Data Science for Coaches
How do you if your training methods are effective? The simple answer is that you track performance over time, and if it’s improving, you know whatever you’re doing is working.
A more difficult question to answer is why your training methods are leading to the observed performance improvements. I’m always skeptical when I hear coaches talk about their mitochondrial biogenesis protocols, maximum lactate steady state progressions, and methods to increase stroke volume. It’s not that I don’t believe that these methods improve their athlete’s performance, but rather that I don’t buy the reasoning behind them.
There are plenty of protocols that should elicit a given adaptation in theory. Still, we put them to the test, we don’t have a reliable way of knowing how and why they lead to performance improvements. As a coach, you may not even care why something works (as long as it does work), but there’s a good argument to be made for why you should care.
At some point, you’re bound to encounter an athlete who doesn’t respond to cookie-cutter protocols. If you don’t understand (a) what that athletes underlying limitations are and (b) how to target that limiter effectively.
One way to better understand (a) the effect of your methods and (b) how they relate to increases in performance is through basic data science methods. In the picture above, we have Delta-SmO2 and maximum power output data from a Crossfit athlete over 36 weeks of training. Delta-SmO2 represents the rate of change of muscle oxygen saturation, which clues us in to the balance of oxygen supply and demand. A more negative value means improved maximum oxygen extraction.
When I started working with this athlete, we identified that their rate of oxygen utilization was a primary limiter factor for increasing their VO2max and maximum power output. In testing, we also found that their maximum rate of oxygen utilization was -4.5 % / second, and their maximum power ouput on an Echo bike was ~1315 watts. Over 36 weeks of training we had them repeat the exact same protocol to improve their maximum rate of oxygen utilization every saturday and each week we tracked their maximum rate of utilization (Delta-SmO2) and their maximum power output, which are both displayed above.
We found 31% increase in maximum oxygen extraction and 20% increase in maximum power output over this period. But, the real kicker is that when we calculated the correlation between their Delta-SmO2 and maximum power output over time the R-Squared value was -0.95, which means a linear relationship between improved oxygen extraction (relative to supply) and increased power output. In other words, as we increased this athlete’s rate of oxygen utilization there was a proportional increase in maximum power output. Furthermore, when we calculated the correlation between their training progress on the weekly repeated session and their increases in power output the R-Squared value was .84. Collectively this gives us a strong understanding of how the protocol we used works, how it changes an athlete’s physiology, and how it relates to increases in performance.
