Top 100 ? I don't think many of those leagues even have 100 CA teams.
You could expand it to "all CA teams". You'd get something similar to SR's strength of schedule numbers, which are about the same.
How do leagues stack up?
- Thread starter Speed
- Start date
You could expand it to "all CA teams". You'd get something similar to SR's strength of schedule numbers, which are about the same.
RandomSoccerFan
GOLD
It's the top 10 teams of each league. Previous lists have used the average of the entire league. In some cases, the results are the same, and in others - they are significantly different. Leagues that have a huge variation from the top teams to the bottom teams are helped more in this view, than leagues that have a much smaller variation from top teams to bottom teams. What it shows is that the top teams, in the top leagues, are pretty close to eachother - moreso than the average of each league. So if the top team in one league were to compete against a top team in another - it would be a good game.
What it does do, is illuminate which leagues consider themselves top leagues, but their best teams aren't great.
For all lists like this, it would be even better if you could drill down to see the 10 teams chosen, and confirm that they are in the same league that you believe them to be in. There is certainly some weirdness, as applied to NorCal.
What it does do, is illuminate which leagues consider themselves top leagues, but their best teams aren't great.
For all lists like this, it would be even better if you could drill down to see the 10 teams chosen, and confirm that they are in the same league that you believe them to be in. There is certainly some weirdness, as applied to NorCal.
TeamDadJokes
SILVER ELITE
Pretty bad look for Elite 64 to be the bottom of the barrel. The only question is what is the cross over to NL PRO as a I know most those leagues have teams in both.
SurfSupDude
SILVER
Exactly. Although — and I’m focusing on the girls side, that’s not really the case with ECNL, as a full 2 goal differential is quite a lot.
But top teams across the next five leagues are about the same (within 1 goal).
RandomSoccerFan
GOLD
Maybe, but keep in mind that even in this age group, in California, the ECNL 2009G teams near the bottom are more than 5 goals off the top teams. While some may believe that the ECNL name alone means good quality teams, there are plenty of teams that would be beaten handily by those in other leagues. What this data is telling us is that the best ECNL teams are favored competitively against the best teams of other top leagues, and there is a measurable gap.
I believe there's also a bit a "banding" of ranking based on leagues going on. To address teams from different leagues need to play each other more often.
Here's an example. If league A club members hold 1-10 rankings and league B club members hold 11-20 rankings + the leagues never play each other or other clubs from league C,D, or E. Teams from league A will tend to stay highly ranked even if they lose a lot and teams from league B will tend to not be highly ranked because they don't play teams from league A. Even worse for league B is they have more to lose if the lose.
The only way I can think of to address this type of situation without teams from different leagues playing each other is to add some kind of multiplier after X number of wins that affects ranking.
RandomSoccerFan
GOLD
I think that thought may have some intended logic to it, but I also think that logic is faulty. The teams aren't banded, and the leagues aren't banded. The results are what they are. There is enough interplay, even if through intermediaries, that the predictions speak for themselves. If the end result of gameplay is that a significant number of teams in one league show better results than a significant number of teams in another league, however the leagues are compared - one is going to show better results both individually and via aggregate. Adding a win multiplier is certainly an option - but remember these kind of tweaks are tested over thousands and thousands of games, and the resulting predictivity can be shown immediately. It's these type of tweaks that have been done for years to optimize predictivity, and making those suggestions because of a team or a league without having the huge amount of actual game data to test them - is just that - it's a hunch that was likely already thought of, tested, and implemented or not.
I think we're saying the same thing.
With interplay between leagues everything works with the current model.
Without interplay between leagues some leagues won't be able to rank as highly as others.
Another way to look at it is a team not in the highest league could win for 4-5 years straight + not rank very highly. Espicially if they never play outside of their league.
Personally I believe leagues will allow interplay less and less over time either directly or indirectly + clubs will be ok with it. It's just easier to state that you're the best vs proving it all the time in games.
Code
SILVER ELITE
It's pretty clear. If your girl is not playing in ECNL, or your boy is not playing in MLS Next, you should be looking really closely at the local NPL teams. Cut out the extra costs and travel, to play the same level of competition as all the other letter leagues. Spend the extra time you have on the weekend getting in individual training intead of riding in a car or flying. Keep trying for the ECNL or MLS team while you play NPL; don't be a sucker thinking there is value playing teams in other Counties/States that are the same as you local teams.
what is the difference between NPL and National League PRO?
RandomSoccerFan
GOLD
We are not saying the same thing at all, and I still think you are completely misunderstanding. There is plenty of overlap in all of the leagues, including ECNL and GA, to show that one is several goals stronger. It isn't because of banding, it isn't because of mathemetical idiosyncracies, it's because the teams in one league are better than the teams in the other league, by a very significant margin. Any lesser league can look to any excuse why their individual team's rating is lower, or why the league as a whole is lower, but there's really nothing to it but excuses that don't hold any water.
It would be a better world, perhaps, if everyone played eachother directly - but this isn't the world we live in. Playoffs are the method that those within a sanctioned league can show they are better than others, and tournaments that attract the top teams from multiple leagues are a glimpse into how it can be. From a ratings standpoint, it's this type of interplay that confirms and readjusts the ratings from team to team, age to age, league to league, etc.
Limiting it to the top 10 would favor a larger league and punish the smallest. For instance, GA seems to have only 9 teams, so you've got the best and the lowest ranked team in GA, versus only considering the top half of ECRL and ECNL. Not sure how many teams the other leagues have, but if they have 15 or more, it would skew them upwards as opposed to having 10 or less. Taking a cursory look at the soccer rankings app's ranking of Calif teams, it seems that the best teams are all ECNL. The next group that shows up most is GA and then ECRL, but they are pretty close. DPL even seems to have far more teams higher up in the rankings than NPL, so that seems hard to square with what this new listing provides.
TeamDadJokes
SILVER ELITE
NL PRO takes top teams from other states that compete under the USYS umbrella and has their own national championship series from it. It consists of top qualifying teams from California from NPL, Elite 64, CSL Premier, and California Regional League to compete in USYS National Championships. NL PRO is the top flight of this championship series. The non E64 teams go through a regional playoff (ie Far West Regionals) on their pathway to the national championships.
SurfSupDude
SILVER
You can't really base it on team names alone. Many NPL teams do not include "NPL" in their team names. In fact, one ("So Cal Blues Call DPL Fly" in SR) is even listed as a DPL team by its name in SR. But if you look at the sources for that team, you will see a lot of "NPL" results.
Look here (GotSport ) for a list of actual SoCal NPL teams per SoCal NPL Fall League standings.
SurfSupDude
SILVER
ETA: Posting G09 NPL Standings to keep the data set consistent. But it still holds true; many NPL teams do not include "NPL" in their team names. GotSport
TeamDadJokes
SILVER ELITE
I think Mark is probably pulling his data from the source of where the games are being recorded. So if it’s got sport then he’s able to parse out which league the data is from and then make his rankings accordingly. What I was trying to get at is, as NL pro for example that many of these teams have cross over within these leagues if you’re not ecnl, GA, DPL or ECRL.
Elite 64, NPL, SoCal, CRL/ Coast, all participate in USYS Nationals which also crosses into NL PRO. So are these results having duplicate team entries skewing the data set?
Elite 64, NPL, SoCal, CRL/ Coast, all participate in USYS Nationals which also crosses into NL PRO. So are these results having duplicate team entries skewing the data set?
Statistics is the science of producing unreliable facts from reliable data.
In this case a specific interpretation was presented. There's probably 10+ different ways to present data which would have implied varying results. That's the benefit of being the one that compiles and presents what's available.
TeamDadJokes
SILVER ELITE
I think you missed my point of having duplicated teams in the leagues mentioned. Specifically NL PRO. No need to mansplain how stats work brother…
I saw it as well + I agree with you. It's just not worth getting upset about. Anyone involved in youth soccer looked at that graphic + instantly discounted it because it doesn't align with general preconceived perceptions.
LIke I said before, the person that collects the data gets to present it however they want.
RandomSoccerFan
GOLD
You're right, there are any number of ways to present the differences in league. One of the common ways that was done before, is taking the average rating, of all teams that were assigned per league. This helped leagues that had a limited top to bottom spread, and hurt leagues that had a large spread. For example, an average NPL team is somewhere near the midpack of NPL 2 (assuming ECNL-RL, NPL1, NPL2, NPL3), while the average MLS N team is somewhere in the middle of MLS N. The top RL team vs. the average NPL 2 team is going to be expected to be a blowout, while the top MLS N team vs. the midpack MLS N team is expected to be a win, but not nearly the difference implied.
Another way to compare, is to ask the question - how do the top teams in league A vs the top teams in league B stack up, and the way this could be done is choosing the top 10 teams by league. Of course - if a league only has 9 teams total, that means all of them and an average of them would show pretty weak, as the top 9 teams in league would include both the top teams as well as all of the bottom teams. Some would say that league that has less than 10 teams in it state-wide is already weak - but that's a different discussion. A potential fix might be to only take the top 10% of teams rather than the top 10, but for a league that has 300 teams, that's comparing 30 teams against another league where the top 10% means less than 1 team. Top 10 seems like the better choice of the two, given only those 2 options - but in either option, the disfavored league would complain about the results.
One of the things that this type of view does show pretty clearly, is that the top teams in several leagues, are much closer together, than the earlier comparison of average teams in league. It is answering a different question, "How would one top team do against another", rather than "How would an average team of League A stack up against an average team of League B". Both questions have value, but maybe the first is a better way to compare leagues, maybe some disagree.
Drilling into the league descriptions to see which teams went where would end at least some of the questions about what each league represents, and confirming that the team association / league association jives with local understanding. It's quite possible now that the underlying data has issues, especially if the particular league is harder to identify - and while it would be good to know how it was done - it would also be good to just know the top 10 teams, so there would be inherent validation over the data, and therefore much of the results.
All of this becomes irrelevant when comparing one specific team against another specific team, where their ratings can be compared directly, and a probability between the two can be compared and displayed. It will always be more accurate in predicting which team might best the other on the day, rather than comparing the leagues they play in to make the same type of inference.