5 answers
Reid
Student
South Hill, Virginia
1
Question
Asked
1658 views
How do I become a sports statistician?
My dream is to combine sports and math and this seems to be the occupation I feel I would enjoy most. I want to ensure that I will take the proper steps to get there.
#sports #math #sportsstatistician #statistician
Login to comment
5 answers
Saman Djabbari
Sr Analyst, Sales & Operations Planning
13
Answers
Irvine, California
Updated
Saman’s Answer
There are a lot of good websites to get you started. Fangraphs.com is a great stats heavy baseball site. Footballoutsiders.com is great for both college and pro football but leans more towards pro football. Footballstudyhall.com is a great stats based College Football site. I don't know if you are on Twitter, but if you are I can recommend multiple Twitter accounts. Many of the math/stats based jobs require some technical skills like python or database management. I'd look into something like https://www.edx.org/, they often have courses where you can learn statistical uses in sports.
Kenton Wong
Manager, Stats & Info at ESPN
3
Answers
Los Angeles, California
Updated
Kenton’s Answer
Read about as many advanced statistics as you can and try your best to figure out what stories they are there to tell. Find an internship in sports in any field to get your foot in the door. Collegiate sports information offices are a great start.
Robert Chung
Demographer at UC Berkeley and UCSF
1
Answer
Updated
Robert’s Answer
This is a wonderful question
TL;DR summary: get a degree in econ, poli sci, or sociology with a minor in computational stat.
Long version: In my day job I do other kinds of statistical analysis but I've also become a sports statistician in the sport of cycling, with peripheral interests in basketball. In the last decade or so, statistical analysis of sports has grown tremendously, spurred by the success of individuals like Bill James, Dean Oliver, John Hollinger, and Nate Silver, among others. You are probably familiar with the work of Bill James in baseball, and the book and film "Moneyball."
The characteristic common to all of these individuals is that they were fans first who subsequently used statistics both to understand and to predict patterns within the sport. From being fans, they had developed a sharp eye to develop their intuition about what patterns were important, and from there they used statistical modeling to organize, understand, and then to test their understanding through prediction. Then they revised their models if their predictions failed to perform well.
This is the basis for all statistical analysis: observation, data collection, analysis, model construction, prediction, testing, and revision.
As one famous statistician, Samuel Karlin, has said, "the purpose of [statistical] models is not to fit the data but to sharpen the questions."
Accordingly, one path to becoming a sports statistician is to develop your observational skills about the sports you're most interested in, then to collect data (or to use data that have already been collected), then analyze the data and develop statistical models from them, then to test your predictions.
As it turns out, this is the path taken by many modern data-oriented academic programs. In particular, since the level of data collected by dedicated systems such as the SportVU system used in the NBA isn't available to students or the public, in these program there is often an emphasis on working with large data sets that have been scraped off the web using something like Python.
Although traditional statistical inference is used, I would say that more modern statistical modeling is probably used more often, especially computational statistics. These are the kinds of statistical approaches taught in the social sciences more than the physical or biological sciences. The difference is, in the physical and biological sciences, much of the statistical methods taught deal with controlled experiments, while in the social sciences more of the statistical methods deal with "natural experiments" that are observational rather than controlled. In sports competitions we rarely have controlled experiments and instead must work with observational data that we have to be controlled statistically.
If you look at the people associated with 538.com, you'll see that many were originally trained as economists or political scientists or sociologists, and in the statistical approaches in those fields. I myself trained as an economist and a demographer, specializing in mathematical and statistical modeling.
You may already have read "Moneyball" and are already familiar with the Bill James Abstracts. If not, I would recommend seeking them out, as well as Dean Oliver's "Basketball on Paper" and the APBRmetrics forum, to see the kinds of statistical modeling that have been used in baseball and basketball. I would skim through these even if your particular sport isn't baseball or basketball -- you would still get a good idea of the kinds of approaches people in the field talk about. In both cases, deep understanding of the game was the foundation; the statistical modeling was built on top of that.
TL;DR summary: get a degree in econ, poli sci, or sociology with a minor in computational stat.
Long version: In my day job I do other kinds of statistical analysis but I've also become a sports statistician in the sport of cycling, with peripheral interests in basketball. In the last decade or so, statistical analysis of sports has grown tremendously, spurred by the success of individuals like Bill James, Dean Oliver, John Hollinger, and Nate Silver, among others. You are probably familiar with the work of Bill James in baseball, and the book and film "Moneyball."
The characteristic common to all of these individuals is that they were fans first who subsequently used statistics both to understand and to predict patterns within the sport. From being fans, they had developed a sharp eye to develop their intuition about what patterns were important, and from there they used statistical modeling to organize, understand, and then to test their understanding through prediction. Then they revised their models if their predictions failed to perform well.
This is the basis for all statistical analysis: observation, data collection, analysis, model construction, prediction, testing, and revision.
As one famous statistician, Samuel Karlin, has said, "the purpose of [statistical] models is not to fit the data but to sharpen the questions."
Accordingly, one path to becoming a sports statistician is to develop your observational skills about the sports you're most interested in, then to collect data (or to use data that have already been collected), then analyze the data and develop statistical models from them, then to test your predictions.
As it turns out, this is the path taken by many modern data-oriented academic programs. In particular, since the level of data collected by dedicated systems such as the SportVU system used in the NBA isn't available to students or the public, in these program there is often an emphasis on working with large data sets that have been scraped off the web using something like Python.
Although traditional statistical inference is used, I would say that more modern statistical modeling is probably used more often, especially computational statistics. These are the kinds of statistical approaches taught in the social sciences more than the physical or biological sciences. The difference is, in the physical and biological sciences, much of the statistical methods taught deal with controlled experiments, while in the social sciences more of the statistical methods deal with "natural experiments" that are observational rather than controlled. In sports competitions we rarely have controlled experiments and instead must work with observational data that we have to be controlled statistically.
If you look at the people associated with 538.com, you'll see that many were originally trained as economists or political scientists or sociologists, and in the statistical approaches in those fields. I myself trained as an economist and a demographer, specializing in mathematical and statistical modeling.
You may already have read "Moneyball" and are already familiar with the Bill James Abstracts. If not, I would recommend seeking them out, as well as Dean Oliver's "Basketball on Paper" and the APBRmetrics forum, to see the kinds of statistical modeling that have been used in baseball and basketball. I would skim through these even if your particular sport isn't baseball or basketball -- you would still get a good idea of the kinds of approaches people in the field talk about. In both cases, deep understanding of the game was the foundation; the statistical modeling was built on top of that.
Kenton Wong
Manager, Stats & Info at ESPN
3
Answers
Los Angeles, California
Updated
Kenton’s Answer
Read about as many advanced statistics as you can and try your best to figure out what stories they are there to tell. Find an internship in sports in any field to get your foot in the door. Collegiate sports information offices are a great start.
Alan Hull
Area Scout at Tampa Bay Rays
3
Answers
California, California
Updated
Alan’s Answer
Be very good at math. If you're a B student in high school, this probably is an unrealistic professsional pursuit. Profession teams want to hire the best candidates out there.
If you are a math, statistics or computer science major/master/professional, be up to date on all the research that's out there. Then, come up with your own research. Come up with ideas, find ways to analyze them and publish them. Many people get hired every year because they found something that can help a profession team gain an edge.
Otherwise, start as an intern and work your way up. Learn computer programming or programs like SQL to help teams crunch data.
If you are a math, statistics or computer science major/master/professional, be up to date on all the research that's out there. Then, come up with your own research. Come up with ideas, find ways to analyze them and publish them. Many people get hired every year because they found something that can help a profession team gain an edge.
Otherwise, start as an intern and work your way up. Learn computer programming or programs like SQL to help teams crunch data.