[saCTF.I] Draft Format discussion

[RickJames]

Dunno about the whole lots of teams, no dev matches thing.

Just based on experience I can tell you that the more teams you have in a draft, the harder it is as a mid-salaried captain. The draft is biased towards high and low salaried captains because the low capts get first picks in both of the first 2 rounds and the high capts are already top players who usually snatch up the last top players left at the end of the first round.

If you're like an $800 captain in a draft with 16 teams, assuming you pick 4th or 5th, the best 4-5 players will be gone before your first pick and the next 15 or so will be gone before you can pick again in the 2nd round. It gets worse as the # of capts increases because there just aren't enough good players to go around. And I'm not just talking about the first couple rounds, since the high and low salaried capts will continue to have pick advantage in subsequent rounds. Top capts will be picking $100s after they secure their top 2nd man (while everyone else is picking up their 2nd picks), putting them into the earliest pick positions in the next rounds while other captains are picking $100s. Low capts will be picking $100s after they secure their top 2, at the same time as mid-salaried capts, and this puts the low capts back into the earliest pick positions because they themselves lower the overall salary of their team. There are just not enough good players left over for mid-salaried capts when they're competing with lots of other capts who are on the extreme ends of the salary spectrum.

I'll give an example, from last saCTF draft:

First round (not including captain round): 1489·Serenity 1467·unrealshots 1456·Rum 1445·riven 1434·Pete 1423·FALLACY 1402·blackout\\ 1380·InfamousRaider
The list of top players drops off dramatically after IR (certainly deceived, morpheus, etc were not on the level of these top 8 in sniper).
I'm picking 4th. Imperious picks unrealshots, pun takes serenity, and ulo takes rum.
Pete and FALLACY had poor availability (as proven when they barely showed up for rain and scream).
Riven is pure frag power and cover O (he's mediocre on D and definitely not a flag runner).
IR was not as playing as good as blackout on D.
So my best option was blackout. Not a bad defender, but certainly not able to carry a team in the way that seren, shots, rum, rain, finalz, or even ramdrop can in sniper (rain, finalz, ram were captains).

Second round (after rain snapped up Pete): 1380·InfamousRaider 1369·Deceived 1336·Steel 1325·ravonix 1314·HuMPtY 1314·Chris 1303·Spriggan 1292·snkydigm 1281·flowX 1270·spydee
There were a few OK underrated guys after spydee, but this is pretty much the list of options.
Snky and IR go before I can pick.
I need an attacker, because if I snap up a second defender my options for a first attacker in the next rounds will be bad. This rules out ravonix and spriggan.
Deceived had low odds of showing up (and, indeed, he ended up noshowing the season for finalz)
Chris was overrated, as evidenced by him falling to the next round and performing mediocre overall.
My options, then, were either Steel or flowX. In retrospect, Steel was by far the better pick but I went with flowie because I trusted him more (figured Steel hated me).
Even if I had gone with Steel, though, and rain had gotten flowX instead of him, can you really compare a team of blackout and steel to a team of rain and Pete or unrealshots and IR or Serenity and snky?
Pretty much no matter what I did here, I had 0 chance of matching the highest and lowest capts. And this is with only 8 teams! Can you imagine what this would be like with 16? Even if we had way more top iCTF players suddenly deciding to sign up and try sniper this time around, I don't see it balancing out.

I think even 12 balanced teams is a pretty optimistic/questionable scenario, let alone more.

[ramdrop]

I picked almost last RJ, I loved the team I drafted, I was very happy with it. All the 'good' players had gone by the time I snapped up Riven, I went for players I knew wouldn't let me down and we made the finals.

Blizz Riven SolidSnake & KyuKen <33

[RickJames]

[PACO]

[InfamousRaider]

To be honest, I think it is completely fair as it is. Regardless of low, mid, or high salaried captain. That same draft, imperious picks first and gets unrealshots. Second round he gets me. Strong start to a team and the rest of the team was decent. Still we didn't make the playoffs. I wouldn't say that Imperious was at any more of an advantage.

Also the more captains you add. The more people who will be gone why then higher salaried players are picking. If there are 16 captains, by the time rain picks last the top 28-32 players (roughly) have already been put on a team.

[rev]

The ability to have larger and smaller disparities between salaries is a must. Whether it's through ranking similar players he same via like a tier or implementing paco's solution is fine..needs to happen though.

[kevinsan1ty]

I like the idea of broadening the spectrum of salaries, especially for some of those top range players. Rankings will have to be really well thought out by people who are active and pugging often in the actual mod, and will matter from top to bottom.

I think most people in the rankings threads just do the top 30-40 players and just copy/paste the rest from the first post. Whoever did tiers in the past was awesome, and bravo. If we're tightening up rankings we should have different threads for different tiers, or maybe just a tiers thread and let the admins make the final call on salaries for individuals within those tiers? Think it's a lot easier to lump 10-20 similar players in to a tier than individually pick who is better across the board (as a community).

[rev]

[PACO]

[ulo]

Firstly I am sorry rain for "hijacking" the thread or rather replying to a matter that's not 100% based on captain signups but since it's been brought up and since PACO mentioned my interest for the topic I can't really resist to give my explanation on the matter.

Okay, RickJames have brought up the fact that mid salary captains does get the bad end in the current draft format and it get's worse with more captains. I will bring up 3 different scenarios to explain this. First out we will have a simple example with no player names named to explain the basics of it. Then I will give a example from our first iCTF on OwnedWell to further explain what happens when the draft goes live. Lastly I will give a more "fun" example based on soccer (most likely the easiest sport to explain from since most people have some knowledge about it).


So here we go with example one:

Let's say we have 3 captains A, B and C. A is a low salary captain, B is a mid salary captain and C is a top salary captain. A is either the lowest or below the lowest "normal" development player on his team, B is a okay high development player and C would be top salary on any team and is a really superstar player.

First thing to happen in the draft is that the captains will have to pay for there own salary. Then A get his first pick, he will naturally go for one of the top players in the league. Then B picks and picks a player left by A. Now lastly C take his pick and can pick a player left by A and B.

Now we come to the interesting thing. All teams have 2 players, however the important players on each team is that A has one really good player for his starters. B has one okay player for his starters. C has himself a really superb player and he also have a second strong player. So C has two good player while A and B is really lacking still in comparison. A get compensated and can stay close since he picks his second player next and is only slightly after C in the picking order of getting starters. B however is left trailing since he neither got to pick first nor even get's compensated for having a lower first starter then both A who picked first and C who has himself and can pick his second player before either of them.

This process keeps going on as the draft goes on. Low and high salary captains keep being ahead and always get the better deal and can "stack" their starters more. (Which has proven to be a winning concept). Now you can imagine how this get's even worse for B if more low salary captain signup since then he will pick his first player even later in comparison. Same goes for more good starter captains which have a higher salary, since they can pick away even more good players before B can pick his second player.



Okay so on to a more hands on example with actual players and captains. I will make a example from the inaugural iCTF league here.

I will compare the highest, lowest and a mid salary captain. There were 16 captains where rain was highest salary so I will pick him as the high salary example, lowest was Pun and he will serve as low salary captain example. 16 divided by two is 8 so I pick the 8th captain which happens to be a perfect mid salary captain in RickJames as example for a mid salary captain.

So the captains buy themselves and in the first round Pun goes for unrealshots which is worth 1500.
A few more captains pick and it comes to RickJames who picks up snap which is worth 1424.
All the other captains pick and it comes to rain who picks up paleface (YES I AM WELL AWARE HE PICKED KEA, however since that is more a tactic from how the draft is worked his "real" first pick of the draft is paleface, which I assume everyone agrees on) which is worth 1356. However rain also has himself who is worth 1483.

Now the question is why on god green earth did rain get to pick up his second starter before RickJames did when clearly he had a starter in himself already which was worth more then what RickJames first starter snap was worth? Secondly Pun also picks his second starter which is Risus who is worth 1179 before RickJames can make his pick and goes for Cyrax who is worth 1330. Now RickJames in the end did get a starter worth more then Pun but this was simply cause Pun picked a lower salary play than was available to save salary for coming rounds, which is needed to be done with the round caps. Rain however still has a first starter and second starter worth more then RickJames was able to pick up, simply cause he was a high salary captain.

To me this is not a good way of balancing teams since the teams who have the "best" players also get a better deal on how to plan for coming picks and can get a "better" (salary wise) player then a captain who is in the mid of the picking order.




Last example is a short soccer example.
(I couldn't be arsed to find player names and good examples for all players, specially since it could end up in a discussion about those players).

Cristiano Ronaldo is captain for one team, the other captains are one a really bad player playing for some division 3 team in England and the last captain is a decent player in a mid team in Premier League.

So the really bad player picks first and get Leo Messi, then a bunch of other players pick and the decent mid player get to pick and "best" player left for him is Wayne Rooney. Now two of the captains have picked and it's obvious to everyone that the team with Wayne Rooney should get to pick their second player next to start for his team. But instead now Ronaldo get to pick his second player and after that the team with Messi pick, then Rooney has to wait for a bunch of other players to be picked before he get a player to accompany him.

Doesn't sound so smart now does it?



So how could we solve this then, also within the limits of our currently used draft bot?

First thought is how can we make it so that picking starters is what's count?
It's actually rather simple, two things has to be done. First is to get rid of the round caps and simply give the captains everything to work with straight up. Secondly captains should not pick themselves in the first round before they start picking instead they should be forced to pick themselves at a predesignated round. This round is the mathematical round in which they would be among the amount of players being picked from top salary. I will explain it with a few fast examples:

If we have 16 teams and a player is ranked as 1-16 he would have to pick himself in round 1 since he is among the players in the first round who get's picked if all captains pick highest available player. If instead the player was ranked as number 40 he would have to pick himself in round 3 as 3*16=48 which is higher then his ranking number and he should mathematically go in that round. So the round is based on the players ranking in the salary and the amount of teams.
(There could be a exception: If a captain is so low salary that he would not mathematically even get picked he of course has to pick himself in the last round no matter what).

Since with this in mind and working on the example above from iCTF rain would pick himself in the first round while Pun and RickJames was free to pick their first starters. Say they pick the same players as above. Now in the second round the only player on each team is the one picked in the first round. So among those captains RickJames now can make his second pick before rain and Pun, which is only logical since he has the lowest ranked starter on the team. Then rain get to pick since he is ranked lower then unrealshots and lastly Pun picks.

I have done test drafts with those premises in mind and the difference between the top 5 starters (with picking highest available by salary player for each captain and captains picking themselves in the right round) ended up being only 122$. This margin is only there more or less since it's and odd number of players, if I did the same thing with sDOM which has 4 starter the difference would be less then 20$ and there it would simply be cause the picking order get's worked around captains picking themselves even if they was no the highest available player.


So why have we not done this?
Firstly it's because to do this captains need to be pick able in the draft bot and any captain could for example pick rain before he even pick himself if they wanted to screw around with the draft. This could be a problem and make for really long drafts if it happens mid draft where it makes more sense to try and reverse it rather then simply restart as would be easiest if it happens in the first round. Secondly it also requires a dual copy of all captains and them being set to captains in the order they are going to pick since the draft bot does a captains round no matter what. Which could be slightly annoying if a captain changes nick on IRC for example. Thirdly it's mainly because people are used to the way the drafts have been done and trying to convince captains that this way is better is not always super easy.


Personally I see no reason to continue with the "old/current" system since it's biased and only works well for low and high salary captains. For example I could not have picked the team I got on in the sDOM league, since I would not have been in the picking order required to get those players.


*Curiosities
All leagues run on globalunreal had the most stacked top starters teams ending up top 4 at a rate of 50% (2 out of 4) or 75% (3 out of 4 ) except saCTF (sniper) which had only 25% ( 1 out of 4). Which made it fairly simply to say which teams would be top 4 at a high ratio of success.
I am the only mid salary captain (mid as not being starter salary wise on his team) to end up top 3 on a globalunreal league. So what I did with those teams was fairly uniqe thing. Can add that sDOM one I was given a salary bump on my starters by over 100$ which in 100-1000 is fairly much <3 snky . The sniper draft was most likely the best one I will ever have and rum was most likely the best player in the entire league, he simply destroyed the enemy defense each and every map.
I am also the only mid salary captain to even come close to having the most stacked starter team, with a second, a third and a fourth place (sDOM I and II along with saCTF).



tl;dr
The current draft system is bad for mid salary captains, if you want to know why read the post!

[InfamousRaider]

Solution:
Don't give captains $1500 for the first round after the captain round. That way the top salary captain can't afford a second starter before anyone else.

Lets say you have a league with 10 players per team and a total salary cap of $8,700 (that was predetermined based on the average player salary).

The round caps would be:
$1500 - captain round
$800 - round 1
$800 - round 2
$800 - round 3
$800 - round 4
$800 - round 5
$800 - round 6
$800 - round 7
$800 - round 8
$800 - round 9
$8700 - Total

This is overly simplified for demonstrations sake but you get the idea. The top players have to wait longer to get their second starter. I don't necessarily think this is more fair but it would get rid of what ulo is talking about.

[ulo]

[Sauron]

[InfamousRaider]

[HULKSMASH]

Ban them for poor mathematical skills

[ulo]

[RickJames]

I like PACO's suggestion about making salaries more reflective of gaps in player ability. It seems to me that could be enough of a change on it's own, and I'd most like to test that. ulo's and other suggestions seem good, too, but it seems like they'd be a lot of work to implement and I'm not so sure about how they'd work in practice.

PS I'd captain, but I can't make a May 31st draft date.

[Monk]

I split this thread as it's deserving of its own topic.

ulo's suggestion is clearly the best. A captain of $1000 should have to pick himself by - say - the 6th round. The $1500 should clearly have to pick himself in the first round, and a $100 captain should be allowed to wait until the last round (or - arguably, any captain <= $x salary where x is the round cap for the final round). It's true as well, that the players that can carry teams should have a bigger salary difference to the other top players.

Code wise however, getting the draft bot to deal with this is quite a challenge. It certainly couldn't be left to 'letting captains not be stupid'.

[d3R]