Group order was determined by seeding.
@Shep The “balanced” algorithm was used for arena selection. Since MP will pair people first and arenas second you’ll end up with a few outliers in a format with multiple games per round.
I’m not entirely happy with how it displays so you can’t navigate to it yet, but here’s a page showing all players and the arenas they played. Bigger circle equals more games played. Hover each circle for a small tooltip stating how many games.
Link: https://matchplay.events/live/finalbattle2016/stats (looks a bit like The Matrix!)
Looks like a pretty good distribution to me, but @bkerins is the mathematician
Just for myself I was surprised to have 10 machines I didn’t play, and 2 machines I played 3 times. Alex Rudder had 12 machines he didn’t play, and ended up on Stars 4 times (out of 13 rounds). Kevin Duffy played one machine 4 times, four other machines 3 times, and 11 machines not at all.
I’d say the one to look at was Sanjay, who played Fireball and Jackbot 4 times each, two other machines 3 times each, and 15 machines not at all (out of 34 total machines).
I suspect there is a way to do this without ever having someone play a machine more than twice, but that may not be correct, and it may take quite a bit of work to pull off. One option is to “re-roll” the machine selection and take the best case out of maybe 100 re-rolls. For Pinburgh, each machine set is assigned, starting with the group where there is the most limited set of options of what they can play.
In any case, thank you for building these tools. They are outstanding.
The pairing algorithms is some of the most fun work in Match Play (most of the work is putting buttons on screens which is more tedious).
MP will do 1000 random “sets” and pick the “best” one. 100 random ones is far too little and there are diminishing returns when increasing it much beyond 1000 (and performance starts to suffer).
There are some quirks and some limitations that makes these repeats happen:
It’s all random so the algorithm can get backed into a corner. I’d like to use a genetic optimization algorithm, but haven’t had the time to explore. It should guarantee better results.
The optimizer assigns a “cost” to each “set”, but cost is linear. I.e. it increases the same amount to assign someone Stars for the fourth time as it does to assign Jackbot to someone for the second time. This would be a simple change that should reduce outliers (adding that one to my list!).
Unlike Pinburgh MatchPlay will go through groups in random order when assigning arenas. This is because unlike Pinburgh MP can’t be sure that there are enough machines for all groups and I don’t want MP to be opinionated about which groups not to assign arenas to.
MP will also not repeat an arena for any “game 1” matches. E.g. if you have three games per round, Stars can be assigned only three times (once for each slot).* The random order then makes it harder to get perfect assignments.
The mathematical models I’ve seen have limitations that make them unsuitable for pinball tournaments. They all assume that no players leave or arrive late and that no machines are replaced during the tournament. If only this was true for all pinball tournaments!
That’s why I’m putting my faith in optimization algorithms. The 1000 iteration random optimizer is “pretty good”, but I’d love to replace it with an algorithm that can obtain better pairings using the same amount of iterations.
* MP also supports machine banks where the procedure is different. Here a bank of machines is treated as a unit and the banks are assigned rather than individual machines.
Another awesome tournament- I’m proud to say I’ve been to every Final Battle and plan on doing it again next year. My only issue is i wish it would go another round or two Thanks so much to Jim, Mark and everyone who contributed.
Regarding future game selections- I wouldn’t mind seeing it go to banks instead. It always seem fairer to me when games in a match are from different eras. Also banks will eliminate a top group from getting all long playing games and dragging out a round longer than it has to be.
On another note- Having a food truck there was great addition. Next year it would be even cooler if it could be arranged to have a food truck serving lunch, then another one serving dinner and a third doing a midnight meal. I’m guessing there would be a enough business to make it worth their time.
I recommend making this exponential: perhaps putting someone on a machine for the third time could cost 3x putting someone on for a second time, and a fourth time could cost 9x. The linear cost makes sense now: once you decide to put someone on a repeat game, it doesn’t matter which, so you’re getting lots of 3-peats and 4-peats. (We should tell Brackelope, too…)
Genetic or other types of algorithms would be amazing, but are crazy complicated to program, and usually when they fail they fail spectacularly!
Match Play Events Open Thread
Isn’t that what makes them great? I agree there would have to be a lot of simulations run before letting something like that loose. Perhaps a formula for acceptable “cost” with a fallback to a random optimized solution since there should be no surprise results for the pairings. In that way arena assignments are very orderly.
We will try to make the game selection a bit better next year. The reason we opted against banks this year was because last year we tried that but with a limited selection of games (who says that when there are 45 setup?) substitutions start getting tricky with all the overlap. We ran into issues doing that. Next year we will be trying to sort games by length of play and give every group one game of different lengths. 1 long. 1 medium and 1 PARAGON hopefully this lets us get another round or two in and creates a bit less repartition for some.
Points are live now: https://www.ifpapinball.com/tournaments/view.php?t=13774
Hope 60 points counts on your record Bowen!
Golf points are up to. 24 points to Mr. Monica for 40 something people practicing and a not full top isn’t too bad.
Keep the suggestions coming.
Please tell me you did this on purpose…
Haha. iPhones don’t like Modica.
I’m dreaming up a 12 hour battle. Dreams dreams dreams or maybe nightmare. Quick question which I think I know the answer to at the end of the 24 hrs do the games just stop right at 24 or do you finish any game started within the 24?
I would imagine you just finish the game but just curious. I originally thought about having a finals but I like the idea of just ploying through!
run something in columbus!
The current round is completed. There are so many rounds that there is generally no need for a final, though with 12 hours of play you might consider it.
I had the exact same idea. The 12-hour Pre-Sanctum Mini-Battle.
Your assumption is correct as Bowen stated. Minimum 24 hours
Last year with 100 really good players we squeezed in 13 rounds of 3 games a round, 4 player matchplay. That’s 3600 competitive games played in a single day. Our games racked up some big play counts. Many people never stopped waiting for the next round.
I bet once you do a 12 you’ll wish you did 24.
If you want any other info on exact format or how we run it let me know. Too some work to find ways to keep players not doing well with no chance to come back to the top to keep playing while also trying to their fullest. I think it’s worked out well so far.
I’m dreaming of a 12 hr stream then 12 hr tourney. Or 24/24 hahah. We’ll see. May be coming to a Columbus near you…If you’re in Ohio now that Clancy’s is out @PinballNarcissist
right now we have 18 games, could prob get that up to 22 or so, i want to have some on reserve for replacements and I want to always have some games open so there is ‘less’ waiting
Waiting just comes from that one group that takes 2 hours to play 3 games while everyone else takes an hour
And when that group just happens to be playing Victory!
How did you guys combat this? I love this format but was wondering how to address this potential issue.