Draft Viewer

[Jason Waddell]

One of the things I do as a statistician is make cute little graphs to visualize data. The graphs show the table positions, colors and records for each player. I can look at one of these and immediately remember decks and various things. It is also useful for people to look at and think about if they should have moved colors, if they were reading the signals appropriately, or whatever. The small circles represent a light splash (1-3 cards usually). 

Also, I know it looks like a total jerkhole thing, but I've 3-0'd every draft since we started making these. So I'm not just hand-picking the ones that make me look good. Also helps for seeing what colors are doing well, and track results. Each of these graphs takes about 2 minutes to make (automated code, just data entry). 

If people find them interesting I could make a website that generates the graphs automatically. 

Draft 1:

Draft 2:

Draft 3:

Draft 4:

[Eric Chan]

That's..  that's awesome.

What do you have to punch in after every draft? Everyone's decklist?

Interesting that you haven't gone less than three colours in any draft so far. Do people not know to punish the greedy?!

[Jason Waddell]

My last deck splashed for Shardless Agent and Cyclonic Rift. The one before for Lingering Souls, Vindicate and Ajani Vengeant. Not super greedy, certainly by my standards. 

As for the code, I just adjust calls like this: 

drawMat(player = 1, label = "Niki (2 - 1)", col = "WBr")

It really takes no time at all. 

Anyways, I have found player skill to be a much more important factor in the results than number of colors drafted. 

[Rob Dennis]

in the suuuuper long-term plans for cuesbey is a "better than tappedout" draft/AI implementation, and it's always great to have ideas for meaningful representations of draft results, so I'd be interested in the code, for sure

[Eric Chan]

Oh, man. Yeah. On board.

Calvin will tell you that all I do all day is complain about the tappedout.net draft AI. I don't even care about good AI - something halfway decent would be a godsend at this point.

I guess I have to come clean and admit I'm a programmer, too. I'm not exactly the type of person who codes at home for fun, or just to learn new things. I leave that at the office. (Boy, my boss won't read this, right?) But for a project like this, I would be ecstatic to pitch in. So if and when you get started, if you need front-end help, or back-end help, or just someone to sweep the floors, I'd be more than willing to lend a hand.

[Jason Waddell]

Sure. I do all my coding in R (statistical programming language), so you'd probably need to re-implement it entirely. 

[Rob Dennis]

not really on topic, but since we're talking cuesbey, just uploaded a big feature (one of the last before initial deployment/release): heuristics, from the wiki:

https://github.com/rdennis463/cuesbey/wiki/Mockups

Note here the concept of heuristics in action. Cube designers always seem to balance around certain assumptions, and this screenshot shows how "assume_on_color_cmc_for_mono_color_hybrids" makes Spectral Procession CMC=3 and "token_spells_are_creatures" makes it a creature 

[Jason Waddell]

Hey Rob, it's looking really good! Quick question. I'm in the midst of a rather sweeping overhaul of my list, and when it's "done" it would be interesting to use this tool to compare the before and after. Would it be possible for me to send you the two lists for you to run the comparison? If so, what format do you need things in?

[Rob Dennis]

I'd be happy to. :-) All iI need is card names before and after in separate text blocks, one name per line.

Optionally, some insight in how you prefer to lay things out, e.g. Do you list colorless creatures separately from non creature colorless? Do you do separate blocks for each color pair/wedge? Stuff like this is customizable and intended to be user editable, the default right now is what's in the screenshots, and then lands/colorless/multicolor is just one lost ATM.

Also, you could let me know what heuristics you'd like on. This link is to a test where I'm asserting that each of them are supposed to be doing and is as long as it's readable (I've tried phasizing that) you can infer what the heuristic is trying to help with:
https://github.com/rdennis463/cuesbey/blob/master/cuesbey_main/cube_diff/tests/test_cards.py#L117

e.g. X spells are often consider infinite mana, instead of the actual cmc: the link is resting that devil's play has that option.

So far it's been coded around assuming singleton, but I needed to fix that anyway and it's not a huge deal.

[Jason Waddell]

Alright, I'll get all those things together when I have the new list to a more stable state. Thanks!

[Jason Waddell]

The win-streak ended, game 3 of my final round. Only 5 people this weeks, so we tried something new: forced shards drafting. Each player was assigned a shard and we went from there. 

[Jason Waddell]

Another draft graph. Played with a bunch of first-timers this weekend. Last week people were complaining about red-white aggro not being strong enough. I basically forced that archetype to prove them wrong. Until Dark Confidant came along, that is.
 

Bonus deck tech: 

[Rob Dennis]

I like how many targets that verdant catacombs has, geez

[Jason Waddell]

If there's one thing I don't do, it's fail to find. 

[Jason Waddell]

Another graph. Hooray 8-mans!

[Nathan Weizenbaum]

How do you calculate the colors for each deck?

[Jason Waddell]

I ask each player "what colors are you in" and then they say "black green splash red" and I write down "BGr" and move on. 

[Nathan Weizenbaum]

Makes sense. That's way simpler than my method.