Rajiv Maheswaran: The math behind basketball's wildest moves
라지브 매헤스워렌 (Rajiv Maheswaran): 농구의 격렬한 동작 속에 숨은 수학
Using advanced data analysis tools, Rajiv Maheswaran and Second Spectrum help make basketball teams smarter. Full bio
Double-click the English transcript below to play the video.
과학에 푹 빠져있습니다.
by the science of moving dots.
in our offices, as we shop and travel
전 세계를 여행할 때처럼 말이죠.
and around the world.
if we could understand all this movement?
굉장하지 않을까요?
의미를 찾아 이해할 수 있다면
and insight in it.
굉장히 능숙한 시대에 살고 있습니다.
at capturing information about ourselves.
sensors or videos, or apps,
with incredibly fine detail.
움직임을 담아낼 수 있습니다.
있는 곳 중 하나가
where we have the best data about movement
or football or the other football,
and our players to track their movements
움직임을 포착할 수 있습니다.
is turning our athletes into --
and like most raw data,
대부분의 원데이터 처럼
and not that interesting.
흥미롭지도 않은 것이죠.
basketball coaches want to know.
알고 싶어하는 것들이 있죠.
because they'd have to watch every second
처리해야 하기 때문에 알 수 없죠.
the game with the eye of a coach.
경기를 지켜볼 수 없다는 거죠.
무엇을 가르쳐야 할까요?
shots and rebounds.
같은 것들을 가르쳤죠.
알 수 있는 거죠.
slightly more complicated.
어려운 것을 가르쳤죠.
and pick-and-rolls, and isolations.
아이솔레이션 같은 기술이죠.
Most casual players probably do.
대부분 보통 팬들도 아마 그럴 거예요.
the machine understands complex events
이해하는 수준까지 와 있습니다.
같은 기술 말이죠.
전문가만 아는 것입니다.
with the eyes of a coach.
볼 수 있도록 가르쳐 왔습니다.
이것을 할 수 있었을까요?
something like a pick-and-roll,
같은 것을 설명해 달라고 하면
인코딩하려면 죽을 맛이죠.
it would be terrible.
in basketball between four players,
이런 식으로 움직이며 일어납니다.
두 명은 방어하면서 말이죠.
without the ball
guarding the guy with the ball,
수비하는 선수 옆을 지나가다가
and ta-da, it's a pick-and-roll.
짜잔, 이게 픽앤롤이죠.
of a terrible algorithm.
he's called the screener --
but he doesn't stop close enough,
충분히 막지 않는다면
and he does stop
it's probably not a pick-and-roll.
아마 픽앤롤이 아닐 거예요.
모두가 픽앤롤이겠죠.
they could all be pick-and-rolls.
the distances, the locations,
위치에 따라 진짜 달라집니다.
we can go beyond our own ability
우리가 알고 있는 것을 설명하는데
Well, it's by example.
예를 들어보죠.
"좋은 아침이네. 기계 씨.
"Good morning, machine.
and here are some things that are not.
이전에 없던 몇 가지가 있네.
그것을 분리해내는 거죠.
features that enable it to separate.
to teach it the difference
차이점을 가르치려고 하면
게 어때?"라고 말하겠죠.
use color or shape?"
그것들이 뭔가라는 거죠.
what are those things?
처리할 수 있을까요?
the world of moving dots?
거리, 타이밍, 속도로
with relative and absolute location,
진짜 핵심이죠.
of moving dots, or as we like to call it,
시공간적 패턴 인식이라고 하죠.
in academic vernacular.
you have to make it sound hard --
it's not that they want to know
알고 싶어하는 것이
안 일어났는지가 아니라
how it happened.
바로 직관때문이죠.
So here's a little insight.
the most important play.
가장 중요한 플레이입니다.
and knowing how to defend it,
경기의 승패를 가르는 열쇠입니다.
and losing most games.
has a great many variations
있다는 것을 알아냈죠.
진짜 중요한 일이죠.
is really the thing that matters,
to be really, really good.
이것이 필요한 이유죠.
두 수비 선수가 있죠.
and two defensive players,
준비합니다.
the pick-and-roll dance.
받아들일 수도 거부할 수도 있죠.
can either take, or he can reject.
외곽쪽으로 빠질 수 있죠.
can either go over or under.
아래쪽으로 뛸 수 있죠.
따라가거나 스크리너를 막거나
or play up to touch, or play soft
either switch or blitz
most of these things when I started
이런 것들을 대부분 몰랐어요.
according to those arrows.
움직인다면 좋았을 거예요.
but it turns out movement is very messy.
움직임은 매우 복잡했죠.
움직이고 변형도 많았죠.
these variations identified
구하는 데 시간이 걸렸기 때문이죠.
a professional coach to believe in you.
어려움에도 불구하고
with the right spatiotemporal features
이런 변형을 확인했죠.
to identify these variations.
almost every single contender
on a machine that understands
기계로 만든 소프트웨어 말이에요.
that has changed strategies
바꾸는 조언을 해 줘
very important games,
경기에서 승리하게 했죠.
30년동안 리그에서 있었던 코치가
coaches who've been in the league
advice from a machine.
수용했기 때문이죠.
it's much more than the pick-and-roll.
훨씬 더 신났어요.
with simple things
하는 일을 다 이해하지 못하지만
much of what it does,
to be smarter than me,
그렇게 특별해지지 않겠지만
can a machine know more than a coach?
코치보다 더 많이 알 수 있을까요?
to take good shots.
슛을 잘 쏘기 바랍니다.
it's a good shot.
대개 슛하기 좋지 않죠.
by defenders, that's generally a bad shot.
or how bad "bad" was quantitatively.
나쁜 게 얼마나 나쁜지를요.
using spatiotemporal features,
시공간적인 특징을 이용해서
What's the angle to the basket?
어떤지 알 수 있었죠.
What are their distances?
그 거리는 얼마인지?
움직이는지 볼 수 있었죠.
at how the player's moving
and we can build a model that predicts
예측 모델을 만들 수 있었죠.
would go in under these circumstances?
나올 가능성이 얼마나 될까?
and turn it into two things:
지금 둘로 나누어 해석하죠.
and the quality of the shooter.
because what's TED without a bubble chart?
이거 없이는 테드 강연을 할 수 없죠.
and the color is the position.
색은 선수들의 포지션입니다.
we have the shot probability.
슛성공률입니다.
슛을 성공하기 힘들죠.
슛을 성공하기 쉽죠.
bad at the bottom.
아래는 능력이 나쁜 선수입니다.
넣는 선수가 있는데
47 percent of their shots,
NBA 평균 선수들이
takes shots that an average NBA player
is that there are lots of 47s out there.
선수가 많다는 거죠.
giving 100 million dollars to
잘 던지는지 아는 거죠.
how we look at players,
보는 방법을 변화시키지 못하지만
변화시킵니다.
a couple of years ago, in the NBA finals.
몇 년 전 NBA 결승 경기에서
20초가 남았었습니다.
there was 20 seconds left.
동점을 만들려고 했으나
came up and he took a three to tie.
리바운드를 잡았고
레이 알렌에게 패스했습니다.
named Ray Allen.
They won the championship.
농구경기 중 하나였습니다.
games in basketball.
the shot probability for every player
가능성을 파악하는 능력을
a rebound at every second
설명할 수 있게 되었죠.
that we never could before.
그 비디오를 보여줄 수 없네요.
I can't show you that video.
그 순간을 재현했습니다.
about 3 weeks ago.
that led to the insights.
This is Chinatown in Los Angeles,
로스앤젤레스 차이나타운이죠.
the Ray Allen moment
that's associated with it.
재현했습니다.
보여주려고 합니다.
of the professional players, it's us,
제 동료라는 거죠.
announcer, it's me.
제가 하고 있다는 거죠.
chance of happening in the NBA
일어날 확률이 약 9%라고 하죠.
and a great many other things.
말하려고 하는 게 아닙니다.
it took us to make that happen.
of every NBA game -- it's not that.
파악하려고 한 직관은 그게 아니죠.
a professional team to track movement.
전문적인 팀일 필요는 없습니다.
player to get insights about movement.
전문적인 선수일 필요는 없죠.
우리는 모든 순간 움직이니까요.
sports because we're moving everywhere.
pick-and-rolls,
알려줄 수 있습니다.
the moment and let me know
any second now.
있는 모든 순간을 말이죠.
our buildings, better plan our cities.
더 잘 할 수 있을 겁니다.
of the science of moving dots,
앞으로 나아갈 거라 믿습니다
we will move forward.
ABOUT THE SPEAKER
Rajiv Maheswaran - ResearcherUsing advanced data analysis tools, Rajiv Maheswaran and Second Spectrum help make basketball teams smarter.
Why you should listen
Sports fans can get obsessed with stats about player performance and game-day physics. But basketball, a fluid and fast-moving game, has been tough to understand through numbers. Rajiv Maheswaran is working to change that, by offering pro basketball teams insight into game data to make better decisions. Maheswaran is the CEO and co-founder of Second Spectrum, a startup transforming sports through technology. He is also a Research Assistant Professor at the University of Southern California's Computer Science Department and a Project Leader at the Information Sciences Institute at the USC Viterbi School of Engineering, where he co-directs the Computational Behavior Group.
His research spans various aspects of multi-agent systems and distributed artificial intelligence using decision-theoretic and game-theoretic frameworks and solutions. His current interests focus on data analytics, visualization and real-time interaction to understand behavior in spatiotemporal domains. Like, say, the spatiotemporal domain around a basketball hoop.
Rajiv Maheswaran | Speaker | TED.com