Tri Talk Triathlon Podcast, Episode 58

The Triathlete’s Coefficient of Determination, and is a high max heart rate good, or bad? It could be the best Tri Talk episode ever! Will it live up to the hype? Let’s find out, today on Tri Talk.

Welcome to Tri Talk your podcast source for triathlon tips, training, news and more. Notable new listeners come from Albania and New Zealand. In Albania, thanks for pointing out that Eric Schwartz, the former US National Duathlon Champion and occasional Tri Talk host, is not 38 years old as he claimed, but in fact 37. Why someone would lie about being older than they really are is beyond me. In New Zealand, I’m begging you, begging you! Please send me some warm weather. Just take a box, open it up to the sunshine for a few minutes, and mail it to me. My goal at Tri Talk is to help you swim, bike, and run faster, to meet your personal triathlon goals. Whether you are an elite or amateur triathlete, we cover sprint distance to Ironman distance. I’m your host, David Warden, and this is Tri Talk Episode 58.

There is quite a bit of pressure today for me to live up to the teaser best Tri Talk episode ever. But let me put it this way: I have never been more excited about presenting a topic on Tri Talk than I am today. I’ll be sharing some information that has been in the works for nearly a year, and tell you all about the Triathlete’s Coefficient of Determination. Plus, is a high heart rate good or bad? Or, is their such a thing as an optimal heart rate, and can you do anything about it. This is an interesting topic and we’ll spend a few minutes on that at the end of the show.

Episode 57 generated some good feedback on the Tri Talk forums. In that episode I suggested that since a study concluded that cyclists had larger hearts than runners, and since larger hearts also were highly correlated to a high VO2max, that perhaps cycling would help to increase your VO2max, via a larger heart size. The user Mad Dog on the Tri Talk Forums pointed out:

“I think there may be a problem with a conclusion you made in the latest episode. Cyclists have larger hearts than runners, larger hearts translates greater VO2 max. Okay so far. But that doesn’t mean that cycling made their hearts larger. Couldn’t it be that cyclists who have succeeded in their sport had larger hearts to begin with and that those who didn’t, didn’t go on to become well-trained cyclists.”

Mad Dog is, of course absolutely right. My suggestion that cycling created the large heart is incorrect logic, because the cyclist could have become a cyclist in the first place because they had a large hear to begin with.

The user Chris on the forums also wrote in on this same topic, reminding us that:

“Optimum cadence on the bike is 80 to 90 rpms, optimum stride rate (running) is about the same. …it’s easer to train for a fast leg turnover on the bike than it is performing speed work. If the cyclist trains and races in the 80 to 90 rpm range then, not just his legs, but his whole body is accustomed to this motion i.e. cardiovascular and respiratory… As a way of cross training, you place him on a bike and have him paddle at a high rpm using light gearing and he will be able to maintain that cadence for a much longer period of time… Speed work is very important for a runner but cycling is an effective cross training method for developing leg speed.”

Thanks for your thoughts, Chris.

 

Let’s get onto the good stuff! This is a long episode today, and we have a lot to cover. Stay with me cause I’m going to go fast. Hang on tight!

I have to admit I am a bit concerned with presenting this topic. First, because I have hyped it as the best Tri Talk episode ever. You may finish listening to this topic and say, “What the heck? That was David’s version of the best Tri Talk episode ever?” Many of you might find it quite boring. Second, it may only be appealing to those of you who really like to analyze data, and who try to squeeze every last second out of their performance. If you really like numbers, this is the episode for you. If not, fast forward for about 25 minutes, or just read the second half of the show’s transcript on the blog. I can assure you again, however, that I have never been more excited to present a Tri Talk topic, and therefore to me, this is the best episode yet.

Several months ago I introduced a survey on the Tri Talk website. My goal was to try and gather some non-traditional data on athletes, and correlate them to race times. What I mean by “non-traditional” data is information that you typically don’t see on a race results listing. It is easy to get data on race times and correlate those to age and gender, because every race listing has that information as part of the race results. But what about other factors such as height, income, years of experience spending, or weight? How do these factors influence race results? Is height or weight a bigger influencer on performance? How much does your income really mean to your performance? Do you get faster at certain distances if you are a little bit older? These are questions that many of us think we know the answer to, but wouldn’t it be nice to see some data behind it? Although we can infer some correlation on weight by using the Clydesdale division, that is a binary input. You are either over 200 pounds or not. It would be even better to see the granularity of all athletes’ race-day weight to get a better correlation of the influence of weight on performance.

Many of you have already taken the survey, so to you please forgive me as I review what the survey asked for the other listeners. The survey asked for the athletes’ age and gender, but also annual income, triathlon spending, years of endurance training, years of triathlon training, and height. The athlete was then asked at which triathlon distances he or she had been the most successful. For those distances, the athlete was asked to put in the exact distance of all 3 legs of that race. As you know some Sprint-distance events are shorter or longer than others, as with all the other distances. The athlete had the chance to adjust the distances so that we could get an accurate benchmark of their Sprint, Olympic or IM event to compare to the other athletes at that distance. The athlete was also asked how fast they were that day, and their race-day weight. Based on this information, I had a database of hundreds of athletes, and their best race times at multiple distances. I could then correlate race-day performance with a host of data points.

Now, before we start to share the results, let me tell you all the flaws in this study. First of all, it is not really a true study. It is not peer-reviewed. The data inputs are voluntary. No one independently verified the athletes’ height, weight or race times. We depended on the athletes’ memory and honesty for all the inputs. What about weather and course topography? Not all Ironman courses are equally difficult, and the time of one course could be much longer than another. I had to use a proprietary formula to “normalize” any distance that was not a conventional distance. Although there is a good sample size of several hundred race times, the margin of error is arguably still too great to be valid. These are all confessed flaws in this study. By the way, all the statisticians listening to the show just fainted. They couldn’t take all the flaws in the data collection process.

So at best this is an interesting exercise that possibly creates more questions than answers. At worst it is a cheap ploy to generate more buzz around the show and gain more listeners. There may be some truth to that.

Therefore, we can’t say for example, athlete’s who had an income of x dollars were faster. We can say that athletes with a reported income of x were reported to be so fast. Or athletes who declared a height of x had a reported speed of y. In short, this is not really a study, it is exactly what it says it is. It is a survey. But a survey that has some fascinating results. To be fair, this is not really data on triathletes, it is really information about triathletes.

Before I share the results, I have to give you a brief overview of statistical correlation, because that is the way I sliced this information, and you will need to know about it to understand the data. Stay with me here, I’m going somewhere with this! It’s worth the wait! This part will take less than 3 minutes.

 

A correlation coefficient is used to describe the direction and the degree or strength of the linear association between data sets. A correlation coefficient can be positive or negative. For example, let’s say we take 365 days of average high temperature for a certain geography. Let’s say we also take the same 365 days and look at snow cone sales. As the temperature becomes warmer, the sale of snow cones increases. There is a positive correlation between warm weather and snow cone sales.

Items with a negative correlation can still be strongly correlated. Over that same 365 days, we could also find that as the temperature gets warmer, the sale of sweaters goes down. That is a negative correlation. Negative correlation does not mean no association, it just means that the linear relationship between the data sets is opposite, whereas with positive correlation the data sets increase together. There is also neutral correlation, where there is no association between data sets. For example, as the weather gets warmer, the sale of CDs from the singing group ABBA does not change. I do love ABBA. The correlation coefficient is always displayed as a number between –1.00 and +1.00, with numbers closer to the either end having a higher correlation.

From the correlation coefficient one can then calculate what is called the coefficient of determination, which is simply the square of the correlation coefficient. This number is displayed as a %, and is also referred to as the R-squared or the R2. And this is the value, the coefficient of determination, that I’ll be using to share with you how an athlete’s overall profile information affects their performance. What we will be doing is looking at what % of the variance in speed can be attributed to height, weight, age, income, experience or spending. Perhaps this will all make more sense if I just get started. Again, this represents the data from hundreds of athletes, which is an excellent sample size. Unfortunately, I had to make a judgment call, and decided that there were not enough samples of athletes at the Ironman distance for the sample size to be statistically significant. I feel really good about the Sprint, Olympic, and Half Ironman data, but there was just not enough Ironman entries for it to be included in this analysis. But stay tuned, I have a way to fix that later in the show.

I’m going to give these numbers to you in no particular order, but in an order that I found interesting. Let’s look at years of experience in triathlon, vs. years of experience in overall endurance sports. At the sprint distance level, the coefficient of determination, or the R-squared between sprint-distance performance and years of tri experience was 8.64%. Meaning 8.64% of the variation in speed at sprint distance could be attributed to how much experience you had in triathlon. While only 6.84% of the variation in sprint-distance performance could be attributed to your overall endurance experience. What this means is that if you are training for a sprint-distance race, you are more likely to perform better if you have triathlon-specific experience than if you have other endurance sport experience. No real surprise. The pattern is the same for Olympic-distance racing. While 10.96% of the variation in Olympic-distance performance could be attributed to your overall triathlon experience, only 5.53% was based on endurance experience. So the gap between triathlon vs. overall endurance experience at the Olympic-distance level is even greater. The take-home point is that if you are feeling like you are slow at these distances, just stick with it for a couple of years and you’ll get faster.

What about the half Ironman distance? While 2.24% of the of the variation in half Ironman distance performance could be attributed to your overall triathlon experience, 3.25% was attributed to your overall endurance experience. Meaning at that longer distance, the experience type is reversed. A longer base of endurance experience is more influential than specific triathlon experience at the half Ironman level.

So, given 2 similar Olympic-distance athletes, one with 3 years triathlon experience and 3 years endurance experience, would likely outperform the athlete with 5 years endurance experience and 2 years triathlon experience. But at the half Ironman distance, the roles reverse, although not separated by much.

Let’s move onto another comparison. The catalyst for this whole project was the ability for me to find out the correlation between height and race performance. Those of you who have been with me for a long time know that I have issues with our current competitive division system. Particularly that we have a division based on weight but not one based on height. This survey and analysis was to be foundation of this argument, and to prove that a weight-based system was unfair to athletes with other physical limiters.

It turns out I was wrong. But, the good news is I was only half wrong.

At the Sprint distance level, 6.76% of the variation in performance could be attributed to weight. Only 0.13% of the performance variation could be attributed to height, which really is 0. At the Olympic distance it becomes even greater, when the R-squared was a whopping 11.84% for height, and only 0.25% for height. Based on these surveys, at least for these several hundred athletes, at the Sprint and Olympic-distance level, height showed almost no correlation, and weight at the Olympic distance level was one of the highest correlations to race performance in the analysis. Again, these conclusions cannot be made universally, but it does mean that of these several hundred self-reported surveys, height played no factor in Sprint and Olympic distance racing, and weight had a small impact.

But, at the half Ironman distance, things change. At that longer distance, only 2.66% of the variation in performance could be attributed to weight, while 3.94% could be attributed to height. This is very consistent with my findings when looking at Ironman race results, where I did spent quite a bit of time analyzing races at that level, and found less than a 5% difference in the performance between Clydesdales and non-Clydesdales, in fact the Clydesdales as a whole were sometimes faster in the swim portion of the Ironman event than their non-Clydesdale competitors as a whole.

Based on these surveys and my own research, it seems that weight becomes less of a factor at the longer distances. I think that part of this can be attributed to the fact that the amount of Clydesdales competing at the IM and half IM distances is much lower than the amount of Clydesdales competing at Sprint or Olympic distances. But, it still does not change the fact that given this same number of half IM athletes, height was a bigger factor than weight at that distance. I hope to be able to get some more surveys back from athletes who have competed at the Ironman level, and we can add that fourth distance to the list to compare.

Let’s take a look at age real quick. It turns out it was actually one of the smallest factors. Only 0.15% at the half IM, 2.21 at the Olympic, and 1.55% at the sprint. However, the typical Tri Talk listener who submitted this survey tended to be a bit younger. Although I got a great cross section of surveys for the other factors, there was a much narrower range of ages, and that would have effected the results. It is interesting to note that although the age of a half IM had the lowest coefficient of determination, it had a negative correlation. Meaning, as the age went up, the race times went down, or the athlete was faster. Whereas with the Olympic and Sprint, as the age went up, the race times went up. Again, with a very small delta, the survey responses indicated that at the long distances, you might actually get faster as you get older, up to a certain point of course.

What I think is the most fascinating part of this survey is the financials. Two financial questions were asked: the athletes’ income and overall triathlon spending. Of all of the 7 factors I looked at: height, income, spending, tri experience, endurance experience, weight and age, the lowest correlation to performance was – income. That’s right, at all distances how much the athlete made had the least impact to performance. With results at 0.87%, 1.03%, and 0.68% for half IM, Olympic and Sprint distances, respectively.

But listen to this. The most influential factor to performance, practically across the board was how much the athlete spent, with results at 5.51% 16.32%, and 8.87% for half IM, Olympic and Sprint distances.

Now, before you freak out, don’t draw any conclusions on this yet. When I shared these results with my wife, she instinctively felt I should not share them. She was worried that it would discourage athletes who can’t afford to spend a lot on the sport, and thought I had a responsibility to research it more before I shared the information. She felt that it would taint the sport. I don’t agree. That same day, at dinner with my brother and sister-in-law, my sister-in-law pointed out something brilliant but obvious. The amount of money spent is likely a parallel to the athlete’s commitment. She didn’t see performance linked to money. She saw performance linked to commitment. Those that spend more, are likely to be more committed in terms of training hours, diet, coaching, research and planning. Especially when you couple that with the fact that having a high income had practically no effect on performance, it was how much you spent, or as my sister-on-law pointed out, how committed the athlete was.

The only conclusion that can be drawn here is this. I’m an idiot. I’m an idiot because when I first created the survey, I should have seen this issue. So, to take this further I have edited the survey to ask 3 more questions. 1) how many hours of training were you performing prior to the event 2) on a scale of 1-10, how committed would you rate yourself to performing well at this event and 3) what your placement was in your competitive division, for example 20th of 50 in your age group division. The first 2 additional questions were added to introduce both a numeric and subjective way to measure commitment, and the 3rd question was added to correct the problem that the race conditions and topography in each race will effect the race time, even for the same distance. By looking at both race times and what % the athlete finished in their age group, I can get a different relative picture of performance.

Now, this is the part where I need your help. People frequently ask me what they can do to help Tri Talk. What I am asking you to do for me in return for this free resource in is to go to the website, and take this survey. I don’t want hundreds of surveys to analyze, I want thousands of surveys to look at. We have the listener base to have up to 14,000 surveys returned. Would that be an incredible sample size or what? The survey system I use also gives me stats for how long the survey took each user. 80% of all of the surveys taken took under 10 minutes to complete. With these 3 additional questions, it will probably move that figure to 11 or 12 minutes. So go there right now to tri-talk.com and click on research.

To those of you who have already filled out the survey, thanks! You’re reward for your altruistic contribution and loyalty to the show has resulted in me asking for – more work!  I made several changes to the survey, in addition to that 3 new questions, and I hate to ask, but if you have already done the survey, can you do it for me one last time? This is the ultimate version of the survey, and you won’t have to do it again.

By the way, I’ll have a table on the Tri Talk blog at this point in the transcript that shows the results in a format that is more easily assimilated than via audio, and you can browse the survey results at your leisure.

This next topic was inspired by 2 of the athletes I coach. Both of them had recently done a lactate threshold HR test protocol to determine their new lactate threshold HR (LTHR). It’s a good idea to test this every 4-8 weeks as it will often change for each sport as your fitness levels increase, and if you use your HR for training or racing, this is critical information.

Both of these athletes express a concern that their LTHR had not changed, and were disappointed with the results. They were worried that they were not increasing their fitness because their HR had not changed. I quickly pointed out that in both cases, the speed at which they had completed the 30-minute protocol had significantly improved, and that was the ultimate indication of fitness improvement. Ultimately, we don’t train just to improve out LTHR, or VO2max, or respiratory exchange rate. We train to get faster, no matter what those indicators of fitness say.

Still, the question is a good one. Why did the performance significantly change, but the HR data did not? The first answer would be environmental conditions, which often have a significant effect on HR, even when working at the same speed or power. I’ve written and podcast extensively on that subject, so we won’t review it again here. The second reason would be fatigue or the amount of recovery before the test. Performing the same test in the same environmental conditions, but on fresh legs would significantly change the HR compared to fatigued legs. Assuming that in this case, the environmental conditions and recovery were the same for both tests, why then would the HR have not changed if the performance did?

Although there could be multiple reasons, including improved economy which was not mentioned, here is what I think happened in this case. There is often confusion in the endurance world that a sign of fitness is a high max heart rate. That is incorrect. It is a sign of fitness to be able to maintain a high % of your max heart rate. Someone with a max heat rate of 200, and who can only hold 87% of that, or 176 beats, for 30 minutes is generally less fit than someone who has a max HR of only 174, but who can maintain 90%, or 157 beats, for that same 30 minutes. Yes, the ability to hold a high heart rate is good, but it is all relative to the percentage of the individual’s max heart rate.

This is supported by the fact that as an athlete begins to train more, his max heart rate begins to come down, not up. According to Wilmore and Costil’s Physiology of Sport and Exercise, a sedentary male with a max HR of 185 will drop his max HR by 2 beats after only minimal training. And, in the same chart in that book, it lists a typical world-class will have a max HR of 174.

This does not necessarily mean that if you have a naturally high HR you are out of luck. With the genetic lot you have been given, you goal should be able to increase the % of that max heart rate that you can maintain, regardless of what your max HR is.

In the case of my athletes, even if their LTHR did not change, if their max HR went down by 2 beats in between the 4-8 weeks, and their LTHR did not change, they were able to increase the % of their max HR that they could maintain, since that max HR theoretically went down.

Now, I bet you might be saying, “This doesn’t make any sense. If my heart beats faster, doesn’t that mean I’m pumping more blood and delivering more oxygen, therefore a higher HR should be good.”

Actually, no. What matters to the heart during exercise is not total beats, but volume of blood passed per beat. If your HR is too high, not as much volume per beat is passed to the muscles.

To demonstrate this, think of your heart pumping blood the same way that your lungs process air. If I were to ask you to inhale and exhale the most amount of air possible in 15 seconds, how would you do it? Would you take 45 breaths as fast as possible, or 10 very deep breaths? You would find that your total volume of air would be highest at the slower, deeper rate. Or, consider when you run, you don’t fall into an extremely high respiratory rate. You take long, steady deep breaths. Your heart works the same way, which is why elite athletes actually exhibit a lower max HR, because it is a more efficient way for the heart to deliver the maximum amount of blood at the lowest cost.

The pitfall I see athletes fall in to is basing their race pace on a HR from the same event the last year, or even from earlier in the same year. For example, if you ran a marathon and averaged 164 beats for the event, you might come into the same event the next year thinking that you should be able to do that again. But if you have been training well, and your max HR is down by several beats this second year, you could be redlining by trying to maintain a much higher % of your max HR than you are ready for.

Don’t get me wrong. The increase of average HR in a LTHR test is always a good sign. Whether your max HR has stayed the same or not, an increase in the average HR over a fixed time means you are now able to maintain a higher % of your max HR. that is great! But, if you find that your HR is staying flat after significant training, that does not mean that you are not making progress. If your times are improving, it is likely that your max HR is coming down, and that is also good.

The take-home point is this. Test yourself often under as similar conditions as possible. Don’t worry about what the actual number is, but use that number in your workouts until you test yourself again. The ultimate feedback from a test like the LTHR test is improvement of the speed or power produced, not the HR number itself. Often perceived effort has to trump the HR monitor when things just don’t add up.

By the way, this discussion is totally different for resting heart rate, and we’ll talk about that another time.

That’s all for this episode, I’ll be back the middle of next month for the next episode. There is one person who deserves a shout-out on the show who has never gotten one. To my wife, Rebecca, thank you for all of your support in my crazy training and business endeavors. I love you, Bec! See you next time.