10 coaches online • Server time: 04:36
* * * Did you know? The best scorer is Utep with 423 touchdowns.
Log in
Recent Forum Topics goto Post Things we dislike of...goto Post DIBBL Awardsgoto Post NO Seasons! 2016 ski...
Games-played based matchmaking?


Ok, so with this monster post, I will try to respond to Jan-Erik's post regarding the big benefits of using a match-maker that uses games-played as the primary pairing mechanism over TV.

Before I get to responding directly to the points, there are a couple of important scoping constraints here:

  • This is primarily for "early" games. 1 through 15 was brought up, but the actual limit could be adjusted if necessary; the point is that it's during the "build phase" and anything beyond that is considered out of scope, and could be done in any way (true random pairing, continued as-is or whatever).
  • Stunty teams are "designed to be bad" and any effect on those will be completely ignored. It's not necessarily that picking them is supposed to be bad, but to keep the discussion clear, they're considered out of scope for the context of this.
  • This is primarily for the Blackbox scheduler. Manual picked gamefinder games is not part of the scope here.

With that being stated, here we go :)

The four stated points are as follows, with my responses and thoughts for each point:

1. Prevents developed teams with high tv efficiency preying on more tv inefficient teams
This point targets the problem of long-lived teams doing "min-maxing" to sit at a low to mid TV range, getting paired versus teams that are newly started, building their team.

While this is possible to some extent, there are two things to consider:
  • This is currently limited within the Blackbox system, where teams within their first 15 games will not be paired against teams outside of that range. The window of opportunity for building a killer team and feeding on the new teams is pretty small.
  • The way the system is set up now encourages teams to think about development in both strategic (long-term) and tactical (short term) ways. For the BBT meta-event, and the loss of "protection" after 15 games, there is incentive to stay very TV lean for the first 10 or so games, and then build up relatively quick to get ready for the "deep end of the pool" once you finish your 15th game.

In a world where the number of games is driving the pairing and inducements are not perfect (very few people think they are); you get a situation where "fast developers" get either a big advantage (if inducements are undervalued), or "fast developers" have to not take the advancements (if inducements are overvalued). In the first case, the fast developers can prey on the slower teams while the other case is the same as where we are now and TV optimization is "king".

So in my opinion, changing to a game-based pairing system will not really "fix" this problem, and mainly just move it slightly to a different meta.

2. Prevents rookie teams from spinning into teams with a developed ballcarrier / killer / guardspam
I honestly have a hard time separating this point from the first one, except this has to do with specific perceived problems in the game rather than general "TV efficiency". Typically, TV-efficiency more or less means these things. A great ball carrier, 1-2 super killers, or a bunch of guard players on rosters such as CDs or Dwarves.

Maybe I'm not understanding the difference here, though, so I'd be happy to have it clarified. As a side-note and to be overly focused on details, the teams that minmax very rarely do so at TV1000k, and sit a little bit above that (say 1200-1300k or so).

3. Allows progression for tv inefficient teams like khorne or nobility where actually skilling certain players or buying positionals is detrimental for your winrate (and thus not progress at all)
This point is a little difficult for me to parse in general. The implicit premise is that increasing TV for these rosters makes them relatively speaking worse because the TV cost doesn't scale the actual playing power of the team as much as other rosters.

While I don't necessarily accept that it's as simple as "higher TV => worse efficiency" being axiomatically true for these rosters, I do accept that TV efficiency isn't equal for rosters as they play and improve.

Either way, though, the fundamental problem wouldn't change with a games-played based scheduler. These same rosters would be forced to take these "bad" skills and positionals to not fall behind even more than the TV based scheduler. The current system lets these rosters stay back for a little bit until they have enough SPP to "push past" this relative low-point.

Note also that this point implies that inducements are weaker than well spent TV, an assumption that we didn't make for point 1.

4. Promotes fast skilling teams and games to get a leg up
I'm struggling to come up with a response for this, because the point implies that "fast skilling teams" axiomatically need to get better in the meta. In my mind that's a very big claim to make and to base a benefit to a game-based pairing argument on. Fundamentally, this would be rosters such as the various Elves (48, 49, 49, 52 percent win rates currently), Skaven (55%), Humans (54%), and Underworld (63%). Probably other fast teams such as Lizards (49%) and Necromantic Undead (51%) as well.

My intuitive response would be "Why would you want to do that?" Most of the teams needing a "leg up" are relatively slow teams who already struggle getting SPPs. Having other rosters get an advantage would make the situation worse for those rosters.

If anything this feels like a gut reaction to "I don't like to play against Dwarves" more than an objective argument to improve the environment.


A closing thought:

Since the premise of these four points is to promote the suggested system as an improvement, there is an omission for how to deal with a few very real effects of changing to a games-played based system. As an example:
  • A team taking heavy casualties and *any point* during these first 15 games would put them in a situation where there's no coming back from. The next game will, on average, be tougher than the one you just played, and your team is worse off. This causes a death-spiral and you are better off retiring the team and not trying to recover it. Clearly, some people will still try, but the average number of recovering teams actively spinning will dwindle to near zero. This causes havoc in BBT, a system that is probably the best liked meta-event we've ever run on the site.
  • A concession will absolutely spike the winner of that match to the skies. You get an incredible amount of gold for that (and the overall assumption here is that TV trumps inducements). This causes problems for the meta, and gives the recipient of that extra boost an unfair advantage. With the current TV-based system, that boost is dampened to a great extent (but not entirely eliminated).
  • People in general are very averse to TV differences in a game. Anything above say 70k is considered game-breakingly unfair and unfun to people in general. Despite how bad TV is at comparing the relative strength of two teams, it's still the "gold benchmark" for fairness. Telling someone "well, you both have 10 games played; playing against that dwarf team is fair" won't make a difference if there is a 150k difference in TV.
  • "Lucky games" where you get a bunch of skills and gold will have a very strong impact on your future games as you'll consistently be playing against weaker opponents, which spirals away with more SPPs to spend. This is especially powerful early on where by game 10, you've effectively become the shark in the pond, much like what you were trying to avoid in the initial point.