and developmental player roster rules?
|
|
|