Data frustrate me.
It’s not the data themselves but it’s how they are being used right now in athletic sports performance and for the last five years.
It was the end of the summer of 2013 when I first starting getting into sports data analytics. I’d just joined a startup that had technology from MIT that collected data from the body continuously: heart rate, blood oxidation, temperature and the hardest of all was blood pressure. Super hard to do continuously; hard like the only way to measure it accurately was to go to a hospital, ask them to stick a probe in your artery and lie quietly.
I jumped in…
So as the product guy, I jumped in and started to learn all I could about how data could be used in health and fitness and besides the always fantastic Sloan Sports Analytics Conference, there were a few sports data conferences popping up here and there around Boston where I live.
Not big, a few hundred folks – trainers, strength and conditioning coaches, business dev types trying to figure it out and a few brave entrepreneurs. And true to my MO, I stood at the back of the room and listened.
A really smart guy from Nike talked about collecting data from sensors on an athlete’s back and shoes and showed us a really complicated dashboard of all the x, y and z data to the third decimal point. There were scatter graphs and radar charts and table after table of data.
I just squinted at all this information and was amazed at what the Nike dude could pull from it – trends and spikes that led to hypotheses and flimsy conclusions.
The voice in my head kept rattling: really, this is just where we’re at now? (Remember, this was four years ago.)
I sat with nerds from Head who were putting sensors on the end of tennis rackets, entrepreneurs creating amazing platforms to collect and review video of the athlete, even a guy from Disney with money to burn who wanted to bring back something, anything to the Sports Complex at Disney World.
And I kid you not – every speaker who was on stage that day understood the goal of data in sports – a lot of them had a single slide with that question:
What am I supposed to do next?
I, the coach, the trainer, the athlete – all of these data are great but what happens next?
WHAT DO YOU WANT ME TO DO?
So time passed and I kept watching the new products – the ascension of Fitbit, the challenges of Pebble, Mio, Misfit, Jawbone, the slow but unstoppable advance of Garmin, of Apple.
But they still couldn’t answer The Question.
So here’s what I’m thinking:
There’s a bunch of ways to slice up the sports performance data market right now. But let me take a shot with these three.
Historical
Collect data and look at it as a record of what you did. I walked this many steps, accelerated this quickly, ran this many miles, slept these few hours. All looking backward and shown in many apps, dashboard and daily circles (what’s with the circles in app UI? Always the circles)
Predictive
With all these data, make a prediction of what is likely to happen in the future: If your data follow THIS pattern, then it’s pretty likely that you’re going to perform like THAT tomorrow. And really, isn’t this what we use data for in a lot of different fields? Business, healthcare, sports, weather. It’s trying to make a really smart guess of what’s going to happen tomorrow, next week, next year based on what happened in the past. And then we craft a narrative that tells a story of the future.
Prescriptive
Use the data to tell us what we should do next. Based on what we’ve done in the past and commingled with the outcomes from thousands of others like us, let the data tell us what our next day’s training should be because they (I’m mean the data – does anyone else think its weird to treat data as a plural? They are, but it’s weird, isn’t it?) already know. Deep Blue beat Kasparov this way – looking at all the alternatives, playing them out and choosing the next move according to the best potential outcome. Why can’t we do this with our data?
If you google “prescription sports data” you get links to healthcare services and white papers so you can’t yet use this word “prescription” in terms of data. But this is what we all want – a coach or an app to tell us what to do next that puts us in the best position to succeed.
And yeah, I get that the world is a messy, crazy, unpredictable place and that even data can’t guarantee that I’m going to score 40 points in the playoffs, or squat 300 pounds in six months. But it can surely show me the best path based on what I know right now, right?
So what I want to do is build that – a way to collect all my data, and yours and everybody else’s data and bung it all into a data container and then figure out how to chart the best path for me to get to where I’m going.
What do you think?
2 comments On Why don’t data tell us how to train?
… that if you can do it, you’ll be a millionaire.
… that if it was easy, someone else would already have done it.
I think the predictive part is the most interesting. If you can make a prediction (other than that I’ll still be a old fat white guy next week as well) that comes to be, I’m gonna trust you a whole lot more. If you provide a prescription, like a million other self-help books, and it doesn’t work exactly, there’s too much data to know exactly what went wrong.
How I parse predictive vs. prescriptive is: predictive = “there is a high probability that this is going to happen based on the data”. Prescriptive can mean: “do this, and there is a higher probability that you’ll get your intended outcome than if you did something else”. This evolves the current state of data tracking from showing where you’re likely to be in the future but no guidance to how to alter your course, to answering the question “what should I do next?”