I have been reading “The House Advantage” by Jeff Ma. The book explains the importance of incorporating data into the decision process. Ma uses examples from his background in sports and as a member of the MIT blackjack team; he was the central character in the best-selling book “Bringing Down the House”.
In one chapter, Ma writes about the approach he takes when consulting with businesses, whether sports-related or not. He summarizes it without using any complex statistical terms, but as taking information from the past to make decisions about the future. Whether in sports or business, that process creates an edge.
This applies to most sports decisions, whether for free agency or the draft. What trends are in a player’s past that bode well for their future? That track record may be long and consistent with plenty of data, as with Albert Pujols or Tim Duncan. Or it could be brief, such as the case of a young lefty relief specialist or a college freshman entering the NBA Draft. Generally speaking, the more the data, the easier it is to project performance going forward.
The quality of the data is also important. When projecting Ubaldo Jimenez’s performance for the season’s final two months, his impressive won-lost record through July is not all that helpful. His ERA is more valuable, but still not the best predictor. The best projections come from examining Jimenez’s walks and home runs allowed, the percentage of batters he has struck out, and factors like left on base percentage and batting average on balls in play. Performance in previous seasons matters too, as that is all part of his track record.
Tuesday, July 27, 2010
Tuesday, June 15, 2010
Digging Deeper
This time of year, you need to look beyond the core statistics to determine true performance. Last June, I tagged Jorge de la Rosa as a pitcher primed for a turnaround. At the time, he was 2-7 with a 5.81 ERA. He finished 16-9 with a 4.38 ERA. I also predicted two arbitration eligible starters would head in opposite directions.
This year, Randy Wells is a candidate for a huge turnaround. His mainstream stats don’t look good (3-5, 5.15 ERA). But a pitcher’s actual performance is better evaluated with advanced metrics, especially in timeframes of less than half a season.
Wells has struck out batters more often than he did last season, when he went 12-10 with a 3.05 ERA. He has also given up home runs and walks less often than in 2009. The problem has been his high batting average on balls in play. His .359 BABIP is 65 points higher than last season. While he has allowed more line drives, this also shows that he has experienced some bad luck and/or poor defense behind him. A low left on base percentage demonstrates that Wells’ hits allowed have been poorly timed. Neither trend should hold up for the entire season, so expect his ERA to improve.
The disconnect between core statistics and advanced metrics isn’t always this great, but it always exists to some extent. Knowing this can not only build value, but it helps predict future performance as well.
This year, Randy Wells is a candidate for a huge turnaround. His mainstream stats don’t look good (3-5, 5.15 ERA). But a pitcher’s actual performance is better evaluated with advanced metrics, especially in timeframes of less than half a season.
Wells has struck out batters more often than he did last season, when he went 12-10 with a 3.05 ERA. He has also given up home runs and walks less often than in 2009. The problem has been his high batting average on balls in play. His .359 BABIP is 65 points higher than last season. While he has allowed more line drives, this also shows that he has experienced some bad luck and/or poor defense behind him. A low left on base percentage demonstrates that Wells’ hits allowed have been poorly timed. Neither trend should hold up for the entire season, so expect his ERA to improve.
The disconnect between core statistics and advanced metrics isn’t always this great, but it always exists to some extent. Knowing this can not only build value, but it helps predict future performance as well.
Monday, June 7, 2010
Stepping Up
Caron Butler's team only lasted one round in the NBA Playoffs, but he did his part at the offensive end. Butler increased his points per 48 minutes figure by 7.2 - from 20.8 in the regular season to 28.0 in the postseason - to top all NBA players (through June 7, minimum: 150 playoff minutes).
Paul Millsap came next at 6.6 more points per 48 minutes, followed by Luc Richard Mbah a Moute (+5.6), Jameer Nelson (+5.5), Russell Westbrook (+5.3), Deron Williams (+5.0), Jason Richardson (+4.5), Goran Dragic (+3.6), Derrick Rose (+3.3), and Dwyane Wade (+2.8).
Among players who have drawn the media spotlight this postseason, Kobe Bryant (+2.1) had the 12th-greatest increase and Rajon Rondo (1.4) placed 16th. Ray Allen has actually scored slightly less per 48 minutes (-0.3). LeBron James saw a drop of 3.4 points, but had still posted the playoffs' fourth-highest points per 48 minutes figure (33.4).
Surprisingly, neither NBA Finals participant had a player in the top 10. The Lakers had two more players - Derek Fisher and Pau Gasol - join Bryant in the top 20. Rondo was the sole member of the Celtics in the top 20.
One key point to make is that most players see their scoring rate dip in the playoffs, mainly because the pace slows. Among the 94 players to see at least 150 playoff minutes, only 32 increased their points per 48 figures.
After the NBA Finals, we will examine points per possession, which adjusts for differences in game pace. We will also analyze advanced metrics to demonstrate who stepped up their all-around game in the playoffs.
Paul Millsap came next at 6.6 more points per 48 minutes, followed by Luc Richard Mbah a Moute (+5.6), Jameer Nelson (+5.5), Russell Westbrook (+5.3), Deron Williams (+5.0), Jason Richardson (+4.5), Goran Dragic (+3.6), Derrick Rose (+3.3), and Dwyane Wade (+2.8).
Among players who have drawn the media spotlight this postseason, Kobe Bryant (+2.1) had the 12th-greatest increase and Rajon Rondo (1.4) placed 16th. Ray Allen has actually scored slightly less per 48 minutes (-0.3). LeBron James saw a drop of 3.4 points, but had still posted the playoffs' fourth-highest points per 48 minutes figure (33.4).
Surprisingly, neither NBA Finals participant had a player in the top 10. The Lakers had two more players - Derek Fisher and Pau Gasol - join Bryant in the top 20. Rondo was the sole member of the Celtics in the top 20.
One key point to make is that most players see their scoring rate dip in the playoffs, mainly because the pace slows. Among the 94 players to see at least 150 playoff minutes, only 32 increased their points per 48 figures.
After the NBA Finals, we will examine points per possession, which adjusts for differences in game pace. We will also analyze advanced metrics to demonstrate who stepped up their all-around game in the playoffs.
Wednesday, June 2, 2010
Cox and Torre Have Lived Parallel Lives
Along the lines of the Abraham Lincoln and John F. Kennedy list of bizarre coincidences, Joe Torre and Bobby Cox have plenty of their own. This has nothing to do with sports analytics, but had to post it anyway:
Joe Torre has three letters in his first name and five in his second name.
Bobby Cox has five letters in his first name and three in his second name.
Torre played for the Braves and managed the Yankees.
Cox played for the Yankees and manages the Braves.
Torre originally signed with the Braves and manages the Dodgers.
Cox originally signed with the Dodgers and manages the Braves.
Both Cox and Torre played third base in the Major Leagues.
Both played for teams in New York City: Torre with the Mets, Cox with the Yankees.
Both managers had their first full season in 1978.
Torre finished last in the NL East.
Cox finished last in the NL West.
Both managers had their second full season in 1979.
Torre finished last in the NL East.
Cox finished last in the NL West.
Torre managed in New York (with the Mets), left to manage other clubs, and returned to New York (with the Yankees).
Cox managed in Atlanta, left to manage another club, and returned to Atlanta.
Torre replaced Cox as Braves manager before the 1982 season.
Both skippers left managing in the mid-1980s to pursue other opportunities: Torre as a broadcaster and Cox as a GM.
Torre had two managerial stints in New York. He struggled in the first one and became one of the all-time greatest managers in the second.
Cox had two managerial stints in Atlanta. He struggled in the first one and became one of the all-time greatest managers in the second.
Both managers have made 15 postseason appearances, more than any other manager in Major League history.
Both managers have had streaks of 14 consecutive playoff appearances, more than any other manager in Major League history.
Joe Torre has three letters in his first name and five in his second name.
Bobby Cox has five letters in his first name and three in his second name.
Torre played for the Braves and managed the Yankees.
Cox played for the Yankees and manages the Braves.
Torre originally signed with the Braves and manages the Dodgers.
Cox originally signed with the Dodgers and manages the Braves.
Both Cox and Torre played third base in the Major Leagues.
Both played for teams in New York City: Torre with the Mets, Cox with the Yankees.
Both managers had their first full season in 1978.
Torre finished last in the NL East.
Cox finished last in the NL West.
Both managers had their second full season in 1979.
Torre finished last in the NL East.
Cox finished last in the NL West.
Torre managed in New York (with the Mets), left to manage other clubs, and returned to New York (with the Yankees).
Cox managed in Atlanta, left to manage another club, and returned to Atlanta.
Torre replaced Cox as Braves manager before the 1982 season.
Both skippers left managing in the mid-1980s to pursue other opportunities: Torre as a broadcaster and Cox as a GM.
Torre had two managerial stints in New York. He struggled in the first one and became one of the all-time greatest managers in the second.
Cox had two managerial stints in Atlanta. He struggled in the first one and became one of the all-time greatest managers in the second.
Both managers have made 15 postseason appearances, more than any other manager in Major League history.
Both managers have had streaks of 14 consecutive playoff appearances, more than any other manager in Major League history.
Thursday, May 20, 2010
Adjusted Double-Doubles
No matter how well a player performs, playing time has a huge impact on per game statistics. Per game metrics remain the favorite of the mainstream sports media, even though playing time varies widely among those considered “regulars.”
A total of 15 NBA players averaged a double-double in 2009-10. The list includes several current and former All-Stars: Tim Duncan, Carlos Boozer, Chris Bosh, Dwight Howard, Pau Gasol, Steve Nash, Deron Williams, Chris Paul, Andrew Bogut, David Lee, Zach Randolph, Gerald Wallace, Troy Murphy, Kevin Love, and Joakim Noah.
Love’s achievement was most impressive considering he only played 28.6 minutes per game. The other 14 averaged 35.1 minutes per contest. Love became just the second active player to average a double-double in less than 30 minutes per game. Lee, the only other, matched his feat in 2006-07. Among the 31 total players to do this in NBA history, Love was the youngest ever.
It’s nearly impossible to average a double-double in less than regular action. Several more players could have reached this level had they seen as much playing time as the first group. So here is a group of additional players that projected as double-double guys based on points, rebounds, and assists per 35 minutes. This adjustment puts them on equal ground with the earlier group, which saw that much game action on average.
Lamar Odom
Brendan Haywood
Emeka Okafor
Udonis Haslem
Samuel Dalembert
Drew Gooden
Shaquille O'Neal
DeJuan Blair
Serge Ibaka
Kris Humphries
Nazr Mohammed
DeAndre Jordan
Louis Amundson
Expect the younger players on this list – Serge Ibaka, DeAndre Jordan, and DeJuan Blair – to emerge as they receive more burn in the upcoming seasons. All they need is additional PT to post big per game numbers.
A total of 15 NBA players averaged a double-double in 2009-10. The list includes several current and former All-Stars: Tim Duncan, Carlos Boozer, Chris Bosh, Dwight Howard, Pau Gasol, Steve Nash, Deron Williams, Chris Paul, Andrew Bogut, David Lee, Zach Randolph, Gerald Wallace, Troy Murphy, Kevin Love, and Joakim Noah.
Love’s achievement was most impressive considering he only played 28.6 minutes per game. The other 14 averaged 35.1 minutes per contest. Love became just the second active player to average a double-double in less than 30 minutes per game. Lee, the only other, matched his feat in 2006-07. Among the 31 total players to do this in NBA history, Love was the youngest ever.
It’s nearly impossible to average a double-double in less than regular action. Several more players could have reached this level had they seen as much playing time as the first group. So here is a group of additional players that projected as double-double guys based on points, rebounds, and assists per 35 minutes. This adjustment puts them on equal ground with the earlier group, which saw that much game action on average.
Lamar Odom
Brendan Haywood
Emeka Okafor
Udonis Haslem
Samuel Dalembert
Drew Gooden
Shaquille O'Neal
DeJuan Blair
Serge Ibaka
Kris Humphries
Nazr Mohammed
DeAndre Jordan
Louis Amundson
Expect the younger players on this list – Serge Ibaka, DeAndre Jordan, and DeJuan Blair – to emerge as they receive more burn in the upcoming seasons. All they need is additional PT to post big per game numbers.
Wednesday, March 10, 2010
MIT Sports Analytics Conference
Here’s a brief recap on last weekend’s MIT Sports Analytics Conference. The conference doubled in size from last year, and again delivered invaluable information for sports agents. This post only covers highlights on basketball topics.
One of the most interesting exchanges dealt with how NBA teams value clutch performance. Rockets GM Daryl Morey said he likes when players have shown strong clutch performance in the past, but he wouldn’t spend millions on a player based on that. Mark Cuban countered by saying he would pay for it, and cited Jason Kidd as one example.
Morey later explained that their research revealed how well Kevin Martin had performed against tough opposing defenses before trading for him at the deadline. Agents may want to emphasize this point for their free agents who excel in this area.
Cuban believes certain NBA teams have an advantage in analytics. Why does he think that? For one, he examines the combinations that some teams place on the court. By comparing that to data that the Mavericks research, he knows which clubs are informed and not, as some of these lineups have poor track records.
Morey also said the Rockets thoroughly research how well a player will perform in their system versus with their current team before acquiring them.
Agents can feel free to contact me for a more detailed rundown on the conference.
One of the most interesting exchanges dealt with how NBA teams value clutch performance. Rockets GM Daryl Morey said he likes when players have shown strong clutch performance in the past, but he wouldn’t spend millions on a player based on that. Mark Cuban countered by saying he would pay for it, and cited Jason Kidd as one example.
Morey later explained that their research revealed how well Kevin Martin had performed against tough opposing defenses before trading for him at the deadline. Agents may want to emphasize this point for their free agents who excel in this area.
Cuban believes certain NBA teams have an advantage in analytics. Why does he think that? For one, he examines the combinations that some teams place on the court. By comparing that to data that the Mavericks research, he knows which clubs are informed and not, as some of these lineups have poor track records.
Morey also said the Rockets thoroughly research how well a player will perform in their system versus with their current team before acquiring them.
Agents can feel free to contact me for a more detailed rundown on the conference.
Tuesday, January 5, 2010
The Power of Park Effects
Park effects have a huge impact on baseball statistics. Yet this area is confusing and often misunderstood. Here are some key points on park effects that agents may find useful for free agency and arbitration.
Beware of Reputations: Citizens Bank Park is thought to be a hitter’s paradise. Yet the numbers don’t back that up. According to the Bill James Handbook, the Phillies home ballpark increased home runs by just 1 percent last year. The impact was far greater from 2007 through 2009 when the park upped homer frequency by 14 percent. But even in this time frame, run production only increased by 3 percent. Now here’s the real shocker: After all the talk about the early season home run barrage in the new Yankee Stadium, the park decreased run production by 4 percent in 2009.
Avoid “One Size Fits All Park Factors”. Ballparks affect different players in different ways. Minute Maid Park is a good park for right-handed home run hitters, but not for left-hand hitters with power. Chase Field, which greatly increases doubles and triples, makes a great fit for gap hitters with speed.
Park Factors Change from Year to Year. Weather patterns and other factors influence park effects. Turner Field increased run scoring by 6 percent in 2008. Last season, when Atlanta had a cooler than usual summer, it decreased scoring by 10 percent.
Don’t Buy the Road Stats Argument. In some cases, teams may point out that a player had comparable numbers in both home and away games to show that his home park did not hurt his statistics. But most players have better numbers at home than on the road, probably due to park familiarity and the negative effect of travel on away stats. Ballparks impact statistics whether or not a player’s home and road numbers look similar.
Beware of Reputations: Citizens Bank Park is thought to be a hitter’s paradise. Yet the numbers don’t back that up. According to the Bill James Handbook, the Phillies home ballpark increased home runs by just 1 percent last year. The impact was far greater from 2007 through 2009 when the park upped homer frequency by 14 percent. But even in this time frame, run production only increased by 3 percent. Now here’s the real shocker: After all the talk about the early season home run barrage in the new Yankee Stadium, the park decreased run production by 4 percent in 2009.
Avoid “One Size Fits All Park Factors”. Ballparks affect different players in different ways. Minute Maid Park is a good park for right-handed home run hitters, but not for left-hand hitters with power. Chase Field, which greatly increases doubles and triples, makes a great fit for gap hitters with speed.
Park Factors Change from Year to Year. Weather patterns and other factors influence park effects. Turner Field increased run scoring by 6 percent in 2008. Last season, when Atlanta had a cooler than usual summer, it decreased scoring by 10 percent.
Don’t Buy the Road Stats Argument. In some cases, teams may point out that a player had comparable numbers in both home and away games to show that his home park did not hurt his statistics. But most players have better numbers at home than on the road, probably due to park familiarity and the negative effect of travel on away stats. Ballparks impact statistics whether or not a player’s home and road numbers look similar.
Subscribe to:
Posts (Atom)
