Background: I started a team league in the Denver area, which copied the exact format of the Seattle based Monday Night Pinball. Teams of 10 players. My fear of smaller teams was that there would be too much discrepancy between the best and worst, even though there are provisions in place (no more than 3 players per team in top 1000 IFPA rank, of those, no more than 2 in top 500, and no more than 1 in top 150).
We’ve played two seasons, and it’s been fine. The league hasn’t grown - it actually shrunk from 7 teams to 6 in the second season. I’m considering switching to smaller teams, and forming an A and a B league, perhaps with restrictions preventing great players from playing in B.
Instead of using IFPA rank, I’ve been considering relying on IFPA rating instead. I have a general sense of how it works, but not enough to answer the following question:
For a hypothetical team league, with teams of six players, would IFPA rating be appropriate to use for a salary cap? The rule would probably be written like this: the sum of the IFPA ratings of the players on the roster cannot exceed X. (I have some ideas about what I can do to try and solve for X).
Maybe what I’m ultimately asking has to do with the rating system…would a 1600 player teaming up with a 1400 player be a reasonably even match against two players at 1500?
Rating is too volatile at the moment to be a good indicator of skill at least at the moment. My rating went from high 1500s to 1400 after I bombed out of Portland Pinbrawl for example.
The difference between a 1600 player and a 1400 player is often as simple as how good their last tournament placement was.
Rating does work for really good players who always place highly, but then they also get restricted by ranking as well.
I’d suggest sticking to using IFPA rank for now.
Echoing @Frisbez’s statement for sure. Rating also is highly situational on what tournaments people play. Placing well in a bunch of small tournaments with lower average skill builds your rating up; playing in larger tournaments with higher average skill makes that much harder.
There’s a good percentage of under 1000 ranked players in Pittsburgh being rated up to 2000th.
@Frisbez and @ScoutPilgrim, what is your suggestion for how to build balanced teams according to ranking using a salary cap like method?
The IFPA site says the decay is .3 per day. Can you explain in a little more detail @pinwizj? All along I was assuming that ranking is “who is playing the best right now” and rating is “who is the best regardless of potential inactivity”…but I guess not?
In Pittsburgh, the Steel City Pinbrawl is about to change from a “X players over Y” restriction to a draft of players. Players are put into value brackets based on their ranking, and a team’s rosters cannot exceed a certain value. Since the old league was hardcore burnt by up-and-comers (and a grandfather clause that let them be treated as over-500 players), the TD has the final say on what a player’s value is.
Can you share what your conversion is from ranking -> value brackets?
I like the idea of a draft, but it may not be possible in my case because you could be drafted by venue that’s 60 minutes from where you live.
I didn’t come up with the system (shout outs to PAPA tech vet Nick Jaquay for that) and the system isn’t finalized yet. I can try to figure out how it’s currently going to be implemented, but otherwise I’m in the dark.
Even if you don’t go for a draft, you can try balancing teams around a certain “budget.” Lock players into their value for the season so that nobody gets forced around mid-season for doing well in other events, and if trades are a thing, make sure no team goes over-budget.
I’m not too well versed in Glicko, but there is a “penalty” for not playing. Your rate would continue to slide due to that inactivity.
In the portland pinball team league I believe in addition to the max 3 top 500 players we added the top 15-20% of players from the previous league season (based on win percentage or some kind of similar maths) to those 3 players. So if you don’t go to many tourneys but rip in league you basically get one season before your skill is properly judged and you get added into the equation.
Also I think this season we added a rule that eliminated the cap of 3 players but put a maximum on total games that top 500 players could play for a team in a night (essentially the amount that 3 players would be able to play). So you could have 4 or 5 top 500 players on a team but they would all play less games. I have some issues with that rule but I don’t think it’s necessarily a gamebreaker.
After looking at the players from last season and sorting on various things, I’m starting to think that Eff % might be the way to go.
I asked Josh about eff% some time ago. Apparently, it is the percentage of first-place points that a player has captured over the 20 results that count towards the ranking.
I’m not sure how well that expresses overall skill level though. For example, if I exclusively play in tournaments where I’m the best player of the people who attend, I’ll have a 100% efficiency percentage. But that doesn’t mean anything in absolute terms. The players I’m playing against could be the top 20 players in the world, in which case I’d be the most skilled player in the world. Or they could all be rank beginners who still double-flip all the time, in which case I’d possibly be a very unskilled player, only just good enough to beat people who know nothing about pinball and have lost their eyesight years ago…
I think Eff% is based on your rolling 3 year activity, not just your top 20.
This makes more sense to me.
I think Eff % more accurately reflects the skill difference between players…and thus their “value”.
Local players with ratings in the 1600s have an average Eff % of 27.
Local players with ratings in the 1200s have an average Eff % of 5.
In a long heads up match between someone in the 1600s and someone in the 1200s, I’m certain the final result would be much closer to 27-5 than it would be 16-12.
Can we get an easy way to pull all of the player data for statistical purposes? I would like to study these relations for fun when I get free time.
If you go nuts, I actually don’t want any decay factored in for what I’m doing. I’ve got one player who has a rating of 1291, but it’s only because she hasn’t been playing lately. Her rating going into the last event that qualified in her top 20 was 1560. That’s a big difference, and something to exploit in a salary cap situation. (Not to say that wouldn’t demonstrate skill on the part of the captain, but for the last two seasons, the best way to field a great team was just to find people who previously competed regularly but haven’t been lately).
Nick and I have been talking about a draft that borrows heavily from the way Ultimate Frisbee leagues (at least in Pittsburgh) do drafts. It puts players into equivalence classes for points, and allows “baggage” rules to allow players to come as a package (i.e. partners or roommates play same day/location) but are still fair for draft purposes. Each round the teams’ picks are ordered from lowest to highest point totals, which balances pick order when 2 highly ranked players baggage. The league officials and teams work to ensure the starting points are fair and account for untested players (don’t play many tournaments) and fast risers where global ranking systems fall down. It’s our first time for pinball, so we’ll see how it goes this fall. @ErinK has participated in this system in Ultimate for years, and it works great for them.
Getting accurate assessment from players about their ability to participate, is critical for fairness. There may be a way to address location preferences this way (i.e. if it’s in the east bay I’ll make 20% of the league nights, but if it’s in the city I’ll make 85%).
Against all events in the past 3 years, not just best 20.
Eff % and Rating are stats that are only meaningful when you are trying to do your best or are trying improve your Ranking. Things like missing league finals, or ditching your classics II entry to focus on PAPA B can pull this down a lot from where it “should” be.
If you’re trying to use something solid for the foundation of the team league, you have to keep in mind that these 2 stats are ones that players can purposefully tank as fast as they can play in events. Ranking will last much longer becaues it’s only pulled down by time.