College star Soares doesn’t mind coming in cold By Frank Dell'Apa in The Boston Globe http://www.boston.com/sports/soccer...lege_star_soares_doesn't_mind_coming_in_cold/ "Soares, the 2010 Pacific 10 Player of the Year for California, is slated to be in a backup central defender role during training camp. But he will have a chance to challenge for a starting spot when the Revolution open their preseason schedule at the University of Central Florida March 17th." I thought the pre-season started February 17th. Maybe that explains why we don't have the other 2 international signings. Pre-season is over a month away!!! Seriously, I appreciate all of the coverage that The Boston Globe has been giving to the Rev's.
The other teams are all playing games while were still screwing around in the bubble, looking for players .... AGAIN. Maybe its time to change this game plan since we kinda SUCK and this tried and tested method has never prepared us for anything
Yup. This Team never started the season very strong and had us all talking in May and June about winning the Triple.
Let's be honest, the only time we've started off very well overall was in '05, and that had a large chunk due to the fact that 3 of our players had been in the USMNT camp.
Does anyone have the time to see if this is actually true -- I don't right now, but I'll try to look at it later tonight if no one else does before then. Qestions are: in SN's tenure, have we actually have a worse record at the start of the season than we've had over the course of that same year for the full season. And do we actually start playing games later, or player few exhibition games, than other teams that start out better. Are there any data to support RevsRule's complaint? Any anecdotes?
We do have a tough start in general because we usually wind up playing more away games to start off then home games, which doesn't help.
How would you even suggest one answer this? Anyway, the first and hardest part, would be finding the data and getting it into a spreadsheet. I got mine here but it is poisoned by USOC, CCL, and Superliga and I don't really want to take the time to filter it all out. After I pasted all of that into a spreadsheet, I tried to calculate a running 5 game average of points-per-game. Then I plotted it across the season. From 2002 to 2010, we were at 1.6 points per game OR LESS every year but 2005, as patfan noted, when we were 2.6. A bunch of people here would say we should be happy with 1.6. Of course, we often hit valleys much lower. If you look at the graphs for every year, you will see peaks and valleys but you will see the maximum 5 game average much later in just about every year. Year Approx Game where peak occurs 2002 28 2003 29 2004 20 2005 8 2006 16 or 28 2007 8 2008 12 2009 21 2010 18 The BIG CAVEAT being that the data is corrupted, again, by CCL, USOC, and Superliga potentially. I tried very quickly to remove the obvious. So, someone with more time will come flame me for inaccuracies but I think the first 5 games of every season are good and I further think the peak comes later than that. PM an email address if you want the file. It was Openoffice. I can save it to xls format but won't guarantee anything.
For some reason I am unable to either rep you, RKane, or to quote your post . . . I do appreciate your efforts on this. You took a more sophisticated approach than I was thinking of. My more simple-minded analysis, which I just did, consisted of looking at the first five games of the season and comparing the ppg for the opening five to the ppg over the whole season. This doesn't show where they may have peaked or valley-ed, but it should answer the question of whether they start out weak compared to how the perform over the whole season. Going back 8 years, to 2003, they have averaged 1.525 ppg over the first five games of the season. In MLS games over those whole eight seasons, they have averaged 1.422 ppg. So they have actually done a little better in the beginning of the season than over the rest of the season, which makes it hard to argue that the pre-season plan is bad. Looking closer, it looks pretty clear that the Revs do not have a problem with starting slow. Ppg in the season-opening five games range from 1.0 to 1.6 ppg, except for the glorious year (2005) when they collected 13 points in the first 5 games (t w w w w for 2.6 ppg) -- of course that was also the high mark for ppg over the whole season, at 1.84 ppg. Interestingly, the two worst starts (1 ppg in 2004 and 1.2 ppg in 2010) came in the two worst seasons (1.1 and 1.06). The other five seasons the starts ranged from 1.4 to 1.6 ppg and the seasons ranged from 1.4 to 1.56. In short, the ppg show pretty clearly that the start of the season is a very good indication of how the season will go, and we are a slow starting team only in the years where we aren't a very good team. Q.E.D.
this seems to have slipped through the mls schedule is late hoopla... jeff_lemieux Jeff Lemieux Shalrie Joseph (left ankle sprain) and Kheli Dube (doctor's appointment) both missed this morning's fitness session. #Revs Shalrie tweeted about it too but i haven't felt like looking for it...
I find your approach interesting and have no reason to disagree. That objective response given, I don't think I'm going to be able to change my emotional reaction that what I see in the spring looks quite crappy, regardless of the results. And that crappy look they have is compared to their opposition - IOTW not just the same rust everyone will have at the beginning of the season. Back to your analysis though. It seems to show that results in the first 5 games are a fairly good predictor of full season performance, ppg being the measure. Do you think we can use that again this year? I do but we've got a bunch of new faces and, just as I can't let go of my emotional feeling they suck at the beginning of the season, I predict many people won't want to let go of the belief that we need a half season to see how these new faces adapt to MLS. Last I checked, 17 is greater than 5. Rep right back at you.
I agree it'll take at least half a season to take the measure of this new team. And I would be very hesitant to put too much stock in the results of the first five games. I am frankly shocked that the data show that the first five games are such an accurate predictor of the whole season. I see what the numbers say, but I find it hard to believe myself. Maybe somebody who understand statistics better then me (you?) can look at what I found and determine whether it has statistical significance?
You're absolutely right, but it's a lot more important to finish the season strong. We've had a few seasons where we've really burned out down the stretch and that's convinced Nicol to take it fairly easy in the pre-season, ramping up gradually. I think that what a lot of people miss, is that when you have preseason games, they do interfere with the work that gets done in a regular practice session (or two, if you're doing double-sessions as we are now). With so many changes, I think the practice time is probably more valuable right now than scrimmages. When you're scrimmaging, 11 guys are playing, but the rest are sitting and watching. Shalrie's tweet: "ShalrieJoseph21 Shalrie Joseph Just getting treatment on my sprain ankle won't be able to play in the second session this afternoon." Doesn't sound like he's too concerned about it.
Actually, I think the close match of first five ppg and season ppg is a bit of a fluke. from my brief glance through the records the sample ppg shifts away from the season avg (sometimes significantly so) if you look at game four or game six instead of game five. Sometime it's up and sometimes it's down, so I don't believe this has any bearing on the question of fast or slow start.
That 1.525 is skewed heavily with the 13 out of 15 we took in 2005. To be fair, if you take out that as well as the lowest point total (5 in '04), you're left with an average that winds up at 1.433 over the first five games. So basically it's saying the team goes 2-2-1 on a normal basis.
Right. And the 1.433 very closely matches the whole-season average over the eight seasons (1.422) -- which I think is a surprisingly strong indication that performance over the start of the season (2-2-1 or 1.4 ppg) is virtually the same as the performance over the course of the whole season (1.422 ppg) -- meaning that the Revs have not had a problem with slow starts.
Well I doubt that very much but I don't have time to look at it now -- I can do the math tonight but I'll be very surprised if looking at 4 or 6 games changes the results from looking at 5. I think the finding is pretty robust that the start of the season is virtually identical with performance over the course of the whole season. But again, I'd be happy if someone who knows statistics can check this for statistical significance. A statistician will know how to check to see if the seemingly strong correlation I found is actually a "fluke", as your suggest.
I wish I had the time to help you out because this has been my career for the past ~9 years, but today is a bad day. As a general rule, the significance of the correlation is dictated by the sample size and there have not been nearly enough seasons to validate the strength of correlation. Correlation also is not a good indicator to use on its own. The number of people who die in a riot is highly correlated with the number of police present, but that's because the riot probably caused the police to be there in the first place. You likely wouldn't say that the presence of police is what caused people to die even though they are highly positively correlated. Anyway, I just wanted to point out that there are way too many factors and not enough examples to make any definitive conclusion about how the Revs approach preseason. I don't agree that they should always take the same approach because I feel a team that has been together for a few years should prepare for the season much different than the team that has entirely new players.
I want to believe that Nicol's reasoning is: "We don't want to tire the players out too much now because it's a long season and we want to save some energy for the playoffs." However, in their 4 MLS Cup Finals the Revs have looked very tired.
This post is great. There will probably never be a coach in place long enough for us to get a significant sample size. We are free to do with our gut! Those damn SN coached REVs alway do too damn little in the pre-season and start the season slow!
Thanks for shedding light on correlation and causation. Those are hard things to keep in mind when following sports teams. On sample size, I used eight seasons that consisted of 40 "early season" games and 244 regular season games. Is that too small a sample to be reliable? I have to say, I was very surprised to see how closely the early season and full season data line up. In the end, if I were SN, I'd take Rkane's joke to heart (there'll never be enough data to trump my personal experience), and I'd go with what I think I know about how to prepare players, how their bodies respond, factoring in the age of the team, etc. I would manage more by experience and intuition than by sabremetrics or soccernomics. But it is fun to try to figure out what the numbers have to say, even in a sport like soccer, which is less measurable than traditional American sports like football, baseball, basketball.
The correlation would be calculated based on average points/game for each 5-game set to start a season compared to each season as a whole, so your "sample size" would just be each season in the sample. I could definitely be missing something obvious in terms of what a useful measure might be or another way to approach this, but off the top of my head I can't think of a data point that would "tell the story." And I agree, I absolutely love statistics and have spent considerable time trying to create soccer-specific data, but the nature of the sport being free-flowing with few events that can be clearly measured makes it extremely difficult.
During major FIFA competitions you can get a treasure trove of game data - heat maps for every player on the field, average positioning point for each player, passing stats (# of passes; # of completed passes; completion % for short, medium, long passes plus crosses as well), distance covered, balls won, balls lost, time of possession. I suspect MLS would see a sizable uptick in fan interest if it provided these sorts of numbers. It would give existing fans a lot more to chew on and give new fans a concrete way of making sense of the game.