Player age on Dec 31 (end of year), weighted by the number of minutes played. 1996 -- 28.49 1997 -- 28.67 1998 -- 28.75 1999 -- 28.66 2000 -- 28.73 2001 -- 28.62 2002 -- 28.24 2003 -- 27.92 The age was essentially constant for the first six years of the league. Only in the past two years has the age begun to drop. I think this is evidence that programs like Project-40 and Bradenton are starting to pay dividends. It's also interesting to note that the drop in age occurred after contraction, while the expansion year was the oldest. Perhaps these numbers are skewed by the presence of aging imported players. At some point, I'll re-run them omitting foreigners and see what happens.
Ok, I ran the numbers for US-eligible players only... 1996 -- 28.01 1997 -- 28.11 1998 -- 28.00 1999 -- 27.90 2000 -- 28.01 2001 -- 28.00 2002 -- 27.77 2003 -- 27.61 The pattern looks pretty similar. Expansion in 1998 might have opened a few doors to younger players in 98 and 99, but contraction in 2002 doesn't seem to have frozen out the younger players.
You wouldn't happen to have a standard deviation for each one of those means, would you? I'd be curious to know if there is a statistically significant difference between 2003 and other years.
[Using the data for US-eligible players] If we assume that all variation in years 1-6 is random, then the estimated s.d. is 0.067 and the mean is 28.01. (Note that any real change would inflate this estimate, hence reducing the significance of future changes.) The drop from years 6 to 7 is by 3.6 standard deviations, and the drop from years 7 to 8 is by an additional 2.4 standard deviations. Conclusion: highly significant Fair question, but that's actually non-trivial to compute ... if you want to compare 2002 to 2003, you have to adjust both years for Preki simultaneously.
Does it make a big difference if you use July 1 rather than December 31? I'm working on some studies with age as well, and wondering if the age at the midpoint of the year makes a difference or not. I guess if someone's birthday is July 7, they're not significantly different on June 30 than on July 14, but the number would be different. Just curious. Probably doesn't matter.
I'm pretty sure it doesn't make a difference, but the only data I have is year of birth, so I don't have any way to confirm.
Thanks for crunching those numbers. Im suprised that the numbers have remained so relatively constant. I'm not suprised that the number was down rather signifigantly this year. It'd be interesting to run some numbers too if you gave a min. minutes played floor. It'd be challenging to figure out the average age of 'starters.' But maybe at a predetermined # of games of # of minutes would yeild some interesting stuff.
Don't know how much use this information is, but if you graph simply the age of each of the 230 players who played minutes in MLS this season (age as of July 1), you get a bell curve with the high point at 26: 16...1 17...2 18...4 19...2 20...7 21..17 22..27 23..21 24..19 25..24 26..31 27..13 28..11 29..10 30...7 31...7 32...9 33...7 34...5 35...3 36...1 37...1 38...0 39...0 40...1 That's obviously unweighted towards minutes played, which is a good way that Phil does it. This is strictly showing the number of players in MLS of a particular age. There were 16 players under 21, and only 18 older than 32. BTW, Phil, when I weighted the average age towards minutes played (and I hope I did it the same way you did), I got 26.52 using July 1, which makes sense because the other 6 months basically gives you another "year" on (probably about) half of the players. But I would imagine the relationship between all the years, if you did the average age by using July 1, would show the same pattern as it does using December 31. And I have everyone's birthdate now, if you want that data (if it would help you at all).
Me too, though I think the average might be masking some interesting effects. My belief (which needs to be verified) is that while the average was staying constant, the US player pool was simultaneously getting both older and younger. At the outset, the great majority of US players in MLS were in the age range from 24 (McBride, Armas) to 31 (Doyle, Trittschuh). As this generation got older, we had a gradual increase in the numbers of over-30 players. At the same time, the league also began to add players like Corrales, Olsen, Wolff, etc., who were able to start contributing as college-aged players. Perhaps the story isn't a shift in the average, but rather a widening of the age range? [Kenn -- it doesn't look to me as if our numbers are in agreement. I'll try to go back soon and double-check what I've done.]
Yeah we may be missing something in terms of turnover of players. You may have hit it in terms of core guys who came into the league young and are still here. Kenn's style analysis would be interesting to see for year per year. Either way I think that the decrease of this year is signifigant and is an example of the MLS 'youth movement.'
Interesting stuff guys. Thanks for doing this. I expect the trend downward to continue over the next few years. I would not be suprised to see the average age drop to 25 and stay there for at least a decade after that.
Why? The core group can only get older. The young kids coming in will have to come in in great numbers to greatly affect the average, won't they?
Maybe I should have said "mean." Over the next few years I think you will see the following three trends: --more players will skip college and go pron at 18 and younger --teams will build like the Fire did this year, with a core of 4 or 5 veterans and with lots of cheap young players --fewer players over 33, and certainly fewer international players over 33. 18 players in the league were over 33. I wouldn't be surprised if that number drops to 10 or ( one per franchise). Preki will be the last 40 year old in the league.
No, I know what you meant by "average." I just wasn't sure how, when the average age has remained fairly constant the last eight years, it was going to drop a full year. I would think a lot of U18's would have to turn pro (and play - if you're weighing it by playing time, the Craig Capanos of the world don't factor into it). The Fire's average actual age was 24. Their average age weighed by playing time was 25. And all those guys are going to keep getting older, not younger. You're going to have to continue to get rid of guys when they hit 27 or so and bringing in 22 year olds (and assuming no concurrent drop in team performance) to bring the average down. As in most sports, great players hold their value longer. Stoitchkov is the second-oldest player in the league, and I don't think he'll see 40. Fraser probably won't see 40. Onstad might see 40. But Preki is a freak of nature and conditioning. Not everyone can do what he can do. I'll have to go back and check and see how much the numbers at both ends of the spectrum have changed over time. It would be interesting to see.
I think that what you are going to see a much more Darwinian attrition at 26 or 27, as guys who simply are not good enough for overseas -- and not really stars in MLS -- yet are making too much money since they have 5 or 6 years of professional experience, are shed by clubs in return for younger cheaper players. The only thing that could change this is a rapid increase in the salary cap, which I don't think is in the offing. I believe that MLS is evolving into a young players league, and rightly so. What's the marginal gain from having, say, a 27 year old non-Nat team player, versus a 20 year old speed demon? Not much. What's the marginal cost? It could be $40 grand. That's a significant amount of money with a salary cap under $2 mill. GMs from around the league have taken notice of the Fire's approach -- watch them start replicating it. Good young players are everywhere, many laying fallow. Nat Borchers. Justin Mapp. And we're not talking Bradenton babies. That's why coaches who know how to handle young players (Bradley, Sarachan, Yallop) will survive and those that can't (Hudson, Jeffries) are toast.
I see your point, but effective players are going to keep their value longer. As long as they have value, whatever increased cap space they eat up will be made up by young players (assuming no increase). Chris Armas would be an example, wouldn't he? A guy who's going to command a salary increase, yet is 31? But there will be some natural attrition if guys like Bocanegra (who is young, but not highly paid, so might not be the best example) go up and out, instead of just out (like Roy Lassiter) and not only leave space, but don't significantly bring the average age down. Like I said, I'm going to go back and check and see what trends are evident.
Chris Armas, though, is a "star" in this league, much as that sticks in the craw of some who find all kinds of reasons to diss him. He is the exception, rather than the rule. You can build a team around him, and feel as though you are getting very good value for your max salary. I am thinking more along the lines of guys like, say, Kerry Zavagnin, Ross Paule, a Steve Jolley, Richie Willams, Richie Kotschau, Carey Talley , a Leo Cullen -- a fairly large pool of guys past 25 who are good players yet not SO great that they can't be replaced by younger players, making, say, 1/3rd to 1/2 of what they are making. Younger is cheaper in MLS and the Fire have shown that with a younger team you can compete and win..and win big. You can build around, say, 5 top vets making the max, and then fill in with lots and lots of young players. They give you depth, and quantity, without much loss of quality.
Well, like I said, let's see what happens with those guys, historically. At what age do players leave the league? Do we have a way of knowing if players leave the league who can still be effective (does that mean we have to track them in the A-League or overseas)? Or if they're leaving not because of age or salary, but because they're not good enough? And because there aren't 16 teams, where a veteran player can usually find a place to play because the gap between Team 1 and Team 16 is big enough that Team 16 can delude themselves into thinking that just because a player is a "veteran" that he can help them, even if he can't? And, does what Dave Sarachan did in Chicago (though some of his hand was forced) mean that Colin Clarke will automatically do the same thing? Or that Steve Nicol will? Or whoever's coaching Colorado in three years? As I said, all interesting questions that need some light shined on them.
A codicil to this..... Trying to do the same type study on the A-League is not quite as simple. While their stat-keeping is much improved over several years ago, I'm still not 100% sure of them, and it's hard to track down players' birthdates. Some teams don't have birthdates on the rosters listed on their websites. I've only been able to track down birthdates for a little less than half (225 of 476) of the players who played in the A-League this year. They ranged from 36-year-old Scott Benedetti of Portland to 18-year-old Sita-Taty Matondo of Montreal. The average age of those 225 players was 29.5. 49 guys were 30 or over (as of July 1, 2003). Only 11 were under 21. Again, that's with less than half of the precincts reporting. Adjusting for playing time, the average drops a bit, to 26.7.
Trying to think of the last time I saw the word "codicil" DOH! Ok, it's a good thing Kenn ran the numbers separately. Embarrassingly enough, every single one of my reported ages is one year too big. Sorry bout that, guys. Memo to myself: sleep deprivation is a bad thing.
Little-Known Codicil to the Faber College Charter I found that I couldn't get my spreadsheet to accurately figure the age of someone born after the date in question - everybody whose birthday was July 1 or after was a "year" too old, so I had to go back through and revise those downward to get the player's actual age on July 1. So, anyway, revising your figures would seem to get us to about the 26-27 range, right? Which is in line with the other numbers. I'm seeing a trend here. I've found some more A-League birthdates, but I doubt I've got half the guys in the league yet.
Re: A codicil to this..... Kenn, the player ages are listed on the uslsoccer.com website. Just click on team and then on an individual player; except for a few exceptions (mainly Riverhawks, unsurprisingly), they're all there. Using July 1 2003 as the cutoff, I got radically different numbers: 25.8 average age, minutes adjusted average of 26.3. Way lower than I would have guessed. Scott Benedetti and Sita-Taty were still the oldest and youngest, respectively. The Canadian teams clearly have a whole lot more very young players than the rest of the league - the only team that comes close to competing is El Paso.
Wow. Well done. Thanks. I'll run those numbers, and if you want to exchange data, send it to me at kenn@mc.net. Thanks. Not surprising that your average age would be lower than mine, but it's interesting that the adjusted for minutes figure is almost identical to mine.
I don't know, I wouldn't have expected the average age and minutes-adjusted average age to have been as far apart as you got. If there are all these old players in the league who aren't getting played, what are they doing on rosters? I'm sure they get paid a good deal more than the 22 and 23 year old's just out of college. I'd be very surprised if the A-league, like MLS, ever had an average age older than its minutes adjusted average age - you're a lot more likely to keep a young player not getting playing time on your squad than you are to keep a 35 year old.
Remember, mine was based on less than half of the players in the league. When I double-check mine against yours, we'll see what we get. And I've asked my guy at USL for the birthdates of the missing guys. Thanks.