|
Braavos36389 Posts
![[image loading]](http://www.teamliquid.net/fantasy/proleague/Images/Banner.jpg)
This is a poll to determine which scoring system is more popular. Feel free to discuss how to improve the FPL system here.
1. "Old" Scoring System - 09-10 PL R1
Win = 2, Loss = -1, Lineup = 1, Ace Selection = 2, Ace W/L = 2/0, Team Win = 1
Original scoring values Lineup and TW. Also doesn't penalize weak players for getting picked as Ace Match. This structure puts value on Ace Selections and makes ace players from bad teams viable options and mediocre players from good teams an OK selection because TW gives them a real boost.
2. "New" Scoring System - 09-10 PL R2
Win = 4, Loss = -2, Lineup = 0, Ace Selection = 2, Ace W/L = 4/0, Team Win = 1
More emphasis on Win/Loss and less on Lineup and TW. Thus, your Anti-Team picks can go positive. We thought this system would turn into more of a "merit based" system, but it turned into rewarding the elite players even more (no more lineup and team wins). This isn't good because it means teams are less diverse (less non "top tier" players). This probably puts too much weight on ace players too, but we're not sure if TL users enjoyed this system more than the original.
3. Hybrid Scoring System - ??
Win = ??, Loss = ??, Lineup = ??, Ace Selection = ??, Ace W/L = ??/??, Team Win = ??
Possible hybrid arrangement, propose your own and give reasons.
Thanks for your feedback! Signups will be open soon.
Poll: Fantasy Proleague System?
(Vote): System 1 - "Old"
(Vote): System 2 - "New"
(Vote): System 3 - Hyrbrid
|
Hyrbrid! Woo!
I like the new one, though both are fine lol.
|
Braavos36389 Posts
Oh I should say, the system will not 100% be chosen by this method, but we will take it into consideration.
|
Baa?21244 Posts
Hybrid; I like the second system with the exception of the fact that someone who goes 1-1 with the win coming from the ace, he gets the same amount of points as someone who goes 1-0 non-ace.
That should be adjusted imo since an ace win should be worth more than a normal win, regardless of whether he lost an earlier game.
|
System 2, but only if you take out the points for Ace pick.
Roro not losing points despite losing twice to JD was such a ridiculous sight that I never want to see again.
|
If you are going to do the new system, you should make a team win worth 2 points, or half a player win, same as the old system for continuity's sake, and to make good teams worth picking up. Otherwise its better to take ace and get an extra point just so you can go after another top player.
|
I liked the old one better, but we really need bad ace players to be punished (HELLO RORO), like +2 for ace, -1 or even -2 for ace loss.
I'd really like losses to be punished more, 9-7 players shouldn't finish with more points than 6-2 in my opinion.
Players from the main team should finish with low points if they are underperforming.
Hell in fact I should have voted hybrid, with wins and losses being the same amount of +/- points. TW and Lineup being worth 1 point each. (Hell, team loss could even be worth -1, but.... hmmm no, a good players shouldn't be THAT handicapped by his team losing, since it's not winners league)
Scores would be muuuuucccchhh tighters, but hey ! Players on a roll would be really rewarded.
|
Too tired to think about hybrid system atm, but 2nd scoring system was really bad, mostly becouse of things mentioned by you.Either someone can come up with something nice in the hybrid or just stick to 1st one imo.
|
Eh, I personally disliked the second one because the elite players did far, far better than other players. There was no chance in catching up if you didn't have them.
|
both are a bit flawed but i cant tell whats wrong.
|
I preferred the old system. The new one puts too much emphasis on ace selection, making the aces of mid-tier teams a lot more valuable than their 5-6 point peers that are on strong teams and don't get sent for ace matches. I like that smart anti-team selection can actually earn you points, but the system overall isn't worth it for less diversity in teams.
I actually think the new system would be better if ace win/loss was 4/0 and the ace selection points were removed.
|
Old system is far better in terms of requiring more overall prediction skill, it's a no-brainer as far as I'm concerned.
|
i like the old system better overall
but this poll seems really biased 
|
Baa?21244 Posts
Removing ace selection points is ridiculous, a player getting picked for ace should definitely be rewarded.
|
i think it'd be better if
Win = 4, Loss = -2, Lineup = 0, Ace Selection = 2, Ace W/L = 2/0, Team Win = 1
since you already get +2 points from ace selection so there's less emphasis on top tier players
|
United States2186 Posts
System isn't as important as the ace part. ace losses must count as normal losses. Just looking at the stats page here shows that there were a lot of people with highly inflated ranks simply because they ended up losing in ace matches instead of normal matches. Leta being 8 points higher than Flash is nonsensical, along with Zero and Roro having very high scores despite going not that well above 50%.
http://www.teamliquid.net/fantasy/proleague/Stats.php?r=5
The ace selection bonus should also be less, not sure on 1 or 0. Guaranteed ace players (Flash, Leta, Jaedong for example) simply heavily outvalue players on teams with 'split' aces like STX, MBC, SKT1 because they have an opportunity to make major point gain (almost as much as winning the match anyway) and at worst lose nothing. This does hurt Air Force Ace and the weaker teams and hopefully at least for ace that could be alleviated with adjusting point values?
As for teams as long as they don't have the hugely inflated scoring system from the one round where the SKT1 joke team nearly won. Overly high team values make it a lot more boring when 1 out of 7 choices is so much more important than the rest.
Also are you guys going to consider changing the selection format at all? There was one idea that gave the choice of being able to pick more than 6 players but paying a -1 penalty for doing so each time. Another promising one let you overstack the anti team in exchange for getting more purchasing power, so making a 20 point anti team would give you 37 points in your main selection. Each would probably encourage a lot more diversity since so many options open up.
|
i think new is "more fair", but old is "easier" to play. I voted for old, i guess that shows somethign about me
|
Braavos36389 Posts
On April 06 2010 07:37 Ver wrote:System isn't as important as the ace part. ace losses must count as normal losses. Just looking at the stats page here shows that there were a lot of people with highly inflated ranks simply because they ended up losing in ace matches instead of normal matches. Leta being 8 points higher than Flash is nonsensical, along with Zero and Roro having very high scores despite going not that well above 50%. http://www.teamliquid.net/fantasy/proleague/Stats.php?r=5The ace selection bonus should also be less, not sure on 1 or 0. Guaranteed ace players (Flash, Leta, Jaedong for example) simply heavily outvalue players on teams with 'split' aces like STX, MBC, SKT1 because they have an opportunity to make major point gain (almost as much as winning the match anyway) and at worst lose nothing. This does hurt Air Force Ace and the weaker teams and hopefully at least for ace that could be alleviated with adjusting point values? As for teams as long as they don't have the hugely inflated scoring system from the one round where the SKT1 joke team nearly won. Overly high team values make it a lot more boring when 1 out of 7 choices is so much more important than the rest. Also are you guys going to consider changing the selection format at all? There was one idea that gave the choice of being able to pick more than 6 players but paying a -1 penalty for doing so each time. Another promising one let you overstack the anti team in exchange for getting more purchasing power, so making a 20 point anti team would give you 37 points in your main selection. Each would probably encourage a lot more diversity since so many options open up.
Both options are good ideas, but we currently don't have the programming manpower available to make large changes like this. Maybe for R5.
I think the biggest issue is the lack of diversity in choosing non-elite players, as everyone has some version of JD/Flash/Leta/Bisu (or whoever is good) and then some middle pointers. Its hard to devalue ace selections because this also hurts the middle/low players who get picked. Not sure how to fix this or "power up" middle tier guys. Maybe we'll just overprice the upper tier guys a lot.
|
I propose Ace Selection = 0, Ace W/L = 2/0
Here is my analysis on the status quo:
Who does it reward and who does it penalize?
The current system rewards anyone who is going to be thrown out as ace. The top few teams are going to finish a lot of their rounds with 3-0 or 3-1, so they wouldn't get to send out an ace player. The bottom few teams who are going to finish a lot of their rounds with 0-3 or 1-3 are in the same boat. The teams in the middle however, are going to have an ace match nearly every time, so their ace player will be getting a lot of points. Also, players on a team with multiple good players such as STX with Kal/Calm and MBC with Light/Sea will be affected as well. Those players will have less ace matches than teams with only one good player. So under the status quo, players like Really and Leta, who are the one good player on a mediocre team, will be getting significantly more points than they deserve, while players like Kal, Calm, Sea, and Light will not be getting as many points as they should be.
Why Ace Selection = 0, Ace W/L = 2/0 is ideal
The ideal system results in every player's points being a perfect reflection of their skill. Under my system, a player like Really or Leta will have a small advantage over Kal or Light in that they have more ace matches, but Kal or Light will get more points in Team Wins. So overall, there is very little discrepancy between two equally skilled players where one is the ace player on a mediocre team, while the other is a good player on a good team. This encourages people to select players based on their skill rather than what team the player is on.
|
i like the new system, it values the win a lot (and makes possible to get points with your anti-team)
perhaps an hybrid system could be the best, some fine balancing can be done. But if this next round is by the new system, it would be fine by me
|
On April 06 2010 07:30 integral wrote:
Old system is far better in terms of requiring more overall prediction skill, it's a no-brainer as far as I'm concerned.
I think the opposite. The optimal strategy in the old system is simply to pick players who get a lot of lineup appearances for your main team and ones who don't play for your anti-team. Neither is very hard to predict. The variance is fairly low considering the best progamers win 70% and the worst still win 30%. The new system has a much higher emphasis on predicting who's actually going to win and lose. As others have pointed out, the added variance means that if you don't pick one or two specific players, you probably have no chance of winning, which kinda sucks. I like the idea of using the new system with ace points toned down.
|
an ideal thing i can think is using the new system with a better balancing trade. The problem of the new system is that once one of your player starts losing, you become more and more stuck with him
|
i would prefer the points given to be the same for players winning and losing,example
if he wins
win +2
appearance +1
team win +1
ace win +4
if he lose
lose -2
appearance -1
team lost -1
ace lost -4
|
My hybrid:
Win = 3, Loss = -1, Lineup = 1, Ace Selection = 2, Ace W/L = 4/-2, Team Win = 1
|
Why don't we remove Ace Selection as a whole since it's redundant? Instead of Ace Selection = 2, Ace W/L = 4/0, we do Ace W/L = 6/2.
We can't remove lineup appearance in the same manner because it makes a difference to the 4th players that didn't get to play because it went 3-0.
|
I like whichever systen gets me a better placing haha. Seriously thought I like the old one.
|
i picked the old one because i pretty much did the same thing both times and got a better result with the old one.
|
Old is much better. One bad night won't screw up your FPL entirely.
Damn you Guemchi!!!
|
I personally liked the old system better (the numbers felt really inflated in the new system), except I would also like to see something happen in the case of an ace match loss. It was just ridiculous that 2 seasons ago players like Roro and Zero did really well points-wise even though they went 0-3 and 1-3 respectively in ace matches (and finished 9-6 and 8-6 respectively overall but were #6 and #11 points-wise).
The whole point is, just give the ace player an extra point or two for winning the ace match. There doesn't need to be any bonus points at all if you go to the ace match and lose.
EDIT: I picked the first scoring system proposed that I saw in the thread:
On April 06 2010 08:27 Grobyc wrote:
My hybrid:
Win = 3, Loss = -1, Lineup = 1, Ace Selection = 2, Ace W/L = 4/-2, Team Win = 1
My problem with a setup like this is that it puts so many points into those ace matches. This is still better than the players gaining a bunch of extra value just because they lost an ace match, but should Flash really be scoring 10 points/night just because he always gets an ace match? I don't see why it shouldn't be a pretty small bonus for going to/winning the ace match since that match is almost always icing because that player has already won a game earlier in the match. The only team that I can think of who doesn't tend to play their ace match players in the regular lineup is FOX, so it seems like this would only result in Flash/Jaedong/Stork/etc being less popular (which is what we want, isn't it?).
|
On April 06 2010 05:49 Hot_Bid wrote:
1. "Old" Scoring System - 09-10 PL R1
Win = 2, Loss = -1, Lineup = 1, Ace Selection = 2, Ace W/L = 2/0, Team Win = 1
2. "New" Scoring System - 09-10 PL R2
Win = 4, Loss = -2, Lineup = 0, Ace Selection = 2, Ace W/L = 4/0, Team Win = 1
3. Hybrid Scoring System - ??
Win = ??, Loss = ??, Lineup = ??, Ace Selection = ??, Ace W/L = ??/??, Team Win = ??
Possible hybrid arrangement, propose your own and give reasons.
I just wanted to put up some examples of what these numbers work out to for certain player stereotypes:
Team Ace ( Flash, Leta)
A good record for an ace player would be 10-3 (2-1 in ace games) and, say, his team goes 8-3.
Old: 20 for wins + 10 for appearances + 6 for ace appearances + 8 for team wins - 2 for losses - 0 for ace loss = 42 points for the round.
New: 40 for wins + 0 for appearances + 6 for ace appearances + 8 for team wins - 4 for losses - 0 for ace loss = 50 points for the round.
-> The new system scores noticeably higher (+19%) for this record.
Workhorse (Really, Mind, HiyA)
I'd say a decent record would probably run to 7-4 for this type of player, giving him a 1-1 ace match record just because. Because they tend to be on all sorts of teams, I'm awarding a "neutral" record of 6-5.
Old: 14 for wins + 9 for appearances + 4 for ace appearances + 6 for team wins - 3 for losses - 0 for ace loss = 30 points.
New: 28 for wins + 0 for appearances + 4 for ace appearances + 6 for team wins - 6 for losses - 0 for ace loss = 32 points.
-> The new system scores slightly higher, but not appreciably (7%).
-> Mid-range players scoring much less in the new system in comparison to ace players (18 point difference vs 12 points in the old system)
Place-filler/Sniper ( HoeJJa, Last, s2)
These guys tend to end up with weak records. Stuff like 3-4 with an unexplainable ace appearance for a loss and a game they just didn't get to play. Again, they show up on every team, but since the players aren't that good we'll go with the other side of neutral, 5-6. And this is a "good" player of this type.
Old: 6 for wins + 6 for appearances + 2 for ace appearance + 5 for team wins - 3 for losses - 0 for ace loss = 16 points.
New: 12 for wins + 0 for appearances + 2 for ace appearance + 5 for team wins - 6 for losses - 0 for ace loss = 13 points.
-> The new system scores a decent bit lower (19%)
-> It's not that many points difference per player but these are the guys filling out your roster so it will add up.
Slumper
Finally I want to consider the guy who goes on a wild slump - the one you don't see coming. The round firebathero or  BeSt starts sucking ass, or whoever your personal favorite inconsistent player is. The ones where the guy ends up with a 3-8 record for the round, including going 0-2 in ace games, and his team crashes to a 4-7 record.
Old: 8 for wins + 9 for appearances + 4 for ace appearances + 4 for team wins - 8 for losses - 0 for ace loss = 17 points.
New: 16 for wins + 0 for appearances + 4 for ace appearances + 4 for team wins - 14 for losses - 0 for ace loss =10 points.
-> The new system scores these guys much lower (42% fewer points)
-> The new system scoring drops these guys' value even below the "bad players" who serve as snipers and guys just there to meet race requirements.
--------------------------------------------------------------------------------------------------------------------------------
My major problem with the new system was that the differences between players became much more pronounced, which is only to be expected when you switch from 3/0 scoring to 4/-2, doubling your point spread. Especially this was problematic because it devalued the mid-range players, making them much less valuable even if they performed and - since they tend to be the players most liable to slumps - much more dangerous to your team as well.
Of course, the old system did have some weak points, mostly that the bad player on a good team could end up making a significant difference (it's far more viable to pick up a player like, say, Barracks for team points only under the old system than the new). Lineup points are not a problem for me: the map orders and play orders are so unpredictable, and there's simply no way to include enough possible trades to offset this; lineup points are vitally important to making teams stable.
--------------------------------------------------------------------------------------------------------------------------------
EDIT 5: Explanatory note: My system is based on the idea of correlating the difference between winner's and loser's points for the ace game with those of a normal game, rather than lessening the difference as both "old" and "new" scoring systems did. </end edit>
I'm perfectly happy with the old system's conception, and upping the winning points seems like the easiest solution to its minor problems. My proposed system would just kick up the win value. I also want more distinction in the scoring of ace games, for a total scoring system like this:
Appearance: 1 point
Win: 3 points
Loss: -1 point
Ace appearance: 2 points
Ace win: 4 points
Ace loss: -1 point
Team win: 1 point
I feel like once you get to the ace match, your performance really really matters and this should be reflected more, rather than less, in the difference between the players. In a normal match, the "old" scoring gives a +3 difference to the winner (3,0), and the new gives a +6 (4, -2), but in an ace match this difference is reduced by the old system to +2 (4, 2), and by the new to +4 (6, 2), when the ace game decides the outcome - this seems weird to me.
With my proposed system, your normal game comes in at a +4 difference (4, 0), while ace is stretched to a +5 difference (6, 1). This may make ace players for poor teams less valuable, but then they're not winning, are they?
--------------------------------------------------------------------------------------------------------------------------------
Finally, I want to demonstrate, so I'm going to take my examples above and rescore them with my proposal:
Ace (10-3 (2-1), team 8-3): 42 old, 50 new, 53 mine
Workhorse (7-4 (1-1), team 6-5): 30 old, 32 new, 37 mine
Sniper (3-4 (0-1), 1 no-play, team 5-6): 16 old, 13 new, 20 mine
Slumper (3-8 (0-2), team 4-7): 17 old, 10 new, 18 mine
A couple things to note:
-> My scoring system does inflate point totals for everybody.
-> Notice that losses do eat into a player's point total, only partially offset by ace appearances (compare the sniper and slumper, who now end up about the same "again").
Point differences:
Old: 42 -(-12)-> 30 -(-14)-> 16 -(+1)-> 17
New: 50 -(-18)-> 32 -(-19)-> 13 -(-3)-> 10
Mine: 53 -(-16)-> 37 -(-17)-> 20 -(-2)-> 18
My system splits the difference fairly evenly between the two systems, so real winners are still rewarded without imbalancing the FPL too heavily in favor of ace players and making other strategies not viable, which to me is the main goal of reforming the scoring system.
Sorry I kind of went on for a while here, I like numbers.
EDIT: Also my system is awesome because it's virtually identical to the one Grobyc proposed up above. I do the ace scoring at -1 for the loss, and that's the only difference. So I'm almost just like Grobyc, except nicer. 
EDIT 2: I'd also consider dropping the ace match appearance point to just 1 point. This has the benefit - if you think of it as a benefit - of lowering the relative value of ace games.
EDIT 4: The scoring then would be:
Appearance: 1 point
Win: 3 points
Loss: -1
Ace appearance: 1 point
Ace win: 4 points
Ace loss: 0
Team win: 1 point
Ace win can't be returned to 3 points (normal win) without counting the winner/loser difference as less than that for a normal game, as mentioned above. The only difference from my main proposed system is that this give the winner 1 less point. It differs from Grobyc's scoring in that it gives the winner 1 less point and the loser 1 more (which messes up the win/loss difference correlation; not good).
EDIT 3:
On April 06 2010 12:30 Qatol wrote:Show nested quote +On April 06 2010 08:27 Grobyc wrote:
My hybrid:
Win = 3, Loss = -1, Lineup = 1, Ace Selection = 2, Ace W/L = 4/-2, Team Win = 1 My problem with a setup like this is that it puts so many points into those ace matches. ...better than the players gaining a bunch ... because they lost an ace match, ...should Flash really be scoring 10 points/night...? I don't see why it shouldn't be a pretty small bonus for going to/winning the ace match since that match is almost always icing because that player has already won a game earlier in the match...
Since I'm claiming my system is basically the same as Grobyc's, the same criticism applies, so I thought I'd reply to it. I'm going to refer to our system (despite the slight difference with ace scoring) as M&G, for Musoeun and Grobyc.
Taken straight up, the ace match is just another game
3/0 (old), 4/-2 (new), 4/0 (M&G).
This is one way to do it; if we take the hypothetical you propose, "Flash already won another game". Let's say Flash wins. He is going to get 7 points on the night (old), 9 (new), 9 (M&G). The other game winner on the winning team gets 4 (5, 5) and the game winners on the losing team get 3 (4, 4). So even with "normal scoring" the ace player is going to almost double his own teammate's score regardless of which system, and more than double the opponent's game winners.
But only almost double: and instinct says the ace match is more than just a normal game; common sense says a player who wins his game and then wins the ace match should get at least twice the points of the other game winner on his team: in short, taking that score (4 old, 5 new, 5 M&G) as a benchmark we say the ace game winner should be getting a score of at least 8 (10, 10).
Old ace scoring gave the 2-game winner 8 points (1+2+2+2+1), which does in fact double the score properly.
New ace scoring gave the 2-game winner 11 points (4+2+4+1), which we discovered helped inflate the ace players' importance since they're earning more than double.
For Flash or another hypothetical game-winner, special ace game scoring is adding 1 point in the old system, 2 in the new system, to the total score as against pure standard scoring. This is a maximum of 22 extra points across a round IF every match goes to ace and IF a team sends the same ace every time and IF he has played every time and IF he wins every match - which is a lot of points on the one hand, say about what a 5-3 player would contribute, and if a player goes on a 22-game win-streak then, DAMN, they earned those points.
The M&G system still gives the 2-game winner 11 points (1+3+2+4+1), This looks like it has the same inflation dangers as the "new" system. I don't believe it really matters though, as the real culprit lies in the losing scores, which the M&G system fixes by correctly correlating ace win/loss difference with the normal game win/loss difference (see above).
|
|
|
|
|
|
EPT
