Correlation between VO2 and watts between cycling and running

Using these formulae from the American College of Sports Medicine (ACSM):
http://certification.acsm.org/metabolic-calcs

Running

VO2 (mL . kg-1 . min-1) = (0.2 . S) + (0.9 . S . G) + 3.5 mL. kg-1.min-1

Leg Cycling

VO2 (mL . kg-1 . min-1) = 1.8(work rate) / (BM) + 3.5 mL. kg-1.min-1+ 3.5 mL. kg-1.min-1

To compare the two, I had a snoop around to see if there was any agreement on a standardized model for watts produced running. There isn't. The best I found was from this conversation amongst well-informed men, and ran with Chung's assumption of a 23.9% efficiency rate, to which I simply plug into a well established calorie formula:
METs = VO2/3.5

METs * kg= kCal per hour

joules\cal = 4.18

EnergyOut = (kCal\hr * 4.18 * .239)

Energy out/3600 *1000 = watts

I noticed immediately when plugging my own snails pace and rhinoceros weight in that my wattage, at least by this calculation, matches my bike FTP nearly to the watt. Sure enough, when returned into the ACSM formulae, my VO2 levels were nearly identical. Obviously, the efficiency part is a big variable. If I want a 1:1 efficiency rate, then 23% works for me. I like round numbers, but it's just a plug. Anyway...

What does this mean? There's a lot of dismissive chatter on training based on VO2Max, and I get the point. I've read several studies now that show VO2Max (assuming it's measured accurately and not estimated, which is one hell of an assumption in some of these studies) doesn't change in elite athletes, and hardly changes at all in amateur athletes. This would support Noakes' work that there exists a 'central governor' wholly different from the supposed 'VO2Max' level. The works aren't as radically different as one would think though; Noakes doesn't reject VO2, just the concept of there being a 'Max', suggesting it's more likely a function of other factors, notably muscle composition and neurological factors. However it's hard to dismiss training without some respect to VO2, if for no other reason than it includes a variable for weight that is absent otherwise. I've personally trained my VO2 tolerance higher and higher, without knowing (or caring) what my 'Max' level was. If elite athletes have been measured with a VO2Max level above 70 l/min, that would at least suggest to me there's a lot of work left to be done for us dopes mucking around in the high 40s and low 50s. Lots of training anecdotes line up with this model as well; If I play with the inputs, 5 pounds equates to 10 seconds of pace in running, which we've all heard before via 1 pound = 2 seconds. If I plug in some of my teammates FTPs and paces, the model correctly guesses their weights. When I put in my peak cycling FTP, I also get the run pace I managed to hold 'magically' one particular race, even though I hadn't been running any more or less leading up to it. I'd simply been riding more. 

So I'm going to proceed with modeling workouts for now not based on TSS, but based on time spent at VO2 levels. I'll make the 'Max' level nothing more than the observed best. If your best 10K run put you at a VO2 of 51.5 l/min and you weighed 170 pounds at the time, then I would expect your corresponding bike FTP to be right around 310 (the formula produces 312, but I measure my FTP in increments of 5, as most do).

There has to be some function for time decay, however, as well as age. Riegel's formula for predicting run times in on the right track for decay, and for age I've been using these for age and gender.

I have nothing for swimming yet, and from my research so far, nobody else does either. It's hard to measure much of anything while swimming (even now, optical heart rate monitors are *just* becoming available for underwater usage at a consumer level), so it seems the science hasn't caught up. I've been using your best 800 yard pace as a proxy, basing percentages off that, and it seems to be a pretty good measure, at least as a predictive tool. I've tested it out on several of my teammates over ~30 different races at different times, and the results have been promisingly close to reality. 

At the end of the day, I'm digging at all of this because TSS seems so inherently flawed. If you need permission to use math, it probably isn't math, it's a marketing term. If you work out and produce a TSS of 90 or 91 - or even 85 or 95, it doesn't mean anything on its own. It has to be looked at as a function of time, and even then, it's only an indication that you might be over- or under-training. It doesn't appear to have an upper bound, because there is no system of debit and credit, you simply accumulate more and more TSS over time. I've seen suggestions that it might have an *optimal* upper bound of 250, but that number suggests it's impossible that any age grouper could finish an ultra distance race in less than 10 hours, when of course we observe otherwise. 

What I'm coming around to is that we are really only training three central things - our ability to create power, our ability to tolerate the toll it takes, and how quickly we recover. That's not just about lactic acid, or VO2Max, or muscle composition, or psychology, or neurology, or heart rate...it's about all of those things. For what I'm seeing, VO2 as a scale does a very good job of linking together performance and prediction across sports. If your last race result put you at 80% of your *best* VO2 at the time, and you manage to raise your VO2 in training since then (including all of the factors that go into it, including the ones you can't directly measure), then you're by definition faster. Whether you did it by tolerating pain better, doing drugs (I'm not condoning this, just saying...), or whatever, it comes out in VO2, of which you can reverse the math to find the pace. 

Poll results

I asked the good folks over at BeginnerTriathlete.com to participate in a survey to get an initial idea of what I need to cover to entice a good number of athletes into the application, and I managed to get 35 responses. Pretty good sample.

89% said they collect training data via a Garmin, or something else.

70% use a garmin, 21% use trainerRoad, 12% use Computrainer, 9% use a Timex (surprising, I didn't realize *anyone* did). Only one used a Suunto. 

30% use Garmin Connect, 37% use spreadsheets, 23% use TrainingPeaks and 23% use BT's features. 

Responses were fully across the board on whether people choose to race with their device, from No to always.

83% said they would switch applications.

The most important considerations to consider in a training planner:

Ease of use
	29	82.9%
Detailed Analysis
	20	57.1%
Compatibility with mobile device
	18	51.4%
Web-based interface
	19	54.3%
Social networking
	3	8.6%
Seamless data syncing (ie to Garmin Connect, Strava, uploading and downloading)
	27	77.1%
Cost
	20	57.1%

Interesting stuff all said. I tried to kick the hornets nest on the forum regarding the Detailed Analysis answer, and it seems like evangelism is going to be a key requirement to getting it accepted.  Keeping it simple and cheap, and linking to everything is going to be key.

Skiba on board?

I've just received word from Physfarm, the holding company for Dr. Philip Skiba, that the GOVSS, SwimScore, and BikeScore formulae are freely available for use. Huzzah! I find these to be excellent and highly credible alternatives to TSS et al, and look forward to integrating them. I've put out feelers to other model authors as well, hopefully we get a consortium, which lends a great deal of credibility to what I'm putting out there.

*update - now, last I heard is that if it's a commercial venture, they want money. That's easy, no. I don't plan to renumerate the Einstein family if I happen to cite the theory of relativity, and the Supreme Court agrees with me. Math is not patentable, and therefore not subject to fee. You can sue me for trademark infringement if I use your trademark in a way that damages your brand, but that's about it. Considering I was planning to endorse the brand and work with the author, that's a far-fetched claim I doubt would make it out of any counselors initial session.

I've since had some far more sane conversation with some lesser known but more engaged researchers, who have been collectively confirming my sneaking suspicion that none of the models do a heap of good. They tell you essentially to go, slow down, or stop. They don't tell you if you're getting any faster, if anything they can't - that part is held constant. In fact if your ability to stand training load increases, you're encouraged to re-baseline. It all makes sense of course - these models are based on elite athlete performance - there's simply not a lot left to gain! All they need is a sound plan (provided by a good coach, not software), good nutrition, and then these kind of models that tell them with some prognostication when they might be at their best to perform. And that's not what I want to accomplish. I'd like to help non-elite athletes improve, and do it based on data. That's a different goal. Thinking Red-Yellow-Green might just be all that's needed. The uncontested models such as TRIMP, VDOT, and the one I stumbled unwittingly on for swimming are more than enough to guide someone through a traffic light. That's not the meat of what I want to offer!