Auction balancing

[ace rooster]

The balancing to wizard thread gave me an idea for a way to get an quantified estimate of where various classes balance against each other that actually makes people put their money where their mouth is. The idea is that we create a party* where only the level of one class is fixed (the class we are balancing against). We go through the other classes in the party, and distribute the roles by a reverse auction. We start by offering the class at a very high level, and then continually lower the level until only one person would play that character at that level (randomise in the case of a tie?). This person then gets to play that class at that level (fixed WBL to the balance point's level). We procede through the classes until there is one person left, who plays the balance point class and level.

It is probably more understandable if I give an example. Alice and Bob will be playing a classic fighter wizard duo, and decide to balance around wizard 10. DM Doug offers the fighter at level 20, and both Alice and Bob think the fighter is better at this point, so they both put their hands up. Because there is more than one hand up, Doug offers fighter at level 19. Both still think fighter is better, so they keep their hands up. Doug eventually gets down to 14 and Alice thinks that she would rather play a wizard 10 than a fighter 14, so she puts her hand down. We are left with one hand up, so Bob gets to play a level 14 fighter. Because we only have two people we are done, and Alice is left with the wizard 10.

Comments on the idea are welcome, but I actually mean to use this thread to find out what people would actually do in this case, so if you could also include your answer it would be appreciated. If someone could express more clearly what I am doing that would be good too. :smallredface:

If the party consisted as follows, at what point would you put your hand down for each of the auctions?

The Party:
A fighter nobleman. Gets double WBL
A druid.
A brass dragon. This one is done by HD, and has no class levels..
A wizard 10. This is the balance build, that the last player without a character gets.

My answer:

Fighter: I would reject at level 17, maybe higher.
Druid: I would reject at level 8. I think I would do better with a level 9 druid than a level 10 wizard, but mostly because I know druid slightly better.
Dragon: I would want to go below 13, but that is more because I really want to play a dragon than because it is better than wizard 10 at that point.

Note that these are mostly pulled out my backside, and are probably way off giving a balanced party. Somebody with more experience of sober play would probably have a far better idea.



* Hypothetically for the purpose of this thread, but if someone wants to actually run it I would be very interested to hear how it went.

[Flickerdart]

This entire premise forgets that multiclassing exists, which is kind of a raw deal for everyone involved.

[ComaVision]

What about if the campaign starts at level 1? Do you tack on NLA until everyone is happy? What if NLA don't make it balance?

[Khedrac]

but mostly because I know druid slightly better.
And this is the potential problem with this idea - a lot depends on the other players and how well they optimise or play certain classes.

In the groups I play in, I tend to be towards the upper end of optimisation present, but not the top - the best optimiser tends to play mundanes (which usually do insane damage).
That said, our games are very low op by the Board standards. I usually play a caster, but I like to go for support casters or rp choices rather than the best possible. For one thing, support casters don't need to break to re-learn quite so frequently.
I also avoid melees because one of the main DMs likes fumble tables...
Also we have more players up for mundane than caster, hence usually caster.

So - what level to play the various characters are really depend on who is playing which other character. A poorly player Wiz 10 won't overpower a group of 8s, even though a well played Wiz 10 will hold their own in a higher level party.

Also, for mundanes even more than casters, the available items makes a huge difference - so do the characters all get the same cash, or does each get the WBL of they end level?

[ace rooster]

And this is the potential problem with this idea - a lot depends on the other players and how well they optimise or play certain classes.

In the groups I play in, I tend to be towards the upper end of optimisation present, but not the top - the best optimiser tends to play mundanes (which usually do insane damage).
That said, our games are very low op by the Board standards. I usually play a caster, but I like to go for support casters or rp choices rather than the best possible. For one thing, support casters don't need to break to re-learn quite so frequently.
I also avoid melees because one of the main DMs likes fumble tables...
Also we have more players up for mundane than caster, hence usually caster.

So - what level to play the various characters are really depend on who is playing which other character. A poorly player Wiz 10 won't overpower a group of 8s, even though a well played Wiz 10 will hold their own in a higher level party.

Also, for mundanes even more than casters, the available items makes a huge difference - so do the characters all get the same cash, or does each get the WBL of they end level?

The idea is not to create an objective assessment, but rather to establish where people put the point that two classes have the same value to them. It is entirely subjective. Would you rather play a great wyrm brass dragon or a wizard 10? The answer will almost always be the dragon. If you ask whether someone would rather play a wyrmling brass dragon or a wizard 10 that answer will almost always be the wizard. For any individual there is a level of dragon where this change occurs, and the idea is to place yours (with no 'right' answer).

It is not a system that accomadates progression, so it would only really work for a one off above level 4 ish. It is not actually intended to be used at all, not least because you need predefined roles.

Wealth is fixed at the level you are balancing around in the example I gave, but it doesn't have to be. Equally one class is used in this example, but it works fine for comparing any builds (including multiclassing) by fixing a level on one and then identifying the level where you would switch from using one to the other.

[ZamielVanWeber]

If the party consisted as follows, at what point would you put your hand down for each of the auctions?

The Party:
A fighter nobleman. Gets double WBL
A druid.
A brass dragon. This one is done by HD, and has no class levels..
A wizard 10. This is the balance build, that the last player without a character gets.

Amongst these 4? Double WBL for a fighter is still going to be under a normal WBL wizard of the same level for most of his existence.
Druid is always a good choice.
Brass dragon with no class levels? Then why bother. The LA is going to hurt and as I gain racial HD the LA goes up too.
Wizard is also always a good choice.

[dascarletm]

It's interesting, but I see a couple problems:
It's subjective. Let's say I'm biased against class X. It will go for a higher level than it probably should. Also, you run into the problem of people trying to outbid, etc.


Auction's irl can see items sold for way more or way less than their actual worth. It seems too inaccurate as a thought experiment for me.

[Quertus]

If the party consisted as follows, at what point would you put your hand down for each of the auctions?

The Party:
A fighter nobleman. Gets double WBL
A druid.
A brass dragon. This one is done by HD, and has no class levels..
A wizard 10. This is the balance build, that the last player without a character gets.

Love the idea, sad that there aren't more responses.

Me, I don't care about game balance. So you can have fighter 20, or fighter 50 and druid 50 for all I care. I never raise my hand for them. I'll take wizard 10, and build it very sub-optimally, thank you very much. Or maybe I might try out the Dragon, but only if I can get it at overpowering levels, and I like the rules for how it / the party levels afterwards.

Now, if it's only about power, and I can do everything I want, and create a cheese fondue, I might still take wizard 10. or I could take fighter 10, and solo an army of tarrasques.

[Jay R]

The assumption of the auction is that we all prefer a balanced party to a powerful one, and will sacrifice a lot of power for that purpose. That's not always the case.

In my group, I suspect that it goes like this.

DM: Who wants the fighting nobleman at level 20?
(Several hands go up. We look at each other, decide Kevin is the best choice, and all hands except Kevin's go down.)
DM: OK, Kevin has a level 20 Fighter. Who would accept the Druid at level 20? (We look around and decide it's Mary. She reaises her hand; the rest of us don't.) Um, right. Mary has a level 20 Druid.
etc.

I would cheerfully play a level 10 wizard with super-powerful allies, in adventures balanced to the level-10 wizard.

[ZamielVanWeber]

Rereading the problem: What happens if there is a role two people want and another no one wants? If we have two players who love melee none of the four want to play a wizard what happens?
In your example, we have 4 classes and 4 players:
Only player A wants class 1, so they take it at face value.
Only player B wants class 2, so they take it at face value.
Players C and D really don't want class 4, so they shred class 3 vying for it.

End result is the party still isn't balanced (2 high levels and 2 low levels) and one to two of the players are going to be unhappy because one had the class he wanted to play give up a lot of ground and the one who is stuck doing something they don't want to do.

This won't end as badly is the party divides the roles before starting, but that assumes a higher degree of party cohesion and still risks someone being stuck doing something they don't want to do.

[BowStreetRunner]

Now you have me thinking along the lines of FFRPG - Fantasy Fantasy Roleplaying Games. A group of DMs each agree to run the identical campaign at the same time and report the statistical results of each session on a central web page. Each FRPG session has the same number of players in the party, and each must fill a certain role, but the specific builds are up to the player. The builds are published on the central web page. The FFRPG players then each draft their own party from the available FRPG characters, although they can only draft a single character for each role. Their FFRPG parties they gain points based on the statistical performance of the various characters in the FRPG sessions.

:smallamused:

[Troacctid]

I think bidding based on starting wealth makes a lot more sense than bidding based on starting level.

[OldTrees1]

If the party consisted as follows, at what point would you put your hand down for each of the auctions?

The Party:
A fighter nobleman. Gets double WBL
A druid.
A brass dragon. This one is done by HD, and has no class levels..
A wizard 10. This is the balance build, that the last player without a character gets.


I foresee a problem and my answer might demonstrate that problem:
Fighter: 10th
Druid: 8th
Brass Dragon: 16HD

The problem is that people do not judge what they want to play merely via balance concerns. Thus when aiming for the kind of character I might want, I end up with 3 wildly imbalanced results(if voting against my clones).

[Jay R]

Rereading the problem: What happens if there is a role two people want and another no one wants? If we have two players who love melee none of the four want to play a wizard what happens?
In your example, we have 4 classes and 4 players:
Only player A wants class 1, so they take it at face value.
Only player B wants class 2, so they take it at face value.
Players C and D really don't want class 4, so they shred class 3 vying for it.

End result is the party still isn't balanced (2 high levels and 2 low levels) and one to two of the players are going to be unhappy because one had the class he wanted to play give up a lot of ground and the one who is stuck doing something they don't want to do.

This won't end as badly is the party divides the roles before starting, but that assumes a higher degree of party cohesion and still risks someone being stuck doing something they don't want to do.

This implies that nobody thinks ahead.

First of all, the last class is the balancing class (in this case, 10th level wizard), and the level is pre-set. Presumably, anybody who let player A have class 1 at level 20 would rather player a Wizard 10 than a Class A 19. So A and B got their choices only because C and D were both willing to play a Wizard 10.

When we look at the classes, before the bidding starts, I want to know what everybody else wants, and maybe make a deal. We both want to be the dragon? OK, I'll let you have it, in return for my Wizard 10 getting first pick of the treasure for a few games.

The theory of the auction is that nobody gets a class and level unl;ess everybody still in would prefer to have the level 10. So whoever is last gets the baseline character that he was willing to play.

[Note that the assumption is that the baseline character is one everyone finds worth playing. If not, you're balancing to the wrong one.]

[Nohwl]

i'd rather have wizard 10 over fighter 20, even with double wealth by level. does the dm increase the level until i'd raise my hand? how does this work with prcs?

[ryu]

I would also take wizard 10 over fighter twenty. Of the options offered I might take druid if it doesn't take too much of a beating in auction. It would have to be a good bit higher than ten to upset the undeniable fact that of the tier ones wizard is my jam.

[ZamielVanWeber]

This implies that nobody thinks ahead.
Who guarantees they do? I have seen people plan together as well as not and am pointing out the system begins to fall apart if they don't.


First of all, the last class is the balancing class (in this case, 10th level wizard), and the level is pre-set. Presumably, anybody who let player A have class 1 at level 20 would rather player a Wizard 10 than a Class A 19. So A and B got their choices only because C and D were both willing to play a Wizard 10.
Or they don't want to play Class 1 or Class 2. I mean, what happens if the only choice both of them like is Class C?


When we look at the classes, before the bidding starts, I want to know what everybody else wants, and maybe make a deal. We both want to be the dragon? OK, I'll let you have it, in return for my Wizard 10 getting first pick of the treasure for a few games.
And if they don't agree to that or if you don't want to play the wizard? The system risks mandating that someone gets the short end of the stick and rewards adversarial play at the start if the party is not immediately cooperative.


The theory of the auction is that nobody gets a class and level unl;ess everybody still in would prefer to have the level 10. So whoever is last gets the baseline character that he was willing to play.
And if none of them are willing?All these hypotheticals require people to be working fairly well together. If there exists one option that no one wants then someone has to lose or the entire process needs to be repeated, which just seems inefficient all things considered.


[Note that the assumption is that the baseline character is one everyone finds worth playing. If not, you're balancing to the wrong one.]
The assumption is that all four are ones that people would want to play and needs everyone to negotiate ahead of time who gets what. At which point why aren't we just gathering the players ahead of time for a character creation session and hashing the whole thing out there so people aren't pigeon holed into something?

[yellowrocket]

What we are looking at is not an actual auction. This was simply a thought experiment on how people value a given class in relation to a wizard 10. Some of you have said you'll never take the fighter 20. Which is fine. You don't have to. That wasn't the point of the post. It was about finding where the balance lies in selecting a class at a variable level vs taking the wizard at level 10.

For me the fighter at 20 with double wbl has access to things that make him competitive with a level 10 wizard.

Druid with ac I take maybe a level or two higher than the wizard only because I'm more familiar w in t tell wizard spell list.

Dragon I think I skip only because I don't know enough to place a relative value on them at this point.

[ryu]

What we are looking at is not an actual auction. This was simply a thought experiment on how people value a given class in relation to a wizard 10. Some of you have said you'll never take the fighter 20. Which is fine. You don't have to. That wasn't the point of the post. It was about finding where the balance lies in selecting a class at a variable level vs taking the wizard at level 10.

For me the fighter at 20 with double wbl has access to things that make him competitive with a level 10 wizard.

Druid with ac I take maybe a level or two higher than the wizard only because I'm more familiar w in t tell wizard spell list.

Dragon I think I skip only because I don't know enough to place a relative value on them at this point.

The reason I devalue fighter so hard is that they don't really get much as they level. Mostly either relatively basic tricks almost all requiring melee range, or bigger damage per round with weapon attacks of some feat based technique or another. I consider them essentially equivalent to wizard six in terms of utility and I feel I'm being generous with that comparison. Also keep in mind that double wealth from level 10 WBL is not equivalent to level 20 WBL by any stretch, and the wizard is more than, far more than, twice as efficient for using their wealth on things I'd consider worth getting. Now lets also remember that wizard power is exponential and ask ourselves what the difference between a competently built wiz 6 and wiz 10 is.

[NichG]

In the spirit of the original premise, here are my price points, decided before reading what ace's are:


Fighter: 21
Druid: 12
Dragon: 21
Wizard: 10

For dragon, its 21 because I think its worth holding out for epic and hoping that someone ends up with either the fighter or the dragon and I get the other.

[ace rooster]

I foresee a problem and my answer might demonstrate that problem:
Fighter: 10th
Druid: 8th
Brass Dragon: 16HD

The problem is that people do not judge what they want to play merely via balance concerns. Thus when aiming for the kind of character I might want, I end up with 3 wildly imbalanced results(if voting against my clones).

Why is this a problem? The system should balance based on how much people want to play various builds, which will be based on many factors not limited to effectiveness. If a build is simply fun to play then the level it may end up playing at a level below what would be 'balanced', but if the player is having fun anyway I don't see this as a problem. The player has already stated in their bidding that they value fun higher (relative to effectiveness) than the other players at the table, so they would be the most appropriate person to play that build.

If you find yourself in a group where everyone desperately wants to play a fighter no matter how ineffective they will be then I don't see a problem with letting someone play an ineffective fighter.

This does highlight another reason to limit this to one off games. There is a lot of scope for making an error in valuation that can mean you are very ineffective due to overestimating the potential of something. In some ways this danger is lowered though, because it does require at least two people to make the same mistake.

As for auctioning by wealth, there is no reason that you couldn't, even in the same party. That particular auction just goes by decending wealth rather than level. It is another experiment for another thread though. :smallsmile:

People gaming the system is always possible (with any system), but for the purposes of this thread lets assume that we are all mature enough to accept that the system is their to help, and play nice (which ironically means not playing nice, ie, no cooperation).

[Jay R]

Who guarantees they do? I have seen people plan together as well as not and am pointing out the system begins to fall apart if they don't.

Oh, agreed. But that has nothing to do with the auction. If the players refuse to plan together the party falls apart under any system.


Or they don't want to play Class 1 or Class 2. I mean, what happens if the only choice both of them like is Class C?

This auction isn't intended for people who will only play a single class. People like that should obviously play the only class they like.


And if they don't agree to that or if you don't want to play the wizard?

I repeat: "Note that the assumption is that the baseline character is one everyone finds worth playing. If not, you're balancing to the wrong one."


The system risks mandating that someone gets the short end of the stick and rewards adversarial play at the start if the party is not immediately cooperative.

The point of an auction is not adversarial. You bid purely for your own goals, not against other people's goals. If I get the Fighter at level 20, that's because everybody else would prefer the wizard to Fighter 19.

In any honest auction, each item goes to the person who will pay the most for it. That is being offered as a rough way to measure people's desires to play a given class and level.

Besides, what "adversarial play" is rewarded? Correct, self-focused, non-adversary-oriented plays are to raise your hand for a class and level you're willing to play, or to lower it for one you aren't. Any other play is either giving away a character you want, or claiming a character you don't want.


And if none of them are willing?All these hypotheticals require people to be working fairly well together. If there exists one option that no one wants then someone has to lose or the entire process needs to be repeated, which just seems inefficient all things considered.

Agreed. But if the DM is forcing a character class that no player is willing to play, then the issue is not the auction; it's the DM.


The assumption is that all four are ones that people would want to play and needs everyone to negotiate ahead of time who gets what.

Not quite. The assumptions are that all four are classes that some player would like to play, if he or she can get it at a high enough level to balance it in his or her own mind, and that there is enough looseness in the players' desires that they can each find one or two acceptable options. Any player who will only play one class has no business in this kind of auction, for the same reason that there is no point price-comparing restaurants if you will only eat at the Olive Garden.


At which point why aren't we just gathering the players ahead of time for a character creation session and hashing the whole thing out there so people aren't pigeon holed into something?

This auction is "gathering the players ahead of time for a character creation session and hashing the whole thing out there." The auction is simply a proposed tool for hashing out character balancing, so each player winds up with something he or she values more than the other players.

But yes, I agree that if players refuse to plan together, and use adversarial play rather than bidding for themselves, and the DM forces classes that nobody wants to play, then the system will break down.

Of course, if players refuse to plan together, and use adversarial play rather than playing for themselves, and the DM forces players to take classes that they aren't willing to play, then any D&D system will break down.

[OldTrees1]

Why is this a problem? The system should balance based on how much people want to play various builds, which will be based on many factors not limited to effectiveness. If a build is simply fun to play then the level it may end up playing at a level below what would be 'balanced', but if the player is having fun anyway I don't see this as a problem. The player has already stated in their bidding that they value fun higher (relative to effectiveness) than the other players at the table, so they would be the most appropriate person to play that build.

If you find yourself in a group where everyone desperately wants to play a fighter no matter how ineffective they will be then I don't see a problem with letting someone play an ineffective fighter.

This does highlight another reason to limit this to one off games. There is a lot of scope for making an error in valuation that can mean you are very ineffective due to overestimating the potential of something. In some ways this danger is lowered though, because it does require at least two people to make the same mistake.

As for auctioning by wealth, there is no reason that you couldn't, even in the same party. That particular auction just goes by decending wealth rather than level. It is another experiment for another thread though. :smallsmile:

People gaming the system is always possible (with any system), but for the purposes of this thread lets assume that we are all mature enough to accept that the system is their to help, and play nice (which ironically means not playing nice, ie, no cooperation).

1) As a DM, I recognize the problem designing challenges that satisfy the player of the strongest character without slaying the weakest character or challenges that satisfy the player of the weakest character without boring the player of the strongest character.

2) As we have been reminded, this relies on that last option being something everyone is at least willing to play. Sometimes such a character will not exist(I can think of 2 players right now that do not have such an overlap). Without such a universally playable character, you run into even greater disparities than my own answers (such as Fighter 4 instead of Wizard 10).

However even when the auction is played fairly (no cooperation and no spite bids) it can (on occasion) result in playable groups.*

*Remember, I am judging this based on the baseline of normal DM adjudicated D&D where each player get's their 1st viable choice with respect to non power incentives and the DM adjudicates balancing the power levels.


Another problem that you have not been fairly addressing:
You say you are only assuming that
1) The last class everyone is willing to play
2) Every other class has at least 1 person willing to play
Lets consider the group of casterplayer, martialplayer, sneakplayer, martialplayer. For this group Bard would be a class all are willing to play despite it not being anywhere near their first choices. So if we populate the list as:

Wizard
Sorceror
Warmage
Bard 10
Well, we get Wizard 20, Sorceror 20, Warmage 20, Bard 10 but 2 players are stuck with characters they don't want.

Instead if we populate it with
Wizard
Fighter
Swordsage
Bard 10
Then we get Wizard 20(remember only 1 player wants wizards at all)m Fighter 8, Swordsage 9, and Bard 10. Now while everyone has a character they want(improved from the first list), we still have dramatic imbalance resulting from 1 class having only 1 fan while the other 3 players had to compete for the other classes.

To avoid these undesirable character, there must be at least 1 configuration of assignments where everyone gets a class they want(this is in addition to the rule that there must be at least 1 class that all players desire).
To avoid these uncontested markets, each subgroup of classes(excluding the last class) must have interested parties that exceed its size.

[Jay R]

We are discussing a system to solve balance problems. If all but one of your group would rather play a Bard 10 than a Wizard 19, balance is not a major issue here. So don't use the balancing method at all. That's like using a sword to build a house and a Periapt of Wisdom as a weapon.


Another problem that you have not been fairly addressing:
You say you are only assuming that
1) The last class everyone is willing to play
2) Every other class has at least 1 person willing to play
Lets consider the group of casterplayer, martialplayer, sneakplayer, martialplayer. For this group Bard would be a class all are willing to play despite it not being anywhere near their first choices. So if we populate the list as:

Wizard
Sorceror
Warmage
Bard 10
Well, we get Wizard 20, Sorceror 20, Warmage 20, Bard 10 but 2 players are stuck with characters they don't want.

Agreed. If people have no interest in balancing, then giving them a balancing method will fail. Specifically, you have assumed that three people would prefer to play a Bard 10 than a Wizard 19, or a Sorceror 19, or a Warmage 19. These people simply do not have balanced characters as a goal, and therefore this method of balancing characters is neither necessary nor valuable.

By the way, how did anybody get the second caster class, since nobody was willing to bid on it, even at level 20?

Also, if the DM forces this group of classes on this group, then the problem is the DM, not the classes.


Instead if we populate it with
Wizard
Fighter
Swordsage
Bard 10
Then we get Wizard 20(remember only 1 player wants wizards at all)m Fighter 8, Swordsage 9, and Bard 10. Now while everyone has a character they want(improved from the first list), we still have dramatic imbalance resulting from 1 class having only 1 fan while the other 3 players had to compete for the other classes.

Yes, of course. The imbalance comes from the fact that these people aren't aiming for balance. A system designed to balance character classes according to individual opinions about balance is useless for people with no individual interests in balance.

I agree that a balancing mechanism for people with no interest in balance is as useless as a chocolate cake for people who won't eat chocolate, or a hoe for people who aren't gardening.

I admit that I have trouble imagining such people. But if they exist, then no balancing method is needed. Let them play the classes they want, all at the same level, since that's important to them, and balance is not.

[OldTrees1]

We are discussing a system to solve balance problems. If all but one of your group would rather play a Bard 10 than a Wizard 19, balance is not a major issue here. So don't use the balancing method at all. That's like using a sword to build a house and a Periapt of Wisdom as a weapon.

Wait. I am confused. What if I have been describing players that want a balanced game(ex: their non caster/caster can contribute) without having to play something they dislike(ex: a caster/non caster). Motivations are rarely so simple as "I only care about power and thus will play the most powerful character allowed regardless of what it is". More often they are more complicated like "I want to play an ___-like character that is able to contribute without overshadowing their teammates and is able to have agency in the world at ___ magnitude of power". 67% of that more complicated motive deals with balance so a balancing method is warranted. Yet
when I went with the simplified "I want to play the most powerful character I would enjoy", your auction system's limited premises ran into problems of imbalance.

Would you please explain how these are not caring about balance?

[Jay R]

Comments on the idea are welcome, but I actually mean to use this thread to find out what people would actually do in this case, so if you could also include your answer it would be appreciated.

Ah - right. It's a specific thought experiment asking for our responses to this specific auction, not proposed for actual use in a game. I've been approaching it wrong.


If the party consisted as follows, at what point would you put your hand down for each of the auctions?

The Party:
A fighter nobleman. Gets double WBL
A druid.
A brass dragon. This one is done by HD, and has no class levels..
A wizard 10. This is the balance build, that the last player without a character gets.

It depends on who else is playing, so let me use two different groups I've played with.

Assuming Mike, Kevin, and Mary.
The fighter will almost undoubtedly go to Kevin. He actively wants it, Mike just wants a powerful character, and Mary wants a powerful caster. So I don't see any point in bidding on this at all. Mike would probably bid it down to about 15.

Either Mary or Mike will want the druid more than I will, so they will probably vote it down to 11 or 12. I'd be hoping that Mike gets it, because then I'd get a very high level dragon, since Mary wouldn't bid on it at all.

The Dragon looks fun. If Mike doesn't really show an interest, I'd aim for this. I'd be willing to take it at about 13-14, but that wouldn't happen. At 16, I'd ask Mike how much he wants to play a dragon. If he really likes it, I'd stop bidding and let him have it at 16.

Assuming a game with April, Wil, and Dirk.
Dirk will simply want power - neither more nor less. I'd expect him to go wizard 10, druid 10, dragon 13, or (maybe) Fighter 20. If he went for the Fighter, I assume that he already has a plan for spending that money effectively.

April wants an interesting character. I'd ask her what she feels like playing, and try to avoid her favorite.
Will would want the wizard 10, druid 10, or dragon 12 or so, and probably would ignore the Fighter.

So I'd take the Fighter 20 if Dirk and April were angling elsewhere. I'd be happy with any of wizard 10, druid 10, or dragon 13, but I'd be angling for the one the others seem to be undervaluing. By paying attention to them, I'd hope to get a couple extra levels on my choice.

Note that I am playing their interests as well as my own, to get them what they want and to increase the level of what I get.