Is anyone using big data to evaluate best practices for running, training or racing?

sloinnorcal wrote:

try this? wrote:

Check out the book called The Secret of Running and their website. Also, the Stryd and "run with power" community has some decent quantitative tools.

https://www.outsideonline.com/2276656/what-running-power-anyway

TL;DC ('click'). Stryd is making advances, but it is still in the territory of smoke and mirrors. Is it really feasible to do for running what happened in cycling?

I have a Stryd. It's a decent device. My impression is that Stryd itself isn't doing that much other than selling devices. Their online platform and app are severely lacking. In many places there are missing very basic features and metrics but simultaneously are trying to introduce a very complex, black box, daily recommendation feature based on ML. I think they are missing the boat.

Some of the people in the Stryd ecosystem are doing cool work. This guy Steve Palladino seems to have a better understanding of training generally and training with Power than anyone at Stryd. Below is a cool anecdote about one athlete's progression. There are a lot of advanced Power metrics builit in, but I do believe that if you had the same data for 1000 - 10,000 runners you could start to test some interesting theories and learn a lot about what works best and what doesn't.

Correct me if I'm wrong, but didn't Jack Daniels basically do this back in the 80s and then published his subsequent book? He was the first to popularize a quantification of fitness through the whole VDOT system and did it in a very data driven way. Not sure what big data would really tell.

Big data is dumb.

Overhyped. It's like the compression socks of the tech world.

If and when you return to your data, I have often wondered if a moving average of daily miles has the strongest correlation to success. If it does, how far back and for what racing distance? This would be a postulate consistent with Ed Whitlock, Yuki and the just 'run more' crowd. I think I could pick the winners at every middle and high school boys' xc race based on the 90 day moving average. In a 400m, not so much.

hank jr wrote:

pop_pop!_v2.2.1 wrote:

How would a 5:30 runner benefit from the run a 10:00 hobbyjogger did? Lots of data out there. Not all of it is useful in simple ways.

They may not, but the 10 minute runner can benefit a lot from knowing what a 5:30 runner does.

I honestly think that running is far simpler than we make it out to be. We may not get anything useful out of it, but I bet they will try, and we can certainly expect them to try to monetize it, that's the whole point of their business.

I am not sure the 10 min runner can benefit that much from what the 5:30 runner is doing. In fact, it could cause more harm than good. Back before 1999 USA Cycling sold training manuals that were yearly calendars telling a ride exactly what workout to do on each day (the manuals were based on a racers category--category was determined by a points system from races). The ignorant assumption was that everyone was the same and would respond the same to training.

What could benefit the 10 min miler is hiring a coach to work with him/her individually if the runner wants to get faster. Some people just like to run though.

I'm no physiologist, but I'm pretty sure you need at least the following parameters in your data set to analyze things properly:

1. Sex, Age, Weight, Height, etc.

2. Diet with time of day

3. Supplements, PEDs, medications, etc

4. Type of exercise with pace or weight numbers, with HR, ambient temperature

5. Non-exercise activity through the day

6. Sleep

7. Illnesses, injuries

... and you need this information updated constantly.

Number 1 is not going to change much, and only needs to be periodically updated. Accounting for a bad knee or other permanent impairment would be problematic though.

Number 2 is almost impossible to track even for the most dedicated athlete. This is where these models will always struggle.

Number 3 is easier to log than 2, but there are a LOT of people using PEDs and they don't like to admit that. The PED reporting issue is likely to destroy any model no matter what you do because unreported usage will throw off all the numbers in a big way.

A garmin HR tracker with the exercise log will cover 4 for those who religiously record their workouts, but who among us record everything we do?

A fitbit like device worn 24/7 could pretty much nail 5 and 6 without too much problem.

How to quantify illnesses for 7? Not easy.

Overall, the problem is super nasty, but not insurmountable. The first step would be to create a standard method for recording these parameters and then make sure all devices and logging tools store data in that format.