I was scanning the chart of Jordan Webb’s HRV analysis, and figured I would share my frustrations and challenges with getting data, analyzing data, and sharing data. The new responsibility of coaches now that we are post Moneyball is data and evidence based coaching. I make 3-6 taps on my smart device during practice, instead of constantly scribbling down notes like I did ten years go. They are attendance, practice success, and medical. Most of the time I find myself doing hands on stuff that is not cool to talk about like moving hurdles, helping with partner activities, or just miming things from a distance. Sometimes I am camera guy. I will be posting a guest blog with Mladen later this week if I can, as we are seeing some interesting trends. I will share reality, even if it’s not as pretty as the HRV chart pictured here.*
Reports, similar to dashboards, should be one page and share the data. Tables are fine, as graphs are often moronic pie charts or line charts that increase questions instead of increasing the answers. If one is to show a chart, show multivariate data, not one factor if possible. The chart that I shared was wrong, while it was pretty, it didn’t show cause. I know show sleep, training load, nutritional consumption, HRV, and medical/mental at the same time. More data doesn’t’ clutter if done right. Clarity not confusion.
Good data tells a story, and timelines are going to be hot topics next year as a new company is working on medical and physiological reporting for coaches that are not experts on injuries and training. The first thing I ask is for a copy of the raw data and then the report summary separately. Not having raw data frightens me as it’s likely to be recall or summarized findings (memory). Record what is unexpected, as I like precording, or inserting data that is expected and on changing what doesn’t meet expectations. We only need the unexpected as much of what we do is more did it get done versus specific data points. Very few data points are outliers, but Malcolm Gladwell and other pop writers focus too much on what is neat versus necessary.
Get longitudinal data if you can, or do an intensive and brief study. Extrapolating too much is a danger but projections are fine so long as you are not banking too much on the results. Daily, weekly, and monthly collections are key. Data collection should mirror farming. Frequency of sampling or testing should match the adaptation curves or time required for further testing. HRV is daily or even more, but power testing doesn’t need to be done every time you are in the weight room. You can’t let sampling or collecting data interfere with the training or replace workouts. We train too infrequently and too wimpy now, imagine the data collecting of plush workouts. Basic stuff like body weight, skin folds, and other gross measures still need to get done. Find faster and easier ways to get legacy data sets as you don’t add time to the process.
Passive data collection trumps active. Chasing people around with lactate testers or swabbing cheeks should be minimized. Too many western (North American) coaches are doing everything but coaching. Train, please. The future is going to be auto sampling and BAN, letting technology do the dirty work. Some elbow grease is required because at the end of the day we need to get the job done. Focus on reducing the steps, as well as athlete participation because doing something boring stinks for everyone, including the geeky coach that reads research.
Google spreadsheets are ok to begin with, but web 2.0 methods are dated and Mobile 3.0 is the future. Simple use of Tapforms is great as it can be exported to .CSV and is dropbox friendly. After you collect the data back it up and start visualizing it with excel or if you have the money a BI system. Regression analysis is nice, but trending is and prediction is beyond the R program. R is free but I would start working with a good stats person. After you get the data analyzed you may have to do infographics or a good visualization program to share the data. Be warned, the best BI systems never do it as well as the custom patch programs that are specialized to handle the data specific to your environment. Even software from equipment tends to be the last thing added at the end of the development cycle. Just get the data in .CSV format and chart or graph separately from the program if necessary.
After the data is shared, decisions and actions must be made or agreed on. I see too many reports without decisions after. Have goals of what you want to happen before you collect the data. Sometimes data shares new and better questions, but goals matter. At the end of the day it’s results that you can prove. Talk is cheap and numbers don’t lie.
* Note: The attached photo is from the BI program Roambi, and is only 100 dollars for the pro version. If you use ithlete, the new dropbox feature allows updated data to be in .csv format so you can update the charts as you need rather quickly. I like seeing the HRV weekly because I live in a controlled environment but teams want real time so I am certain that the developers at ithlete will make some rather innovative changes in the next few months for more rapid aggregation, as the entire direction is going to real time. With Roambi you can tap different trending options for different perspectives. A new sport specific app will be coming out with a massive analytics engine under the hood but still have the gestures of good programs. Many think the attached HRV data is too good to believe and even accused me of photoshopping! This is the future of Data. Simple, rapid, deep, and multivariate. I think Roambi Pro is great for yearly reports but don’t expect it to autoupdate.
Discuss entry