[UNOFFICIAL] Balanced Team Generator

[Crater Creator]

I’d like to present everyone with a faster, easier way of picking balanced teams.

https://docs.google.com/spreadsheets/d/16_tMQkTEUG-hnxx6ArtYVTzhbnfZjFWd4hYFMeHLH7E/edit?usp=sharing

This is a Google Docs spreadsheet anyone can use. Just enter the names of 8 people that want to play, and it selects 4 players to team together to make balanced teams. Copy/paste the result to the full lobby, and then team up accordingly.

We’ve been trying this over the last week or so. In my humble opinion, it’s worked faster and easier than random teams with adjustments, which is what we’ve been doing. It also has advantages over team captains. Now, anticipating people’s questions...

How does it pick the team?

The math isn’t complicated. The spreadsheet simply considers all possible teams and picks the two that are most balanced, based on ratings assigned to individual players.

Where do these ratings come from?

The ratings were originally taken from an earlier spreadsheet made by a long-time member of Larry’s group. When I asked him about being credited, he preferred not to be named.

How are the ratings determined?

According to the writer of the original spreadsheet, they are “just my subjective rankings [last updated around the start of 2019] based on experience, cooperation/attitude, leadership, infected scores, win ratio, tank rock throw score, & a couple other things.”

Since receiving the original spreadsheet, I have moved the ratings onto a new scale, added new players that weren’t in the original list, and made manual adjustments to ratings I thought were no longer accurate.

Why should we base teams on someone’s subjective ratings of the players?

Yes, the ratings are subjective. But the way I see it, this is just a way of automating the crux of what team captains do anyway. When you are team captain, you’re expected to pick players from the pool, based on how each would affect your chances of winning, to come up with a competitive team. If players aren’t weighed against each other, you can only get balance by dumb luck.

Hey, my rating is low! I don’t appreciate being judged and/or humiliated like this!

In order to build people’s trust that this spreadsheet can make balanced teams, I think it’s important to be transparent on how it works. That’s why the ratings are visible. I’m not putting this out there to fuel bragging, trash talk, or arguments. If you’re tempted to react to the ratings that way, just don’t. Please remember that half of any group is going to be above average and the other half is going to be below average. It’s not personal, it’s just math. I know I’m certainly not the best player among us, and that’s okay.

I consider the spreadsheet a ‘living document’, and I am implementing a process to incorporate feedback and make the ratings more accurate over time.

Okay, well I think the numbers are off. How can we make the ratings more accurate?

At any time, publicly or privately, just let me know that you think players A and B should be closer together or farther apart in ranking. I will nudge those two players’ ratings accordingly, no questions asked. I will adjust anyone’s rating except your own. Any adjustments won’t take effect until the next game.

By doing things this way, the ratings are harder to manipulate, vanity/ego has less influence, people are less singled out, picking teams isn’t slowed down, and the median score doesn’t drift.

[Beelzebub]

Its set in edit mode, dont know if that was intended.

[Beelzebub]

just dont want to accidentally mess with something

[Imperial]

Looks good thanks crater.

[Crater Creator]

Thanks guys! Yes, it’s deliberately made editable, so that anyone in the group can use it. Specifically, anyone can punch in player names in the 8 cells of Step 1. I encourage any of us to do just that, since I won’t be around for every game. Everything else, though, is not editable. At least for now, I’ll be the gatekeeper for making any changes.

[Crater Creator]

So I've gotten some feedback from some of you, which is great! It's already making the spreadsheet better. I removed users that have been banned from the server, and I've adjusted ratings on two users based on feedback from multiple players.

Remember, please put your calls for ratings adjustments in the form of "A and B's ratings should be closer together" or "A and B's ratings should be farther apart". This helps you think beyond singling out one player, and it prevents inflation or deflation of the overall median rating.

At this early stage, I'll be making relatively large adjustments, and as time goes on these adjustments will probably get smaller as the ratings get dialed in based on more people's assessments. So if you want to have more influence, send me your feedback sooner rather than later.

[Profile Name]

Captains, dedicated preferably. If we can't manage that then honestly, no spreadsheet is going to save us. I've said my part, now I'll leave it alone.

[Thane]

I appreciate your efforts Crater!

[Crater Creator]

An update for everyone.

One thing I didn't anticipate was how many players weren't already in the spreadsheet. Don't get me wrong; this is a good thing. Not having the same 8 people every night is part of what keeps the game fun. But it has highlighted the need to come up with a rating for players that don't have one.

I've decided a reasonable compromise between speed and accuracy is to put on my 'talent scouting' hat, watch the 'new' player for one game, and then assign them a provisional rating based on that. That gives a reasonable starting point, and then as always the ratings can be refined through feedback from players, like you.

In other news, Derailed has his own spreadsheet going. It's been great to see, because it provides another data set: another person's comprehensive perspective on the ratings. So I've used it to nudge a lot of player's ratings in the Balanced Team Generator spreadsheet, and while there are outliers the two largely agree now.

Ideally the spreadsheet would reflect input from many players. Everyone has a slightly different philosophy and would weight different factors a little differently. The same can be said for any captain, when we do captains. But I realize not everyone has the time to weigh in on every player, which is why I haven't given everyone a spreadsheet to fill out for homework. ;)

Lastly, now that the ratings are effectively cross-checked by three people, I'm ready to take your input on players singly. So instead of comparing X and Y, which people haven't been interested in doing, you can just tell me X is rated too high or too low now.

TL;DR:
'New' players will get a rating after watching them for one game.
Props to Derailed for his own full set of ratings.
You can just say "X is too high/low" now.

[Imperial]

way to go crater!

[Crater Creator]

I enhanced the spreadsheet today, so that it prints a team for all 8 players. Previously it only did one team, so you had to extrapolate that the other 4 should go to the other team. This should reduce wait times in lobbies a little bit more, since it'll be easier for all players to see their team instead of only half of them. As soon as either side is teamed up, the lobby leader can start and the rest will be assigned to the other team.

[Crater Creator]

Since the BTG doesn't lump people into tiers or otherwise segregate them, it has sometimes produced results we're not used to seeing. In particular, it's put the two highest and two lowest players on one team, and the middle four on the other team. This is because the combined rankings of the two teams are closest with those teams.

Some of you probably subscribe to the logic of, "Wow, this teams gets both top-tier players while the other team gets none? They have a clear advantage!" while others of you probably think, "Wow, this team has two players to carry while the other team has none? They have a clear disadvantage!"

In any case, I think it feels somehow 'weird' to a lot of us to set up a middleweight team against a combined heavy+lightweight team. Regardless of how teams are chosen, I do see a benefit in each team having a 'captain', in the narrower sense of having at least one relatively high-level/well-regarded player that can take a leadership role for their team. The group dynamics/chemistry of both teams will theoretically be better that way, over a team with 2 natural captains versus a team with no natural captains.

So I am in the process of updating the BTG algorithm to specifically choose a 'captain' for each team first, and then fill in the other six players. In the first iteration, it will choose captains from among the top four players. It will give equal weight to how highly ranked two potential captains are (the '2 best players should be captains' approach), and how close in ranking they are to each other (the '2 players of equal tier should be captains' approach).

I don't claim this will be exactly the same as captaining, but hopefully those of you who prefer captains can see it as meeting you part way. I think we'll be more comfortable with the results this new formulation produces, and that's important to me, since my goal for BTG wasn't to be a radical departure from what we're used to.

[Crater Creator]

I've finished a round of improvements to the spreadsheet. Briefly:

• The top two players are now always split. Based on your feedback, BTG will no longer put the top 2 and bottom 2 players together on a team.
  
  
• Player names are now sorted in descending order, so that their relative positions in the spreadsheet are more apparent. The top player has a * by their name, for those that like to look up to one person as captain/leader/shot caller/what have you.
  
  
• BTG can now also suggest who should be captain. Spreadsheet captains are the same idea as spreadsheet teams: it can produce a quick, good enough result when we don't want to spend time arguing about what to do. It picks captains among the top 4. It will often pick the top 2, but it can also pick others when the top 2 are uneven in ranking. It also automates the coin toss to see who picks first.

BTG will now produce results that look like this:

Survivors: Alpha*,Charlie,Foxtrot,Hotel. Infected: Bravo*,Delta,Echo,Golf. -BTG

-or-

Captains: Alpha (Survivor) VS. Bravo (Infected). Bravo wins the toss and picks first. -BTG

[Thane]

Nice

[Crater Creator]

Thanks for the recent feedback on ratings for players old and new. In addition to tweaking those values, I've finished another round of improvements to the spreadsheet.

• Using BTG is quicker than ever, because now you can input partial names. You can type "woot, Crater, Bio, Imp," and it will know you mean wootlink, Crater Creator, Biohazard Entropic, and Imperial.
  
  
• In support of this, I updated the roster to use 'full' player names. So e.g. whether you think of him as "Piss" or "Blasta," you can type in either one.
  
  
• If you enter a name not in the list, you'll see a warning that the player doesn't have a rating.
  
  
• BTG now uses the median rating by default for anyone without a rating. It's still best to get a new player on the list with even a tentative rating (hence the warning). But this way BTG makes the best guess it can when it has missing data, instead of just producing an error.
  
  
• Formatting/readability improvements.

[Knee]

Hello, Mr. Crater.

How exactly were the data compiled? Did you analyze replays and record each player's average clear time? How is pacing different from clear time (unless you are counting the ground covered between SI attacks, in which case how is this data measured? Did you measure average distance traveled between SI attacks per player?)

How exactly is "communication" measured? Did you record each player's mic time per game and control for relevant and nonrelevant mic usage?

Is tank ability quantified by recording how many wipes each player got on average when controlling the tank per X amount of chances with tank?

I'm assuming common clear is also averaged per 100 games?

Is "shop time" controlled for relevant shopping and frivolous shopping? If so, how is such distinction determined? What if a player has previously scouted the location of a molly while on the SI team and quickly grabs it as a survivor between SI hits? Is educated shopping considered "shopping," or does it only count against you in the pejorative sense?

What is the lowest ranked player, bwab, gets on a team with heff and has phenomenal clear time and pacing, is bwab awarded unfairly for heff's effort?

If heff has poor support from his teammates while playing tank, does he lose "tank ability" points?