Club Team Tiers

[Jamisfoes]

The app does a remarkable job at predicting game outcomes. I looked at one age group for Surf Cup and found that it correctly predicted the outcome of 25/27 games, with only one game that was truly an upset (the other two teams differed by less than 0.5 ranking unit so was basically a toss up). There were several games were the difference in ranking between the two teams' rankings was less than 1 and it correctly predicted the outcome in all but one of those games.

I follow several teams on the app and am familiar with a couple of different age groups and from time to time teams get shuffled and their results intertwined. This could be a glitch in that app (that I believe relies on how teams' names are entered for competitions), especially when team names are similar or shift between seasons or it could be users incorrectly attributing results to the wrong team. If you notice an issue with a particular team and know the team well the app gives you the ability to correct mistakes (add or remove sources). I've found that as the season progresses and new teams have the chance to play the rankings do a really great job and become more accurate with time.

I agree this app works most of the time. If I just use a coin to predict the result, I can get it right 50% of the time. Is 90% prediction hard to achieve? Don’t know. It will depend on how far apart the teams are ranked. If the first ranked team play the second ranked team, I suspect soccer ranking is as good at predicting as my coin.
What I do know is a team with that record got grossly misplaced at the top of the ranks. I want to know what happened there. I am not against Soccer Ranking but I don’t like the response from RandomSoccer that the calculation fixes itself after a few iterations. But for $10 a year, maybe my expectation is too high.
It’s also true that this app is not endorsed by any soccer governing body yet we are citing it like the Bible.

 

[polygon]

I agree this app works most of the time. If I just use a coin to predict the result, I can get it right 50% of the time. Is 90% prediction hard to achieve? Don’t know. It will depend on how far apart the teams are ranked. If the first ranked team play the second ranked team, I suspect soccer ranking is as good at predicting as my coin.
What I do know is a team with that record got grossly misplaced at the top of the ranks. I want to know what happened there. I am not against Soccer Ranking but I don’t like the response from RandomSoccer that the calculation fixes itself after a few iterations. But for $10 a year, maybe my expectation is too high.
It’s also true that this app is not endorsed by any soccer governing body yet we are citing it like the Bible.

The thing is, does any of this really matter? Who cares if a team mistakingly gets placed towards the top of the ranking for 24 hours because of some glitch? It seems you're the only one who really cares. If you don't like that app then don't invest the $10 annual fee. It's a cool little tool that gives parents, tournament organizers a means of seeing roughly where teams fall in the landscape of youth soccer. This is especially cool in that it allows a way of comparing teams who play in different leagues (or in different states) who rarely have the opportunity to play each other. And 90% prediction rate is much better than 50%. As I mentioned in my previous message, in my exercise of comparing predictions with actual results, there were 7 matchups in Surf cup where the difference in the ranking of the teams was less than 1 and it predicted the correct outcome in 6/7 of those games (86%)...overall for the group it predicted the correct outcome 25/27 times (93% of the time). That's pretty darn good! It even does a good job of predicting outcomes across age groups.

I think it's an amazing tool!

 

[focomoso]

but I don’t like the response from RandomSoccer that the calculation fixes itself after a few iterations.

But that's literally how the algorithm works. Imagine you have 4 teams in your pool and you don't know anything about them, after the first round of games, you have to give the same value for every goal. Say A beats B 1-0 and C beats D 1-0. A and C will be tied for first and B and D tied for third. Then in the next round A beats D 10-0 and C beats B 1-0. You now have to reevaluate even the first round results because you now know that B is better than D and so A's win over B has that much more weight. That's how all of these algorithms work. It's an iterative process that works its way through as more results come in.

And again, the app just updated to the new age groups, so there's bound to be some messed up data in there as teams names change.

It’s also true that this app is not endorsed by any soccer governing body yet we are citing it like the Bible.

I don't see anyone doing that. All of the matchups are given in likely outcomes. Nothing's set in stone and 90% accuracy is impossible given the randomness inherent in soccer. It's the best app out there for this, though, so it's the only thing we have.

 

[Jamisfoes]

But that's literally how the algorithm works. Imagine you have 4 teams in your pool and you don't know anything about them, after the first round of games, you have to give the same value for every goal. Say A beats B 1-0 and C beats D 1-0. A and C will be tied for first and B and D tied for third. Then in the next round A beats D 10-0 and C beats B 1-0. You now have to reevaluate even the first round results because you now know that B is better than D and so A's win over B has that much more weight. That's how all of these algorithms work. It's an iterative process that works its way through as more results come in.

And again, the app just updated to the new age groups, so there's bound to be some messed up data in there as teams names change.



I don't see anyone doing that. All of the matchups are given in likely outcomes. Nothing's set in stone and 90% accuracy is impossible given the randomness inherent in soccer. It's the best app out there for this, though, so it's the only thing we have.

I agree with everything you say. It’s the best app out there for this purpose. I am just bothered by its placement of a team whose record doesn’t justify it. If that team had beaten TFA pro once then I can see it. But that team has 6 games under their belt against bad teams yet they got placed 4th in California. Still would like an explanation for this. I don’t see how updating to a new age group can be used to explain this glitch.

 

[RandomSoccerFan]

But you've already received a likely explanation, along with the advice early on that if it wasn't satisfactory, contact support and complain yourself. Being accusatory to posters here who have nothing to do with the app other than being happy users hasn't been terribly productive, no?

If ratings are out of whack, the best answer since this app was released - and prior when it was YSR on the web - has always been to have some patience and see if it resolves itself over time. If not, get in touch with Mark. Having a hissyfit about a single suspect rating probably isn't the appropriate response. It may take a few days after the weekend before the crawler gets to the particular score site to load the data. It may take a day or two as all the data is run through a few times if there is something weird. None of this is realtime - or has to be. Ratings/rankings aren't going to change markedly from week to week, or in many cases even month to month.

I hope he can track down why some of the initial manual ratings sometimes look like they got the math wrong if you compare each by hand (like this one was), but as expected, after a few recalcs it has resolved itself, and now shows as attached. I personally don't think this is due to the age shift on Aug 1, as it shows up from time to time if you merge teams. But the age shift is a very valid reason why rankings are a little wonkier right now, especially in 06/07, and those do get better as more data comes in each new season. Ultimately, none of this is a big deal mathematically, and as other posters tried to share - it's not worth getting upset about. But you do you.

 

[Dale Earnhardt Jr Jr]

What's the name of this app?

 

[RandomSoccerFan]

Soccer Rankings.

16 page thread on it here on this site.

Apple app store

Google Play store

 

[focomoso]

What's the name of this app?

It's just called "soccer rankings".

 

[sigmabody]

And again, the app just updated to the new age groups, so there's bound to be some messed up data in there as teams names change.

Question I'm curious about: does the app by chance pull data from GotSoccer for team ID's, so as to "notice" when a team name changes?

I've noticed that, for my club in specific, the team names get updated per season (to align with the leagues/flights the teams are competing in), but this doesn't seem to be tracked in Soccer Rankings. So at least in our case, someone else looking for team rankings for our teams would potentially be misled, as the rankings don't currently align with the actual teams (ie: the teams are "inheriting" ratings from other teams which previously had the same name).

It seems like this would be possible to do; curious if that's something the app does, or the developer has considered.

 

[RandomSoccerFan]

Yes. the app pulls GotSoccer/GotSport team IDs (or more specifically, the crawler, rather than the mobile app). It used to rely more on the GotSoccer IDs, referring to them as a pretty good source of truth. However - that data has gotten worse to the point that relying on it was causing more matching problems than would be expected, as team's GotSoccer IDs are often just as hosed when teams don't keep that information correct and have various team ID's there that may or may not be the same team. GotSoccer's policy of charging $25 to fix any issue (that may or may not have been caused by their crappy UI), doesn't help their data stay particularly clean. So from what I remember, Mark has shared that they rely on that data field quite a bit less than they used to.

 

[maxpower]

Question I'm curious about: does the app by chance pull data from GotSoccer for team ID's, so as to "notice" when a team name changes?

I've noticed that, for my club in specific, the team names get updated per season (to align with the leagues/flights the teams are competing in), but this doesn't seem to be tracked in Soccer Rankings. So at least in our case, someone else looking for team rankings for our teams would potentially be misled, as the rankings don't currently align with the actual teams (ie: the teams are "inheriting" ratings from other teams which previously had the same name).

It seems like this would be possible to do; curious if that's something the app does, or the developer has considered.

When you look in the "Sources" tab on your teams page it doesn't have the previous iterations of the specific team? For my boy's team that is only 1.5 years old there are 10+ different names across the different tournaments and leagues, and the dates that those names were used for.

 

[RandomSoccerFan]

I've noticed that, for my club in specific, the team names get updated per season (to align with the leagues/flights the teams are competing in), but this doesn't seem to be tracked in Soccer Rankings. So at least in our case, someone else looking for team rankings for our teams would potentially be misled, as the rankings don't currently align with the actual teams (ie: the teams are "inheriting" ratings from other teams which previously had the same name).

You (or anyone else with knowledge of which team is actually which), can try to add correct data sources, and delete incorrect data sources, so the team name that holds the team data that accurately represents the games that they have played. But it's ultimately a losing battle as long as the club is reusing identical names for what are actually different teams over time. They will almost certainly be brought together automatically, and incorrectly, at some point in the future and will have to be resolved multiple times per season.

Your club's method of naming teams has some foreseeable and negative consequences, but perhaps it's completely unintentional. A friendly chat with whoever is registering these teams and explaining what you are seeing might educate their choices. But for any number of reasons, it's better to name a team:

2012B Purple People Eaters Elite

and keep using that team name as the boys progress from U10, U12, U14, whatever.

Rather than:

U12 Purple People Eaters Silver Division

and have to change it next season to:

U13 Purple People Eaters Gold Division

even if it is essentially the same team.

 

[Code]

Soccer ranking app currently has Pateadores b2014 pre ecnl at 4th in the state and 10th in the nation. Major app glitch. I would like to know what went wrong there. We need to start questioning who is behind this app and if we can trust it.

If you look at the list of Unranked Teams and scroll down, you can only get into the letter "c" before the list reaches its limit. Which means hardly any of the teams in that age group have had their game sources assigned. 2014 won't be accurate until some people who know the teams get in there and clean it up.

 

[Code]

I agree with everything you say. It’s the best app out there for this purpose. I am just bothered by its placement of a team whose record doesn’t justify it. If that team had beaten TFA pro once then I can see it. But that team has 6 games under their belt against bad teams yet they got placed 4th in California. Still would like an explanation for this. I don’t see how updating to a new age group can be used to explain this glitch.

Get in there and clean up the age group. The accuracy of predictions will improve significantly.

 

[Code]

Question I'm curious about: does the app by chance pull data from GotSoccer for team ID's, so as to "notice" when a team name changes?

I've noticed that, for my club in specific, the team names get updated per season (to align with the leagues/flights the teams are competing in), but this doesn't seem to be tracked in Soccer Rankings. So at least in our case, someone else looking for team rankings for our teams would potentially be misled, as the rankings don't currently align with the actual teams (ie: the teams are "inheriting" ratings from other teams which previously had the same name).

It seems like this would be possible to do; curious if that's something the app does, or the developer has considered.

Sources of game results are assigned by app users, not the app itself. The app only does auto updates to a team when league scores are updated in a source that has already been assigned team. Sometimes if the team maintains the exact same name from one year to the next, and plays in the same tournament one year to the next, it will automatically update.

 

[RandomSoccerFan]

I've seen that it will automatically update much more often than that, and one of the pieces of data it tries to match is that gotsport team ID. For instance, if a club and team name is updated properly in Gotsport, it is almost immediately updated correctly in SR, even if the team/club name is completely different. Mark might share stats if someone was curious, about how many data sources came in and were matched automatically, vs. how many come in and stay unranked until someone touches them. IMO, if too many stayed unranked and required manual intervention - the system wouldn't scale as it does.

 

[sigmabody]

Sources of game results are assigned by app users, not the app itself. The app only does auto updates to a team when league scores are updated in a source that has already been assigned team. Sometimes if the team maintains the exact same name from one year to the next, and plays in the same tournament one year to the next, it will automatically update.

Yeah... I think this is going to "break" for our club, since they are reusing team names for different teams across seasons, and the teams are playing in the same tournament(s). It'll obviously correct itself over time, but in the short term, it'll at least be a bit misleading for opponents, relative to GotSoccer ratings at least (which track by ID, and which have been updated by the club).

I have no direct influence over the way the club does team names, but I can suggest that they consider maintaining one name which is not tied to current league/rank, to help tracking apps.

I'd also be happy to help update the data in this instance, but I don't know how. I am but a grateful normal parent user of the app, providing pseudo scouting reports for other parents.

 

[Slobodan]

Sometimes you just need to let things go.

Here's a fun one you'll enjoy...

My youngest started playing competitive and at a recent tournament I brought my chair + sat about 4 feet off the line to watch. Several of the other parents did the same. When the game started a couple of the parents actually stood in front of the other parents that were sitting in chairs while the game was going on to yell at their kids while walking to and down the line.

Sigh...

Maybe that's why you have friends and I don't. I'd have asked those parents if they wanted a flag and were staying with the last defender. When they looked at me stupidly and said, "what are you talking about?", I'd have said, "sit the fuck down... and not in front of me."

 

[Carlsbad7]

Maybe that's why you have friends and I don't. I'd have asked those parents if they wanted a flag and were staying with the last defender. When they looked at me stupidly and said, "what are you talking about?", I'd have said, "sit the fuck down... and not in front of me."

I really really wanted to say something like that. But, they're all new to youth competitive soccer + all younger parents with $$$ + this is their first kid. It will be more fun to watch them get more and more crazy over the years. If they all hate me now we won't get to be "friends".

 

[RandomSoccerFan]

I'd also be happy to help update the data in this instance, but I don't know how. I am but a grateful normal parent user of the app, providing pseudo scouting reports for other parents.

It's trivial if you'd ever want to attempt. (It does require the Pro subscription, for $10/year; it doesn't work if using the free version without the sub). When you're on a team you're troubleshooting, click the Sources tab at the top, and it will list all of the data sources (tournament, leagues, etc.) that are already assigned to that team. You can find out exactly which ones they are by clicking on them, and if any of them still have a live web page, you can even click that button to see the results on the web. If any of them aren't actually this team, just hit the Edit button within the app, then the trash can next to the wrong data source, and hit save. That's it - that data source is no longer connected to that team. Finding data sources that you know about (a tournament the team was in, or a league that started), but isn't connected to the team, is just as easy. Click the Add Missing Sources button at the bottom of the app, search for what you're looking for, and if you find one - just add it to the team and hit save. To do it for your own team takes minutes at most per year. But.... then you notice that your opponent's data isn't complete, so you fix them. And then you notice that there are a bunch of teams in your conference with wrong or incomplete data, so you fix them. And the tournament that you just went to was mostly pulled in OK, but there are a few dozen teams that didn't get the info for some reason. And pretty soon, if you have the mildest of OCD tendencies, you find yourself as yet another of Mark's unpaid volunteers helping keep the data closer to accurate than it would be otherwise.

(similar instructions on the FAQ for the app here)

 

[sigmabody]

It's trivial if you'd ever want to attempt. (It does require the Pro subscription, for $10/year; it doesn't work if using the free version without the sub).

I only have/use the free version for now, but I will keep this in mind; I don't buy many apps, but have been considering this one. Your OCD note is on-point, though: I already do a lot of research online, and I am a software developer professionally, so that might be a deep hole if I started...