Kona, Baby!

As a postscript to my blog post on Ironman pacing, I’m happy to report that Athlete B took his considerable fitness to the new Ironman Canada in Whistler and got himself a well-deserved Kona slot. I’ll leave aside the specific strategies around slot allocation, late-season Ironman races, etc., and look at the numbers:

  • Whistler is an impossible course on which to ride evenly. It is very hilly and rolling in the first half, flat in the third quarter and uphill in the final quarter (by quarters and halves, I mean by time). Athlete B’s VI (Variability Index) was 1.11 for the entire race; NP was 214 (almost exactly the same as he had in IMCdA). Bike time was 5:19 compared to 5:14 at IMCdA.
  • In his division, he had nearly identical placings after each leg. IMCdA: 35th after the swim, 16th off the bike, 11th at the finish. IMC: 36th out of the water, 18th off the bike, 9th at the finish. Of particular note was a 3:22 run at IMC vs. a 3:26 run at IMCdA after what is, on paper, a harder bike course.

Although it isn’t possible to divide the Whistler bike course into comparable segments, we can look at how Athlete B rode the course by looking at peak power numbers. Climbs are usually going to produce the higher power numbers, so as we might expect, his Peak Power for 20, 30 and 90 minutes all come in the hilly first half. What is interesting, though, is his Peak 60:

IMC-AthleteB-peak-60

This graph illustrates the he actually had his best 60-minute power in the final part of the race, which on this course effectively means he rode negative splits (there isn’t enough time in the final quarter of his ride to have his Peak 90 there).

One other interesting data point to look at is cadence relative to power.

First half:

  • Peak 20: NP 253W, avg 247W, avg cadence 80 rpm
  • Peak 30: NP 251W, avg 244W, avg cadence 79 rpm
  • Peak 90: NP 227W, avg 205W, avg cadence 82 rpm

Final quarter:

  • Peak 60: NP 227W, avg 215W, avg cadence 71 rpm

This correlates with the cadence drop that we saw in IMCdA — Athlete B’s natural inclination seems to be to reduce cadence rather than gear. Based on the athlete’s heart rate during this segment and his subsequent solid run split, however, it doesn’t appear as though this had any negative effect on his overall race.

Regardless, now he can put this data to work in Kona, which is the best news of all!

The sweet smell of Kona

The sweet smell of Kona

Ironman power fall-off: is it the gear or the cadence?

First off, this post isn’t groundbreaking exercise physiology research; it’s just numbers analysis brought to you by my friends at TrainingPeaks — they provide the software, and I provide the analysis to my teammates.

What I’ve been looking at is the power files from a recent Ironman race, one in which the athletes involved were well-trained, experienced performers at the Ironman distance:

  • Athlete A, a former pro now racing as an age grouper, who won his division at the 2013 Ironman Coeur d’Alene by a pretty substantial margin.
  • Athlete B, a younger age group male, who set a PR at this year’s Coeur d’Alene race and finished 11th in the most competitive division, but only 5 minutes out of 6th.

Athlete A is very new to training and racing with power — you’d call him “old school” in his training and racing methods. He’s used to racing by feel, and this has served him well over distances from sprint triathlons to Ultraman, yet he hasn’t done an Ironman run split that reflects his considerable ability as a runner. My working theory for this was that he spent just a little too much energy in the bike leg, which is his forté (though he also swims in the low- to mid-50s, so he’s not exactly a slouch there either). A couple of us on the team basically forced him to get a power meter, so with a few rides’ worth of power numbers, including a sprint triathlon the weekend before Coeur d’Alene, Based on what I could see, I guesstimated his “ride-to” power for the Ironman distance at 255W. I could tell he thought that sounded low — in a lab environment back in 2007, the analysis based on blood lactate tests done on a CompuTrainer had him pegged at 325W for Ironman power. But in trying to ride to that in a summer Ironman race, he had cooked himself by mile 70. In the meantime, we have all sorts of real-world data from the pros at Kona, the best of whom don’t even average 300W. “Try it,” was about all I could say.

His race went great, but not perfectly. He was 2nd out of the water, but started cramping and estimates that it took 20-30 minutes into the bike to fully rid himself of the leg cramps. During this time, he was passed by a number of guys and was in the for-him unusual position of having more riders around him than he was used to. Once the cramps subsided, he did some work to get back into his more usual front-of-the-pack position, and ended up coming off the bike in 1st with a 5:07 split. On the run, he was able to run sub 3:30 — something he hadn’t done in about a decade — and win the age group by a comfortable margin. Kona, baby!

So the question was, how did the power-pacing experiment go?

If we look at his TrainingPeaks file, we see a number of things:

IMCDA-AthleteA-entireride

  • Normalized Power of 231W — considerably under the 255W I had suggested for his ceiling.
  • Variability Index of 1.07, which means he rode fairly evenly (1.00 is perfect), especially for an undulating course.
  • TSS of 268 and Intensity Factor of .72, meaning he left plenty in the tank for the run (normal target TSS is 280-310; ideal IF varies according to how long the athlete takes for the ride).

The great thing about looking at Coeur d’Alene is that it is a two-loop bike course, so we can compare his first and second halves to see if the effort was truly uniform. If we zero in on Peak 30 and Peak 120 (best power for 30 minutes and 120 minutes, respectively), we see that those both occurred in the first half:

Peak 120 shows an NP of 250 — much closer to the target 255. Note that the average cadence was 77 vs. 78 for the entire ride.

IMCDA-AthleteA-firsthalf

Peak 30 occurs within Peak 120, and it involves a sustained climb, so there’s a spike in power up to 265W for that period. The cadence for this segment also equals the average cadence for the entire ride, at 78 rpm.

IMCDA-AthleteA-firsthalf-peak30

 

The timing mats had Athlete A doing the first half in 2:32:19 vs. a 2:35:04 for the second half, so there was a slight positive split. Lots of factors could account for this, though — a shift in the wind being the most likely external culprit. However, if we look at the same part of the course where the athlete did his Peak 120, we see a marked difference in power output:

IMCDA-AthleteA-secondhalf

  • NP of 219 vs 250 in the first half
  • Average cadence actually went up from 78 to 80
  • VI identical to that in Peak 120 at 1.04

Athlete A reported being aware that he was not comfortable pushing the same wattage in the second half, and the numbers would indicate that he likely backed off the gear but increased his turnover slightly. This strategy seemed to pay off, as he was able to run sub 3:30 off the bike for the first time in a decade, winning his age group by a massive 24 minutes. He used no watch or heart-rate monitor during the run; his major reported issue in that discipline were foot blisters he sustained early on and had to deal with for the bulk of the run.

Athlete B put together a similarly strong race and reported feeling as though he could have gone a bit harder in the bike. If we look at his data, we can put some numbers to that perception:

IMCDA-AthleteB-entireride

  • A TSS of 219 and IF of .65. This is very low indeed for a guy whose bike split is 5:14 on a hilly course.
  • VI of 1.06, so ridden very consistently given the hills.
  • Cadence averaged 80 rpm, 2 rpm higher than that of his elder teammate.

Athlete B reported feeling as though he were “holding back” the entire ride. If we look at the Peak 120 numbers, we do see a difference between the halves:

IMCDA-AthleteB-firsthalf

  • NP of 226W, 11W higher than his average for the entire ride.
  • Cadence of 82, or 2 rpm higher than his average for the entire ride.

Peak 30 occurs at approximately the same segment as the Peak 30 of Athlete A:

IMCDA-AthleteB-firsthalf-peak 30

Relative to Athlete A, Athlete B actually held back a little more in his Peak 30: he was 8% higher than his NP for the entire ride, whereas Athlete A was 14% higher than his NP during his Peak 30.

Yet we do see a similar fall-off, both in average speed and power, between the halves. The timing mats show a 2:35:09 first half vs. a 2:39:12 second half, and the power numbers do point to the cyclist’s own output as a significant factor. Looking at the same segment we had in Peak 120:

IMCDA-AthleteB-secondhalf

  • NP of 209W, 17W (or 7.5%) lower than in Peak 120.
  • Average cadence of 78 rpm, vs 82 in Peak 120.

What’s interesting is that this fall-off isn’t simply a “late in the ride” phenomenon; if it were, we would expect a significant drop in power after 90 miles. What we see instead, if we look at the Peak 30 segment during the second half, seems to be more of a conscious “holding back”:

IMCDA-AthleteB-secondhalf-peak 30

  • NP of 213W vs. 233W — 8.6% lower than in Peak 30.
  • Average cadence of 78 rpm, vs 82 in Peak 30.

This suggests that Athlete B used similar gearing in the two halves but simply dialed back his cadence in the second half to keep the same RPE (Rate of Perceived Effort), whereas Athlete A actually increased his pedaling tempo in the second half but decreased his resistance. Athlete B, like Athlete A, did have a very good run split in the mid 3:20s, though it was about 10 minutes slower than he believed himself capable of.

A sample set of two athletes in one race obviously doesn’t hold up to any scientific standards by which one could draw any conclusions. Both athletes are very fit but do not put in the type of volume that professional Ironman triathletes do, but their type of training does correspond well with that of talented “recreational athletes” — those who don’t do it for a living. I believe there are several factors at play:

  • Both athletes ultimately defaulted to RPE as an upper bound for their output. In other words, they maintained the same perceived effort. But natural fatigue and environmental factors (e.g., warmer temperatures as the day went on) increase RPE naturally, so the athletes had to reduce their output in order to keep their effort constant.
  • We don’t have heart-rate data for Athlete A, but we do for B, and we actually see his average heart rate drop 3 bpm in the second half equivalent to his peak 120.
  • Different athletes achieve “output relaxation” in different ways – some drop the cadence and some drop the gear. The fact that the more experienced athlete dropped the gear and increased his cadence is interesting, but we lack sufficient field data to draw any conclusions from that.
  • What I believe this does suggest, though, is that “negative splitting” (increasing effort and output as the race goes on) must be done consciously and must be practiced in training. I’ve now seen many more power files from Athlete A, and this is a technique he has been practicing in his training rides as he prepares for Kona.

On a final note, I also looked at my own race from Ironman Arizona, which is a three-loop course, but one that is much flatter than Coeur d’Alene’s. If we look at Peak 90 (120 would be longer than the loops took me), we see:

IMAZ-firstthird

  • Lower NP than either Athlete A or B by a lot. That’s because they are better riders. 🙂 But they also weigh 10% more than I do, and IMAZ being an easier course, I had a faster bike split (5:04).
  • Higher average cadence than the IMCdA athletes. That might be a function of my coming from a running background, or might simply reflect the course differences between the two races.

Then if we look at the final third of the race, we see this:

IMAZ-lastthird

  • A drop of 8W in NP (from 198W down to 190W).
  • A drop in average cadence from 88 to 85.
  • Time loss of about 60 seconds.
  • Surprisingly, superior VI (1.01 vs 1.03).
  • Rise in average heart rate of 8 bpm (going from 125 bpm to 133).

I definitely recall making a conscious effort to work a little bit hard in the final third, but not by much, since I still had the run to go. Perhaps that is reflected in the rise in heart rate; more probably, though, the heart rate has to do with my high sweat rate and resulting dehydration (my training partners can attest to my “king of sweat rate” claim).

I’ve been making an effort to incorporate negative splits into my own Kona preparation as well. I did actually negative-split the Kona run back on a very hot day in the 2009 race, but the 3:56 run split was nothing to write home about. Due to the heat, I made the decision to approach the run from the outset in survival mode, which meant I walked every aid station from the first one on. With my sights set a little higher for this year’s race, a negative-split run will be a difficult feat to repeat.

“Coach Ian”

I had a pretty good 2012. Actually, a really good 2012, at least from an athletic perspective:

  • Two Ironman finishes, including the infamous 2012 IM St. George
  • 3-minute Olympic-distance PR of 2:15:01 at Malibu
  • 22-minute Ironman PR of 10:04:24
  • Best age group placings ever at Ironman events: 10th at St. George, 7th at Hawaii 70.3, 3rd at Arizona
  • Qualified for 2013 Kona at Arizona
  • All at age 50 🙂

Ok, turning 50 definitely helped with the placings and the Kona slot, but the rest is pretty unusual, particularly because I’m not a newbie to the sport — I did my first tri in 1983 and have done more than 100, many of which you can’t even find on Athlinks because they predate the WWW (“that’s right, sonny, back in my day we got our race results on paper, mailed weeks or months after the event, and we liked it!”). So what made the difference?

Two things:

  1. Emphasis on the bike.
  2. Measuring almost everything.

Triathlon, especially an event as long as the Ironman, is all about the bike. Well, sort of. You have to get through the swim without losing a ton of time, and of course you have to close the deal with a good run, but like many triathletes I come from a running background. That has always been my strong suit. But the bike is where you can gain the most time and set yourself up for your best run. So I started last year by riding 5-6 days a week, mostly on my trainer. Using, in the beginning, two Spinervals Super 6 programs to establish my base.

Measurement and planning was all thanks to TrainingPeaks, whom I had the pleasure to represent as an ambassador athlete in 2012 and again this year. I have limited training time, like pretty much everyone else, so I have to make it count. Here’s a good picture of 2012 on my TrainingPeaks dashboard:

TP 2012In the pie charts on the upper right, you can see visually the emphasis on the bike, both in distance (which you would expect) and in duration (which you would also expect, though to a lesser degree). You can also see, in the widget on the bottom right, my two most important metrics: Intensity Factor (“IF”) and Training Stress Score (“TSS”), both of which were invented by the TrainingPeaks folks. What the blue values (TSS per week) show is how I built early on for St. George, then recovered, then built big time in the late summer/early fall for Arizona. All at a fairly steady average IF of above .75 and below .85, which for me at least is a sweet spot for Ironman training.

Along the way last year, I also guided the training of one of my teammates, the infamous Mikey, who shows up in many of my “Garage of Pain” and other key sessions. He had a similarly stellar year relative to normal, other than some bad luck at Arizona, but the point is the program worked — not just for me.

It was such a success, in fact, that we’re rolling it out to the entire Team Sheeper crew of athletes this coming year. We’ll be using TrainingPeaks for early season fitness and strength building, and then for specific programs aimed at goal half Ironman (Wildflower, Hawaii 70.3, Vineman 70.3) and Ironman (Coeur d’Alene, Lake Tahoe, Kona) events. I’m really excited to play a role in combining the team traditions of fun and hard training with a data-driven approach for managing training load within busy lives and professions. I’ve lived it, and continue to live it, and now I get to teach others.

So I guess I’m “Coach Ian.” All I need now are some stretchy coach’s shorts and a whistle.

If you live anywhere near Menlo Park, come and check it out! We don’t bite…much.

Rise of the Brand Ambassador

This might be my first personal blog that talks about work. Or sort of about work. It’s also about triathlon. But only sort of. (Reader rolls eyes and awaits another disjointed blog post. Or stops reading altogether.)

My work specialty is text analytics (a combination of computational linguistics and business intelligence), mostly applied to social media these days. What that means is that we at Attensity analyze the content of what people say on social media along with all of the other “social graph” data: e.g., how influential people are, how things get retweeted, liked, +1d, etc. The critical part is how to accurately map random human language into structures that correspond to meaning, so that they can be counted, tracked and trended in useful ways. Oh, and do that on many thousands of posts per second without falling behind.

What’s come out of that work, besides a lot of variety (I’ve worked with the Fortune 500, the intelligence community and major media and entertainment companies) is a rapid-fire introduction to the business side of social media. We’ve applied our technology to everything from following the U.S. presidential elections (starting with the GOP primaries) and Arab Spring to who’s getting voted off each week on American Idol or The Voice. There’s a lot of subtlety to what goes on; it’s not just about identifying positive and negative sentiment. Particularly important is the role of influence.

Influence is harder to measure than it might seem — it’s way more complicated than how many followers you have on Twitter or friends on Facebook. Entire companies (e.g., Klout) have been built around the attempt to quantify influence, but even their presumably sophisticated metrics don’t ring entirely true to many. What’s clear is that influence is topic specific — if you look at the most-mentioned celebrities on Twitter at any given moment, for example, you’ll almost always find Justin Bieber at the top. However, on election day, if you looked at election-specific tweets, as we did for Bloomberg, the top celebrity aside from the candidates was Jay Z. (You can watch the video to find out why.)

(The reader is wondering when we’re going to talk swim, bike and run. Patience!) Topic-specific influence has created novel new ways for companies to market their products and brands: among them, the brand ambassador. If you think about traditional advertising, celebrities are often used as brand ambassadors, but celebrities are expensive. The social brand ambassador, on the other hand, doesn’t need to be a celebrity per se; they just need to influence a sufficiently large network of people on a particular topic.

Which brings me around to triathlon. I’ve found myself, quite unexpectedly, in the position of having become a brand ambassador. Not once, but twice already, and I’m likely to pull the trigger on a third. Why is that? There are many faster guys out there than I, though I’m reasonably quick for a 50-year-old age grouper. I can think of a few reasons:

  • I’m part of a community. I am very active on my triathlon team, and I race within my local community in addition to bigger races elsewhere.
  • I’m active on social media, but not overactive. I try to be interesting and honest, without oversharing. Hopefully I succeed more than I fail.
  • I am a gearhead. I will try almost any new product if I think it will give me an edge. I would never endorse a product just because I got it for free or heavily discounted — my litmus test is would I use this if I had to pay full price for it? Actually, in the case of TrainingPeaks and many of Wattie Ink’s sponsors, I am and already was a customer and avid user.
  • I work with other athletes to help them improve. I am eager to share what I’ve learned — which tools to use, which training sessions are most effective for a particular end goal — and to see my advice through to implementation. I think I’m most proud of the level I got my “Garage of Pain” training buddy Mike to this past year, even compared to my own results.

What’s ironic for my day job is that — so I’m told, anyway — one of my company’s investors at one point laid a bunch of printouts of various of my Facebook posts down during a board meeting and opined that it seemed that all I did was train and race. (If that were true, btw, I should have much better results than I’ve had.) Notwithstanding potential jealousy (he’s…um…not exactly the fittest individual on the planet) and probable violation of European privacy laws (he’s not a Facebook friend, so had no right to access any of my posts), he was missing the point on one of the central themes of a company he’s invested a lot of money in: social influence.

Becoming an influencer isn’t hard: write about what you know, be passionate, interesting and real, and connect your social presence back to a community of people at least some of whom you know in real life. Oh, and occasionally kick some ass in a triathlon or two. 🙂

An interesting side note on the confluence of work and hobby: the evening before the Wattie Ink Elite Team selection was announced, I got the following DM on Twitter:

wattie dmWhat was funny about that was that I had been on set at The Voice a couple of nights before, thanks to our work in media. While I had made a couple of random references to it on Twitter, I had mostly posted about it to my Facebook friends. So anyway, Wattie did his homework.

Maybe I’ll get him to be my lead investor next time. 🙂

On the upswing

After the Hawaii 70.3 race, I brought down the volume while injecting a little intensity, partly in the form of aquathons but also with some hard bike and run sessions. In july, it was time to officially kick off the second half of the training year, pointing towards Ironman Arizona in late November.

That meant injecting some long, flat rides into the program. We in the Bay Area are blessed with a lot of beautiful, hilly terrain to ride on, but most Ironmans are pretty flat and involve consistent pedaling over 5+ hours. My team’s normal weekend rides involve some long climbs and long descents, and the issue that presents for those doing Ironman is the uneven nature of the effort. Feast or famine, so to speak.

So a few of my teammates and I who are doing Arizona have taken to riding long from Lake Almaden in San Jose, going through Morgan HIll, Gilroy, Hollister and beyond, in order to get long, consistent pedaling time. The last one we did was 112 miles in 5:45 of riding time, which equals an Ironman bike leg. That went pretty well, and 10 or so days later, the benefits of that long ride and the previous one two weeks before that one seem to be kicking in: I feel like I’m on fire.

Analyzing things with TrainingPeaks, part of the reason is that I’ve pushed my base up from where I was earlier in the year when I was getting ready for Ironman St. George. The telling number is Chronic Training Load, or CTL. Prior to St. George, I was hovering just over 80. Currently I’m at 95, and it’s still on the rise thanks to a couple of pretty big weeks. Here’s the graph:

The 70.3 World Championship is in Vegas in a week and a half; I won’t be tapered for it, but hopefully I can maintain my “on fire” form until then. We’ll see!

Ready or not…Hawai’i, here I come!

It’s coming up on three weeks since The Toughest Ironman EVER™, and it’s one week until I hit the scary age of 50. A couple of days later, I’m racing Lance Armstrong I will be in the same race as Lance Armstrong at the Ironman Hawaii 70.3. Actually, I’m pretty sure I was in the same race as Lance back in 1987 at the President’s Triathlon in Dallas, but I can’t find any documentation for that online. I know that I did the race, but it might have been 1986 — one’s memory fades when one gets to be my age. Anyway, that was when Lance was a 15-year-old triathlon pro and I was a mediocre 24- or 25-year-old age grouper in the relatively early days of the sport. Now I’m 50, goddammit, and I’m gunning for him. 🙂

Not that he should be sweating it. I’m not even sweating it, because this wasn’t even originally on my race schedule. I love going to Hawaii, but I’m a little snake-bitten in this particular race — it’s one of the harder half IMs on the circuit due to the heat and winds (and even the swim can be pretty choppy). So I don’t really have great expectations going in, which sometimes means you end up having a great race, and other times…

I’ve been using TrainingPeaks to monitor my IMStG recovery, and I have done a few key workouts — mostly some intensity to remind the muscles and cardio system what it’s like to go hard. Here’s the latest Performance Management Chart:

I ran quite a Training Stress Balance deficit by doing IM St. George, but a week-long work trip in Europe after the race forced me into little post-race training (though not exactly rest and recovery), and I’ve gradually reintegrated training. Still, I sometimes feel as though I’m teetering on the razor’s edge between fitness and illness. What I do know is that my power numbers on the bike are still down, particularly at the high end. On the other hand, running-wise I was able to knock out a few 1km repeats in around 3:40 (about 5:55 mile pace) on Sunday before deciding not to press my luck, so I’ve still got a little leg speed.

We’ll see — I’m looking forward to a nice time on Maui and the Big Island, and the race will be a little lab experiment.

Racing Makes You Old Before Your Time

In a month or so, I’m turning 50. Gulp! I see something like this on the horizon:

Actually, I did get to drive that car while last in Honolulu, and I’m married to a young (all of six weeks younger) blonde, so I’ve pretty much ticked the boxes on my midlife crisis.

My first goal race of the season, Ironman St. George, is coming up on May 5, and in triathlon you race in the age group based on your age at the end of the year, which means this will be my first race in the M50-54 division. Even though I won’t technically be 50 yet.

Six days before that, I’m entered in the Big Sur 5K, a nice scenic little run held in conjunction with the marathon. There I’m racing as a 49 year old. In running, you race the age you actually are.

Theoretically, I guess I could jump into another running race in mid May and race as a 49 year old once again, after having raced IMStG as a 50 year old.

I wonder if this is that “fountain of youth” that Ponce de Léon was looking for.

Balancing Quantity, Quality and Incomplete Data

I’m a data-driven guy, both professionally and in my training, but in both sides of my life, numbers alone don’t tell the entire story. Words, feelings, tone, etc. all provide a qualitative dimension to what would otherwise be pure numbers, which you can interpret often any which way you want (“lies, damn lies, and statistics”).

A recent example of numbers not telling the whole story is my company’s analysis of Twitter for Super Tuesday — the numbers of positive posts for each candidate alone were not accurately predictive of outcome in 5 of the 10 states. However, if you look at the data more closely, you’ll note that we did get the most populous states — the ones we had the most geo-coded data for — we did get it right. That at least puts our 5-5 record in a slightly better light. 🙂 We did go out on a limb with USA Today in a pre-Super Tuesday article as well as a post-event analysis, and we ourselves didn’t believe we would get every state right, in particular the ones where we had very low numbers of tweets and where the difference between candidates wasn’t large. In this case, more data would likely have been better.

More interesting, though, is the qualitative data behind it — what did people say about the candidates and issues that would explain the why behind the results? We provided some of the insight here, but we have a lot more behind the scenes that we haven’t yet made public.

With my triathlon training, balancing the quantitative with the qualitative is a different exercise. I’m an experiment of one, so large numbers of experiential data from other athletes can tell you a lot about what has worked in general, but you still need to map the numbers to your own situation. High volumes of training that work for some athletes break other athletes.

This is where I’ve been using TrainingPeaks’ Performance Management Dashboard to help me look at how I’m progressing over the course of my buildup to Ironman St. George. I’ve been investing a lot of time and effort in cycling, since I come from a running background and have that leg pretty much nailed as long as the bike doesn’t take too much out of me. I’ve been following Coach Troy’s Spinervals Super 6 challenge, and it’s worked well for me — I improved my 20-minute threshold power number by 13 watts in the first six-week phase, and am looking for more in the second phase.

This improvement hasn’t come without cost, though. I had a sub-par half marathon in Austin in mid February (a 1:28:30, which normally would be a moderately hard training run, but on that day was “all out” on a pretty hilly course), and in general I’ve been feeling pretty tired on a lot of days.

Looking at my Performance Management Chart, I can see a couple of things happening:

Image

First, my CTL (Chronic Training Load, the blue line) has been on a steady increase over the past 90 days. This is a good thing. I’m averaging above 70 TSS points a day on a consistent basis (the CTL in my case is a weighted average over the past 42 days). The pink line, my ATL (Acute Training Load), reflects my activity over the past seven days. The difference between these is my Training Stress Balance (TSB), the yellow line. You can see that that has been negative for some time, which pretty much explains why my one race wasn’t good and also why I generally feel tired.

This week, however, I’m on a little family vacation in Hawaii, and I didn’t take my bike (I’m on Oahu, which is not the most bike friendly of islands), which means I’m using the stationary bikes in the fitness center where I’m staying and not putting in really long rides. I also don’t have power data being uploaded (the bikes do display watts, however), so my TSS scores are being calculated based on heart rate data, which is ok but not the same. Anyway, the bottom line is that it is a true recovery week for me, and the numbers bear that out.

Qualitatively, I’m starting to feel less tired. After a few hectic days (the Super Tuesday stuff mentioned earlier) of “working from paradise,” I’m also now catching up on sleep (9+ hours last night), and that also seems to help. Which brings us to the topic of “incomplete data”: although TrainingPeaks provides for the analysis of nutritional data and other metrics, it’s a lot more work to enter that stuff, so I basically don’t do it. However, that means that my analysis by definition is based on incomplete data; how well you eat, how much you sleep, how stressful other parts of your life are — these can all affect your training, your relative feelings of fatigue, even your day-to-day level of motivation to train hard.

It’s just like with the primaries and social media — we know where some tweets are coming from, and we can analyze the language in them to know who and what they’re talking about, but we don’t, for example, know whether they’re actual voters, whether we’re looking at a second or third account from the same person, exactly what their age, sex and political/religious/etc. affiliations are, so the data is incomplete. But in the aggregate, it still tells you a lot about the issues facing the candidates, their relative chances of success both now and in the general election, and which messages are resonating with the public.

So I guess the lesson is that data will necessarily be incomplete, but it’s still useful if you know what to listen for.

Early Season Base and Spinervals Super 6

I confess with both my readers as my witness: I’m not a big fan of aerobic base building. Like a lot of things I know are good for me, I know I should be doing it before I move into the fun stuff — intervals and threshold — but I just find it mind numbing. Particularly the long aerobic ride. Particularly solo.

But one needs to work on one’s “limiters,” as the data-driven endurance coaches say, and one of mine is definitely cycling endurance. Other than Ironman Arizona 2010, all of my Ironman rides have been far below what I’m physiologically capable of doing. It doesn’t matter that you’re a decent runner if you lose 30-60 minutes in the bike. What I did differently in 2010 was do some well-timed volume on the bike in that high Zone 2 / low Zone 3 area, so that seems like something I should get back to.

To kick off the year in style, I’ve been following the Spinervals Super 6 program. I’ve been a Coach Troy fan for many years and got a chance to meet him in person at an Ironman breakfast in Kona this past October, though of course I felt like I already knew him thanks to many hours of suffering through his workouts in my garage. Basically, the program is six weeks of six rides per week with running and strength work thrown in for good measure (it doesn’t address swimming — for that you’re on your own). During the week, it’s got a mix of intensity (my fave) with steady aerobic work, but the Saturday ride gets long. Today’s was very long indeed — 100 miles in about 5.5 hours, for those hardcore types who can stay on the trainer that long. Short option was 3-4 hours. I enlisted the company of my good friend and teammate Mike, who like me has Ironman Arizona on the docket this year.

I barely made it to 3.

It started fine and controlled, but in the second set the intensity went up, and the gearing Coach Troy was using seemed too high on the fluid trainers we had — I hit over 300 watts in the last minute of the first repeat. This is higher than I can even maintain in a sprint distance triathlon, so well above lactate threshold. My heart rate never really recovered from that; where my aerobic efforts had been below 130, I was consistently staying above 140 for much of rest of the workout, which meant I had burnt a lot of matches during that set. Effectively, I bonked and made it to 3 hours only by force of will.

Looking at the workout in TrainingPeaks, the graphs tell the story. In the bottom graph, you can see the correlation between heart rate and power as the workout goes on; in the third hour, the power goes down but the heart rate stays high. In the scatter graph (one of the new beta features I have access to), if you plot power and heart rate data points, you see a wide fan-out of power and heart rate at the upper end of the X axis (heart rate); normally you would expect a smoother upward trend on both axes.

What this tells me is that I have a lot more long aerobic work to do, and it also illustrates what happens when you go into the red zone during a long aerobic effort — you burn enough matches to end your day prematurely.

Live and learn. 🙂

Catching up in the New Year

I’m perennially running late, and so it is with race reports, years in review, etc. 2011 was a pretty good year, training-, racing- and otherwise.

Highlights:

  • Two marathons within 13 days, by far the shortest time I’ve ever had between races of that magnitude (previous shortest was 11 weeks.
  • Second Hawaii Ironman (tenth Ironman overall), with a 37-minute time improvement
  • Being selected as a TrainingPeaks Ambassador for 2012

I think I’m probably proudest of the marathon double (3:03:08 at Boston and 3:04:57 at Big Sur), especially since a calf strain severely curtailed my training leading up to Boston — I averaged 11 miles per week from the beginning of February through Patriots Day. Compensated for by lots of cycling, swimming and muscle memory, of course.

Still, I think that double hurt me in Kona. My cycling never really came together, especially for longer rides. And my bike crash in Big Kahuna didn’t help, either; I still have some scars from that one, literally.

For 2012, I am renewing my focus on cycling strength, kicking things off with the Spinervals Super 6 on TrainingPeaks. I’m also having fun beta testing some new TP features (more on that in another post). As a software company CTO, though, I have a bunch of other suggestions; I’m reining myself in for now.