Monday, January 30, 2017

Match Structure

Currently, match structure at a Magic tournament is this: Play best-2-out-of-3, unless you run out of time. Except 2HG and other multiplayer formats that go long, in which case you play 1 game, until time. Or if it's a feature match at the end of a big tournament, then you play to best-3-out-of-5.

The winner gets 3 points, and the loser 0. Or if it's a draw, 1 each.

What if instead, players played 3 games or until time runs out and scored 1 point per game they win?
In that case, match winners would usually score 2 points instead of 3 (but could score 1-3), losers could score 0-1 points, and a drawn game would score nothing but wouldn't invalidate previous games.

That's clearly fair. Is it more or less fair than what we've got now?

Because it reduces the average spread between winners and losers, it makes it easier to come back in a tournament. If you eek out the match two games to one, you get 2 points and I get 1. If you win every match of a seven-match tournament that way, you'll end with 14 points. If I lose 1-2 twice more but win 3-0 four times after that, I end with 15 and earn a better standing than you. It's subjective, but that sounds better to me.

A few effects:

++- Players wouldn't drop as early, because they still have hope. This is clearly better, though it does have the side effect of preventing players who will ultimately lose from side-drafting or going home as early.

--+ More rounds would go to time. That's a not a factor at large tournaments already sure to do so, but would cost small tournaments maybe 5-30 minutes across the entire event. On the other hand, players spend more of their day playing Magic.

---+ Slow decks get penalized significantly. If I can guarantee I win every match, but only have time to finish one game in each, my score will only be 1/round, which is only as good as winning 1-2. This has the positive side effect of making games faster, but is seriously detrimental to control decks and could upset Magic's core metagame balance.

+ We can use the exact same match structure across all formats, instead of tweaking it for multiplayer, for large decks, or for feature matches. (We'd still give feature matches more time on the clock to accommodate more games.)

-- By reducing variance, we reduce the spread in finals scores, increasing the number of ties, and thus the need to use tie-breakers.†

+ Going first game one is impacted when you're more likely to go to three games. I believe it should be slightly less beneficial here, but I'm not sure.

± Sideboards become more relevant. (I don't know if that's good or bad.)

+ It will feel much better when you can redeem yourself/deck by winning the third match after losing 0-2.

+ You will feel masterful when you win a match 3-0.

- You'll feel worse when you lose 0-3.

+ Drawn games are less impactful. If we go 1-1-1, we each score 1, the same as now. If we go 1-0-1, the winner gets 1 point and the loser 0, where currently the winner would get 3, which isn't quite right. If we go 0-0-1, we get 0 each, which is rough, but perhaps not unfair.*

* What if instead of a draw granting 0 points to both players, it grants 1? 1-1-1 would score us 2 points each; 1-0-1 would score the winner 2 and the loser 1; and 0-0-1 would score us 1 point each. I think I like that better. It's been called to my attention that awarding players a full point each for a drawn game would too often lead players to draw games they could win.
…What if a draw scored the most intuitive way possible: ½ point each? That's ugly, which is why I didn't mention it until now, but let's just see: 1-1-1 scores us 1.5 each; 1-0-1 scores the winner 1.5 and the loser 0.5; 0-0-1 scores us each 0.5. That's clearly the most fair, but I'm not entirely sold on it being worth decimals/fractions. One last note: A record of 2-0-1 is also possible; the winner would get 2 or 2.5 points depending on how we score draws and the loser would get 0 or 0.5.†

† Maybe we offer a bonus for actually winning a match. Suppose the match winner gets an extra point (or ½ point). This defeats much of the point and I don't care for it, but it's a potential compromise to keep variance up and ties down.

w is winner bonus
d is draw points
w=1 d=0 w=½ d=½ w=1 d=½ w=0 d=½
record win lose win lose win lose win lose win lose
0-0-1 0+d 0+d 0 0 0.5 0.5 0.5 0.5 0.5 0.5
1-0-0 1+w 0 2 0 1.5 0 2 0 1 0
1-1-0 1 1 1 1 1 1 1 1 1 1
1-1-1 1+d 1+d 1 1 1.5 1.5 1.5 1.5 1.5 1.5
1-0-1 1+w+d 0+d 2 0 2.0 0.5 2.5 0.5 1.5 0.5
2-0-0 2+w 0 3 0 2.5 0 3 0 2 0
2-1-0 2+w 1 3 1 2.5 1 3 1 2 1
2-0-1 2+w+d 0+d 3 0 3.0 0.5 3.5 0.5 2.5 0.5
3-0-0 3+w 0 4 0 3.5 0 4 0 3 0

My conclusion is that this sounds a bit better than the current system, but it's far from being a mandate, and may not be quite worth the effort of change.

What other effects did I overlook? How do you rate them differently?
Do you think system is better? Worse? How much? Why?


  1. Edited 2/1/17 to include helpful observations from Twitter:
    That scoring a full point each for a draw rewards intentional or purposeful draws too much.
    That going first will be affected.
    That sideboards will be used more.
    That more tiebreakers is worse than originally rated.

  2. Easy solution to the convoluted draw math: Award two points for a game win, one for a draw, and zero for a loss. It'll probably take some mental readjusting to get used to such high point totals (If you go 5-0 at a GPT, you can have anywhere between 20 and 30 points - more than most people get at GPs nowadays!) but it's nothing the system itself couldn't handle.

    One other issue is that this will still make Swiss rounds be scored differently than single-elimination rounds, which doesn't solve the "feature match at end of event" problem.

    1. Doubling is a nice way to avoid halves.

      I didn't even think about single-elimination. Do those track points at all?

    2. For event reporting and coverage purposes, yes. The story of Pro going 3-0 is quite different from the story of Pro eking out a 3-2 win.

      I'm not sure if points affect tiebreakers after a top cut.