16 coaches online • Server time: 07:48
Forum Chat
Log in
Recent Forum Topics goto Post Secret League Old Wo...goto Post Creating a custom to...goto Post ramchop takes on the...
SearchSearch 
Post new topic   Reply to topic
View previous topic Log in to check your private messages View next topic
sk8bcn



Joined: Apr 13, 2004

Post   Posted: Jan 01, 2009 - 15:44 Reply with quote Back to top

[quote="boch]What I´m saying is that this was a matchup on the verge of being unplayable. 15 TS is just a lot at that level of team development while a match between a team with 185 TS versus a 200 TS one is less unequal.

Therefor there might be a point in supplementing the hard cap of TS with a percentage cap as well.[/quote]

Well, question is:

Make a harder TS cap means a harder time to get a match up.

and 15 TS difference is not always an impossible match up.

So it's questionnable if a harder TS cap is a good thing. I've been non-scheduled once. It's frustrating to think that you have another 30 mins to wait.


By the way:
One handicap is 5 TS worth. IMO, if I get matched at -10TS+ 1 handicap, it's an ok match up. Cutting the cap makes recovery games even harder.

_________________
Join NL Raises from the Ashes
boch



Joined: Feb 15, 2006

Post   Posted: Jan 02, 2009 - 00:22 Reply with quote Back to top

I do see the point that a percentage cap can mean fewer games at low TS. On the other hand why use a hard cap in the first place? If a 15 % difference in TS is fair at TS 100 why isn´t it fair at TS 200?

Allowing a TS difference of up to 30 for a TS 200 team will certainly make it easier for the bot to set up matches and according to your argument that is the main thing isn´t it?

What I´m saying is that a hard cap at 15 TS leads to more unbalanced matches at low TS than at high TS.

By the way:
I don´t understand your comment on playing -10 TS and getting a handicap (for effective -5 TS?). I haven´t said anything about an effective TS difference of 5 being in any way problematic whatever the TS of the respective teams.
treborius



Joined: Apr 05, 2008

Post   Posted: Jan 02, 2009 - 14:03 Reply with quote Back to top

boch wrote:
I do see the point that a percentage cap can mean fewer games at low TS. On the other hand why use a hard cap in the first place? If a 15 % difference in TS is fair at TS 100 why isn´t it fair at TS 200?

Allowing a TS difference of up to 30 for a TS 200 team will certainly make it easier for the bot to set up matches and according to your argument that is the main thing isn´t it?

What I´m saying is that a hard cap at 15 TS leads to more unbalanced matches at low TS than at high TS.


I agree (and I think most do).

I think alot of the discussions about what would be a fair cut-off for creating matchups that are in a "fair range" are suffering from the fact, that people don't differentiate between the 2 fundamental (and not directly related questions):

1. Should the cut-off be an absolute or a relative value (relative to the team's total TS) or even a combination of both.

2. What should be the exact value (either in terms of absolute TS or % of team's TS)

Personally, I'd favor a combination of absolute + relative cut-off like cutoff in TS = 4 TS + 6% of team's avg_TS.

So, for example:

1. 2 teams with avg_TS = 100 -> cutoff = 10 TS
2. 2 teams with avg_TS = 150 -> cutoff = 13 TS
3. 2 teams with avg_TS = 200 -> cutoff = 16 TS
4. 2 teams with avg_TS = 250 -> cutoff = 19 TS
boch



Joined: Feb 15, 2006

Post   Posted: Jan 02, 2009 - 16:55 Reply with quote Back to top

That sound reasonable.

Just for the sake of the argument:

Why have an absolute component?

You could just go with a cap of for example 8%.
treborius



Joined: Apr 05, 2008

Post   Posted: Jan 03, 2009 - 12:08 Reply with quote Back to top

boch wrote:
That sound reasonable.

Just for the sake of the argument:

Why have an absolute component?

You could just go with a cap of for example 8%.


well, i think ideally everyone would like to have a max. ts-difference of something little like 1% or 1 TS, but that wouldn't allow many match-ups at all.

So what i think we should be aiming for is the "best" compromise of having no TS-diff. in every match-up vs. having every coach getting a game each time (s)he is looking for one.

Suppose there are about as many teams lfg in the, say, TS=170 range as there are in the TS=100 range...

Further, suppose for the TS around 100 - range it suffices to have a relatively small cutoff of 10 TS to match "most" coache's teams in that range.

Also suppose it is a similar advantage playing 110 vs. 100 TS as it is playing 187 vs. 170 TS (not sure this is true, but just suppose this for now), then we might not need to have a max. TS-cutoff of 17 in that range, but 10 would suffice (since we supposed, that there are about as many TS170-teams lfg as there are TS100-teams).

So effectively with an absolute component, the matching would get more precise the higher the teams TS is. (which, of course, would be desirable in general and especially for coaches who invested alot of effort in a team's development and rebuilding a team that was taken apart at TS=170 is harder than rebuilding at TS=100).

I suspect, though, that the amount of teams lfg is getting thinner with higher TS, so my shot was aiming at compromising those aspects as well.

(sounds a bit complicated, but i hope i could make it clear Wink )

EDIT:

To make it more clear, i think choosing a good cutoff is about maximizing these 2 (competing) goals...

1. probability of getting matched (rises with absolute TS-diff, rises with # teams available)
2. fairness of the matchup = relative TS-diff. (declines with rising relative TS-diff.)

...and also minimizing this one...

3. cost of rebuilding the team = expected # games needed to rebuild team (in terms of TR/TS/FF/gold) in case the team suffers from the fact that it got an unfair matchup. (rises with TS-diff, rises with TS)

EDIT2: 'reason why i think that cost of rebuilding a team (also) rises with TS is, that when your team get's mangled heavily it's often not so bad at low TR/TS because it might just take you 2..3 games to get the money and a few spp to replace your rip'd wardancer, while it might take 20 games at a higher TR/TS - range to replace that same wardancer and earn those 50 spp it had accumulated on its way up to, say, TS=200
TheCetusProject



Joined: May 25, 2004

Post   Posted: Jan 03, 2009 - 16:07 Reply with quote Back to top

For the point 3, it might be worth on the other hand bearing in mind that a high TS team probable has enough players that a serious mangling won't leave them too understrength, whereas a mangling for a new team could well be an invitation to a long sequence of manglings.

Also slightly confused by why you think that your absolute + relative combination gives more potential matchups... if you calibrate a relative system so that TS 100 can play up to TS 110, then in a relative-only system, TS 150 can play up to 165, 200 up to 220, and 250 up to 275, all greater ranges than your suggested system.
treborius



Joined: Apr 05, 2008

Post   Posted: Jan 03, 2009 - 17:28 Reply with quote Back to top

TheCetusProject wrote:
For the point 3, it might be worth on the other hand bearing in mind that a high TS team probable has enough players that a serious mangling won't leave them too understrength, (...)


understrength is really relative though - if you play a team around TS=250 loosing 1 Player with >50 spp will probably still leave the team "very strong" (in absolute terms), but severly weakened (in relative terms when compared to its original strength).
Also, the amount of effort to regain that original strength is probably a lot higher on average for a high-TS-team that lost one of its good players.

TheCetusProject wrote:
whereas a mangling for a new team could well be an invitation to a long sequence of manglings.


i think you're forgetting the fact that we're talking about Black-Box where you face a team of (ideally) equal strength each time you play a match so when you lost 1 player at TS=100 which was worth TS=10 then you'll face an oppo who's TS is also around 90 next game.
(I'm not saying it couldn't happen that you loose another few players until your team is almost dead, but it's more probable that you'll pretty soon return to your original strength than with a high-TS-team, i think - mostly it's "just" about getting the gold to afford new players)

TheCetusProject wrote:
Also slightly confused by why you think that your absolute + relative combination gives more potential matchups... if you calibrate a relative system so that TS 100 can play up to TS 110, then in a relative-only system, TS 150 can play up to 165, 200 up to 220, and 250 up to 275, all greater ranges than your suggested system.


i think you misinterpreted my post(s) - i didn't claim that "my" combination of rel. + abs. TS-value would produce more potential match-ups and especially didn't claim that my example of having a cutoff of 4 TS + 6% TS gave more match-ups, but was claiming that that kind of combination was a better shot at compromising 1..3 with the values 4 TS and 6% TS just being fairly close to what we currently have (fixed cutoff of 15TS) - a bit lower at lower TS-ranges and a bit higher at higher TS-ranges - just giving an example of what i was aiming at in principle...

...maybe something like 3 TS + 5% or 6 TS + 8% TS would be a better pick of values, though - it was just an example of having rel. + abs. components in general...

(i also picked those numbers, because i think that 15 TS is too high for low TS-ranges, but could be a bit higher at higher TS-ranges because currently match-ups are probably less frequent at >200 TS due to a lack of teams in those ranges - BlackBox is only a few weeks old, afterall)

EDIT: a few edits due to entangled language Wink
TheCetusProject



Joined: May 25, 2004

Post   Posted: Jan 03, 2009 - 17:47 Reply with quote Back to top

For the first part, my thinking was linked to my perception that TS doesn't handle under-strength teams too well, and doesn't really appreciate that playing under-strength is quite bad for your team's long term health (since TS doesn't care about team health). If you lose a few players in your first blackbox match then you have a few Ogre beatings to look forward to.

For the second part, I will do some calculations in a bit to see the relative merits of the various posible systems.
treborius



Joined: Apr 05, 2008

Post   Posted: Jan 03, 2009 - 18:23 Reply with quote Back to top

TheCetusProject wrote:
For the first part, my thinking was linked to my perception that TS doesn't handle under-strength teams too well, and doesn't really appreciate that playing under-strength is quite bad for your team's long term health (since TS doesn't care about team health). If you lose a few players in your first blackbox match then you have a few Ogre beatings to look forward to.


if you're talking about TS<100 teams: i don't know if TS-calculation is less valid in that range, but if you get mangled far below TS=100 it's always a good guess that you'll be back at old strength with a new team w/o any recovery-matches required - so possibly only 0 matches to get back to where you started from (in terms of TS) - not too much of a rebuilding effort (except for creating that new team) Wink

TheCetusProject wrote:
For the second part, I will do some calculations in a bit to see the relative merits of the various posible systems.


sure, looking forward to that...
TheCetusProject



Joined: May 25, 2004

Post   Posted: Jan 03, 2009 - 18:31 Reply with quote Back to top

Boo to retiring injured new teams! Teams should only be retired because of erratic mood swings, not results!
treborius



Joined: Apr 05, 2008

Post   Posted: Jan 03, 2009 - 18:39 Reply with quote Back to top

TheCetusProject wrote:
Boo to retiring injured new teams! Teams should only be retired because of erratic mood swings, not results!


i boo 2, but fact still is that it doesn't take any effort to be back at where u started from Smile
Display posts from previous:     
 Jump to:   
All times are GMT + 1 Hour
Post new topic   Reply to topic
View previous topic Log in to check your private messages View next topic