This is interesting... 1⃣Don't overestimate predictive power of @MLS table only 3 months into the season.👇Shift months in interactivehttps://t.co/KE5R7YxzbT pic.twitter.com/H8s75bWxUR— Steve Fenn //\\ Mast: StatHunting@skrimmage.com (@StatHunting) June 1, 2017 https://public.tableau.com/views/ML...e?:embed=y&:display_count=yes&:showVizHome=no
Yeah, a lot of stuff in there that I have mentioned in the past. Basketball is less random because there are so many "chances", etc. Baseball playoffs are ridiculously random due to the small sample size and the randomness built into the sport (a weakly hit ball sees its way to the outfield, a ball crushed on the nose goes straight to the centerfielder, a team has 9 hits all distributed in one inning and scores 5 runs, the other has 9 hits, one per inning, and scores 0 runs). If he had done MLS and not Premier League, I think it would have been even more random than NHL, because there are even fewer chances.
LOL. Hockey results are more random than basketball? The NHL's reigning champion is tied at 2 games to 2 in the finals. The NBA's reigning champion is down 0-3 games. Both reigning champions managed to make it back to the finals, however -- because it's not random. The good teams play for the title and the bad teams don't. All that randomness in soccer, by the way, is why the Quakes have failed to qualify for the playoffs the last four years and likely won't again. In the NHL, Montreal won five straight championships in the late 1950s, followed by four straight in the 1970s, with a few championships sprinkled in during the interim. Montreal was succeeded by four straight championships by the Islanders, followed by a long reign of random domination by Wayne Gretzky's Edmonton Oilers -- the guy known as the "Lucky One." LOL. Statisticians obviously don't understand sports (and Vox doesn't tell you how many years of data they're using), and they certainly don't understand hockey (the NHL is a century old), where the most skilled players with the most will to win usually prevail. Ask Patrick Kane or Sidney Crosby. Or the Lucky One. Or ask their enforcers, the guys with the fewest teeth. Maybe we can come up with a formula about how the number of fake molars is related to winning percentage. LOL.
I just skimmed the video but I don't think they are talking primarily about championships, which are inherently skewed by small sample size. It is more about regular season results. In the NBA the best team can go, what was it, 73-9, whereas in MLB a great season would be 100-62. Why? Because there is a higher weight on randomness vs. skill in MLB vs. NBA to determine the result of any one game. As they pointed out, that is primarily due to the number of scoring chances. The more scoring chances, the more that skill can assert itself over chance. I think they did touch on the enhanced degree of randomness in playoffs due to small sample size. The teams that have shown themselves to be best in NBA in the regular season often wind up in the finals whereas in something like football or MLS, due to small sample size + high degree of randomness in the sport you often have lower seeds getting to the championship game. When is the last time a #8 seed in NBA made it to the finals? Then we have MLB and MLS and NFL where wildcard or lower seeds can go on runs and get to the championship series.
The title of the video is, "why underdogs do better in hockey than basketball," which is somewhat misleading if we're talking about "football or MLS." The NHL and NBA play essentially the same number of regular season games, and feature playoff formats that now mirror each other, four rounds of best-of--seven. Ironically, the Vox video speculated that because basketball players can be on the court all game while hockey players enter in and out via shifts, great hockey players cannot carry their team in the same way great basketball players can, which supposedly makes the results more random and subject to upsets. Except the last two nights transpired in quite the opposite way. On Wednesday night, the NBA's most dominant player of this generation, Lebron James, played nearly the entire game on his home court, and lost, making defense of last year's championship nearly impossible. Tonight, the NHL's most dominant player of this generation, Sidney Crosby, played intermittently on home ice, subbing off periodically because that's what hockey players do, yet contributed to half his team's goals in leading them to a critical rubber-game win, and putting them in a position to repeat their title in the next game. Here's a different theory that explains the results in both the NBA and NHL the last couple nights: in each case, the best team won. Do I get tenure?
I was not that keen on the number of minutes argument, but I don't think counterexamples of single games prove anything. This is about looking at several seasons and the trends over those several seasons. I don't know much about hockey but I can clearly see the enhanced role that randomness plays in say MLS vs. NBA, and I think number of scoring chances is the main reason.
Of course counterexamples of single games don't prove anything, but neither does "looking at several seasons," especially when we don't know how many seasons or which ones. Rick Barry's NBA is not Steph Curry's NBA. Yet, both those guys are fairly atypical alpha-dogs for basketball champions, by the way. And the Warriors were underdogs when Barry won. Likewise in the NHL, the Broad Street Bullies, who were champs when Rick Barry was sinking free throws underhanded, played an entirely different brand of hockey (satirized in Slap Shot) than do any of today's teams, because the rules regime is very different.
Sure things change over time and there can be significant variances in different leagues of the same sport. I was initially confused by soccer's relatively low randomness ranking until I realized that rankIng was for EPL, where the skill differential between top and bottom teams is so great that it overcomes the higher randomness factor that is inherent in the sport due to the low number of scoring chances.
Would it kill these people to do actual write ups of their findings? Maybe there is one, though I couldn't locate it via the tweet. It's undermining what possible impact these efforts would have because they can't be bothered to take a few words to actually explain what they are doing. It's really frustrating to look at plot graphs with literally no context. I get that not everyone can write effectively, but it comes across as incredibly lazy to me to just throw a chart up and go: this is valuable.
A highlight of our trip to Azteca was a brief session with USSoccer High Performance Director Jim Bunce. He hired in from the EPL around Feb-2017 and talked about what USSoccer is doing to improve youth development. A large part is to avoid but also recover from injury, since injury is unavoidable. So they are outfitting not just the MNT but also youth players with wearables to monitor physiology as well as performance (speed, distance, etc). They are also doing biobanding so that kids can be evaluated on their true bio age rather than their calendar age. At the EPL, he developed a database system used by all clubs, which pinpointed hamstring and ankle injuries as incurring the greatest number of playing days lost by youth players. He attributes the hamstring problems to the fact that most soccer players do not activate their glutes to play. They are trying to get youth coaches to develop glutes early. He said that each EPL club spends at least $20M on its youth teams annually. Here's a NYTimes article from when the guy was hired: https://nyti.ms/2kAxKDo
So, I've scraped statistics for about 11000 matches to prove that goals from corners are useless rarity.
The analyst thought the conversation would stop there, but to his surprise, it went one step further. Olsen brought the expected goals numbers to a meeting with his players. "This is where we are expected to be; this is where we are. Let's not worry about it," Mairs remembers Olsen telling the team. I don't know if "Let's not worry about it" is really the right attitude to send to players, regardless of what the stats say.... He cites Patrick Mullins as a good example of where he feels added value. Last season, D.C. needed a forward. But there was a problem: "Top tier of strikers in MLS and around the world, they cost money," Mairs says. "We haven't got any money. We're looking at guys who fit a profile: peak age (24-28) so we can develop them. Salary-cap friendly. [Mullins'] expected goals and ability to hit the target were big. That was something that I pushed." D.C. pulled the trigger on a trade with New York City FC, and the new arrival proceeded to score eight goals in 14 games for the club. Boy that Patrick Mullins vs JJ Koval debate from 2014 is not looking so great... how are Koval's grades in dental school?
I realize this probably belongs in the bone broth thread....... http://www.charlotteobserver.com/news/article163167508.html Looks like the use of Sparta is paying off big time for the Quakes.
Good article! But I can't believe the Quakes' soft tissue injuries are down by 80% -- seems like we have even more absenteeism than usual.
That's a pretty good write-up. Of course, the downside of keeping the data collected and analyzed from other teams is that the general public will also have no access to it. It will be fun to try to reverse engineer it all, though. Hopefully people in the community take that as a challenge unto themselves. Assuming there are dividends being paid off by this partnership, of course. I imagine it will be quite a task and I am a bit curious what would happen if this partnership just collapsed after a couple years. It can take a lot of time to bring in academy players, so spotting trends might take some real effort.
This graph is interesting. Does anyone keep metrics like this for the Quakes? Everton's top and bottom performers after 4 matches. Whatever the problem is, it doesn't look like its Rooney. pic.twitter.com/XZMOIemO46— Twelve football (@twelve_football) September 9, 2017
Non-penalty goals plus assists per 90 minutes (min. 1,000 minutes). pic.twitter.com/M4pkDFEsNh— Ben Baer (@BenBaer89) September 28, 2017
Vako doesn't have 1000 minutes yet, but right now, with 606 minutes, he would be 5th on the list, with 4g, 2a in 606 minutes. So where are all those folks telling me Vako was a bust?
Certainly not me, I've thought Vako looked good pretty much from the moment he arrived. The problem is that he's only played 606 minutes.
Yeah, I didn't think it was you. I won't call out names but there are plenty in the Vako thread. I was thinking he is a liability on defense - he takes plays off, walks around a lot. But somehow he had like 15 recoveries, 2 interceptions, and a couple of tackles. That is a lot of defensive plays. I don't know if I've ever seen 15 recoveries. I guess the best way to spin it is that he's "economical" with his energy.