Training Intensity Distribution
The relative amount of exercise you do at different intensities is called your training intensity distribution (TID). If you spend 75% of your time running easy, 15% running somewhat hard, and 10% running very hard then you have a 75/15/10 TID. There's a lot of interest in identifying the optimum TID for each sport and each athlete. The optimal TID will depend upon many factors including genetics, training history, capacity for recovery and adaptation, nutrition, event demands, and timing in the training progression. When we look at the practices of successful endurance athletes, there are some patterns that emerge.
Elucidating and curating the training of top athletes provides useful insight into the training that generated sufficient adaptations for those individuals to be successful. This approach, however, when espoused as the likely best approach for unsuccessful or pre-successful athletes is an application of survivorship bias; it ignores all of the people who trained that way and were not successful. It also fails to capture the likelihood that underlying biomechanics, physical build, and fiber type, among others, all layer into the response equation. What helps athletes who are predominantly slow twitch, which we can assume of all elite endurance athletes, may not be the best approach for amateurs who may have a higher percentage of fast twitch fibers and no chance of reaching the world stage, but who do wish to achieve their personal best. For example, someone with a sprinter phenotype is not necessarily going to optimize their endurance potential by training like someone with a marathon phenotype. Nevertheless, it is useful to understand the training intensity distributions of top athletes because, when patterns emerge, these are likely to serve as actionable starting points for athletes to adopt if those athletes have similar potential.
We begin with the foundation of endurance training, and that is volume. You will not become a great ultra-distance runner by running 30-40 miles per week. You might get away with 30-40 miles per week sometimes, but, at some point in your career, you're going to have to put in quite a bit more volume if you're going to really get good at ultra-marathons. And, if you're putting down 80-mile weeks, for example, you can't tolerate running most of those miles at your 10K race pace; that's not sustainable. I laid the groundwork for this thought experiment in recent episodes dedicated to discussing mileage and volume. We begin our exploration of TID, then, with the understanding that we have to run significant mileage, and most of this mileage has to be at easy or moderate effort. Running small mileage isn't enough and running a lot of miles very fast isn't sustainable. This obvious observation results in a TID for ultra-marathon runners where the majority of the training is easy to moderate intensity.
What do we do with the rest? One option is to do more of the same; make all our training easy or moderate. A lot of people will do pretty well on that program. This approach will give you most of the endurance capacity you need to complete ultra-marathons. Over time, it will even help you to get a little bit faster at running them. But, the evidence shows that athletes benefit from adding some faster or high-intensity running into their program. And, while some high-intensity running is beneficial, doing more is not better and may even be detrimental, especially if it takes time away from more running at lower intensities.
In the laboratory we can measure two transitions in metabolism that occur as we progressively increase output - like running faster and/or greater inclines. One of these transitions occurs at the transition from what bioenergeticists call the moderate domain to the heavy domain. It can be identified by a rise in blood lactate from resting levels to higher values and occurs roughly at 2 mmol blood lactate on average. It can also be identified as an increase in exhaled carbon dioxide relative to oxygen consumption or as a rise in oxygen consumption above the extrapolated linearity apparent in lower work rates. That last method is a little wordy and heavily scientific but it's actually the one I like and use most. To put it into easier to understand terms, it's where you begin consuming more oxygen per unit of work output than you did at lower efforts, which is why I've dubbed it the economy threshold. The higher transition is the one above which a relative steady state is not achievable and is synonymous with critical power. We can call this the fatigue threshold because fatigue arrives at a highly predictable moment for any higher pace if some of the physiologic capacities have been properly measured for that athlete. These transitions define the bounds of three bioenergetic domains called moderate, heavy, and severe. Exercise is sustainable for many hours in the moderate domain, some hours to less than an hour in the heavy domain, and less than 30 minutes if we're clearly in the severe domain.
It becomes tricky to equate these domains with race distances because sustainable paces don't line up perfectly with the underlying measures of these bioenergetics across the entire spectrum of athletes from novice to elite. Said another way, a recreational runner or short-distance specialist may finish a marathon with high lactate levels while an elite may run the entire race and never get much above 2 mmol/L despite running much faster and at a higher percentage of their maximal aerobic capacity. But, as a very rough guide, and averaging heavily, the first transition may be roughly at marathon race pace and the second at roughly 10K race pace. I give these not as guides for you to try to implement but only as a way of helping to map what we're talking about onto tangible metrics you'll be more familiar with. So, roughly, the moderate domain is below marathon pace, heavy is marathon to 10K, and severe is faster than 10K. Again, any individual may differ significantly from these estimates, which become more accurate in a population-average, but it gives you something rough to anchor on.
This recap of intensity domains is useful because these are the three domains that have been most often used when quantifying TIDs. If we accept that most of our training will be in the moderate domain, the question is how much of the rest of our training will be distributed in the heavy and severe domains. The possibilities are:
a) running most of it in the heavy domain and less in the severe, something called pyramidal training (PYR) because the graph of the PYR TID slopes like one side of a pyramid, or
b) running a little or not at all in the heavy domain with most or all of it in the severe, called polarized training (POL) where the graph looks like an inverted-J.
There is one other major type of TID worth mentioning. It's called threshold training (THR). This is where the largest amount of running is in the heavy domain with less in moderate and severe domains; THR TID graphs look like an upside-down V or U. It may be useful for those running marathons or half-marathons for some peaking weeks but THR training is not really sustainable because running a lot of mileage at THR puts you at high risk for eventual over-reaching.
You've probably heard one of those patterns talked about a lot because it's been popular lately, POL training. One of my goals today is to show you why the evidence for POL training has been misunderstood, it's not how most top endurance athletes train, and it's not how most people should be training most of the time.
The under-appreciated factor in early papers that popularized a POL TID is that the studies quantified training by a "session goal approach" (Seiler, 2010) and not by training time or distance. In these studies, each training session counted once, regardless of the duration or distance for the session. Each workout was assigned a training intensity for the entire session based on the main goal-intensity within the session. For example, a 2-hour easy run counts as 1 moderate-domain session and a 1-hour run with 30 minutes of intervals at half-marathon pace was counted as 1 heavy-domain session. A track workout with 15 minutes of 400-meter fast repeats would count as 1 severe domain session. You can see immediately that the "session goal approach" tells you the number of sessions aimed at hitting intensities in the three domains but it doesn't tell you about the amount of training time, or distance, that athletes actually spend in each intensity.
One of the first and most cited early studies to do this was by Seiler and Kjerland in 2006, which I call the study that launched a thousand misunderstandings. When you read the abstract of that paper, all the authors tell you is that the intensity distribution of training sessions was 75, 8, 17%, which means that 75% of sessions were kept moderate, 8% had heavy domain bouts, and 17% of sessions included severe-domain intervals. That word 'sessions' is the key and it's the word that has gone unappreciated by most of what I see and hear in the general running public when POL TID is recommended. When you read the paper, you have to dig deep into the results, and then you discover the actual TID of those athletes was 91, 6.4, and 2.6% by time (using heart rate records). Those are massive differences: 75/8/17 by sessions but 91/6/3 by time. If you, like so many, simply gloss over that word 'sessions' in the abstract, you come away thinking that you're supposed to spend about 20% or so of your time in the heavy or severe domains with most of that above 10K pace when, in fact, the athletes in that study spent only 6.4 and 2.6 % of their time (that's 9% total) at heavy and severe intensities, respectively. They spent less than 3% of their time running fast, not 17%.
As we explore more studies, we find another failing in communicating the science to the general public. Upon closer analysis, the majority of what has been counted as severe-domain training was around 90% of maximum heart rate (88-92%), which is approximately the heavy-to-severe border, approximately critical power or the fatigue threshold. It's a misapplication to use these findings to recommend severe-domain training because most listeners of that message will run their faster workouts too fast - like 5K pace - rather than near 10K pace, which is more consistent with the research.
If you want only the take-home message, it's this. Based on 15+ years of research studies, the TID that seems to produce great endurance performances, based on time or distance, is a PYR approach, where the clear majority of running (often more than 85 or 90%) is moderate - slower than marathon pace - most of the rest is in the heavy domain - roughly up to 10K pace - and only a few percent is above 10K pace. If you graph your time or distance spent at all intensities over a week or more of training, by heart rate or power for example, this will look like a line or curve where more time is spent at any lower value when compared to any higher value.
Now, we'll explore the nuances and studies to date, beginning with studies of runners, then moving on to other endurance disciplines, and finally on to reviews and meta-analyses.
In 1999, Billat et al. studied 8 subjects who performed 4 weeks of normal training with one session per week doing some intervals at the lowest velocity that would elicit V̇O2max (that's just barely into the severe domain). They followed this training with 4 weeks of overload training where they did the interval session three times per week. The conclusion was that intensification of training did not improve factors associated with performance. In other words, one hard session per week was enough to maximize gains.
In 2001, the same group reported that top-class marathon runners ran ~78% of their weekly kilometers slower than marathon pace, ~12% at marathon to half-marathon pace, 6% at 10K pace, and just 4% at 3K pace. This is a PYR TID. I've actually seen this study cited as evidence supporting POL training by those who promote session-based TID. Why? Because if you don't pay attention to the time or distance spent training and only bin each workout based on the top speeds of each session, then there are more sessions with very fast running than with just somewhat fast running. But, as I've just given you the numbers, you can see that these athletes run less at faster paces than slower paces.
In 2005, Esteve-Lanao and colleagues showed that regional- and national-class runners' training was PYR, with a 71, 21, 8% distribution of their training time. Two years later, the same group reported that increasing the amount of time spent in the middle - the heavy domain - from a TID of 81, 12, 8% to 67, 25, 8%, did not enhance performance.
In 2012, Stellingwerff published case studies of 3 elite Canadian international marathon runners over a 16-week period before a marathon race. Based on subjective ratings of perceived exertion, the distribution of sessions was 74, 11, 15%, which is POL, but the actual time spent in each domain reveals a PYR TID. This difference is because athletes report perceived exertion for a session with a bias to the harder parts of the session and we get a skewing of outcomes when we count each session as one intensity anchored on the hardest parts.
In 2013 and 2014, Tjelta reported a PYR TID for top runners in two studies across all time points of training, which spanned up to 2 years.
In 2014, Muñoz et al. studied 30 recreational runners who performed either POL (77, 3, 20%) or THR (46, 35, 19%) training for 10 weeks. Both groups improved 10K race time, and the improvement was not statistically different between the two groups though the authors say that the POL TID was more beneficial (statistically) when they excluded the POL group athletes who did the least amount of training. Unfortunately, there was a large difference in the amount of moderate-domain training between the two groups; i.e., the THR group wasn't PYR. I include mention of this study because it may well be that the TID matters less when the runners are less well-trained to start, and that's important to consider if you are relatively new to running or otherwise would consider yourself recreationally trained.
In 2015, Manzi and colleagues reported the TID of 7 recreational marathon runners, which was 76, 17, and 6%. This is PYR training, and what I found more interesting is that they stated, "There was a significant correlation between total time spent in zone 1 (meaning the moderate domain) and the improvement at the running speed of 2 mmol/L" lactate. This is important for ultra-marathon runners because ultras are run at an average output that is roughly this intensity. They found that the amount of time spent running slower than that intensity was significantly correlated with improving speed at it. There is a common misconception that I regularly hear. It's this notion that to enhance a physiological capacity, ability, or measure, you have to train harder than it and accumulate time at or above it. There's no evidence to support that notion. It may be that you can enhance some capacities or measures by training harder than them, but there's no evidence that doing so is required for improvement.
I'd like to further illustrate this point using an example of just one factor that supports enhanced endurance capacity, capillary density, which is the number of capillaries (tiniest blood vessels) in contact with each muscle fiber or within a voxel of muscle tissue cross-section. Given the architecture of microvascular control relative to motor unit dispersion, it is predictable that easy exercise generates sufficient shear stress to increase capillarity even to fast twitch fibers despite fast twitch fibers being recruited less at lower intensities. Specifically, arterioles control flow into capillary beds that serve fibers from multiple motor units, which includes adjacent slow and fast muscle cells. One need only recruit slow fibers sufficiently to generate arteriolar dilation to increase shear stresses in the capillary beds that also serve fast fibers, and thereby create a stimulus for angiogenesis to fibers that may not even be recruited. Because the shear-stress signal may have a minimal duration effect (a need to be 'on' for some period of time before new capillaries will result) short, high intensity exercise is less likely to add capillaries than longer total durations of low intensity exercise. This is a long way of explaining how longer duration of moderate running can improve blood flow delivery capacity even to those fibers that are relatively less aerobic and thereby improve support for long endurance performances where those fibers become necessary late in a race. This is one reason why high-intensity interval training cannot substitute for long runs at moderate paces and why a larger volume of your training for endurance performance should be at easy-to-moderate efforts.
In 2017, Esteve-Lanao and colleagues published another study, this time comparing the training of marathoners and full-distance triathletes. They found that marathon runners trained using a PYR approach of about 75, 16, and 9% of time in the three domains.
In 2019, Pérez et al. studied the effect of two different TID programs on measures in ultra-endurance runners. Twenty recreational ultra-endurance runners trained for 12 weeks in two groups: POL (80, 4, 16%) and what they called THR but is actually PYR (67, 34, 0%). Though there were some differences in laboratory measures, it turns out that the two groups were not actually of equal abilities or capacities at the start of the study. Why do I mention a study if it was so badly done as to not even have comparable groups? Because the training was in advance of a 93K ultra-marathon, at which neither group performed better than the other. This may be another case of almost any training being equal to any other in recreational athletes or it may reflect the realities that there are many other factors that have a large impact on performance in ultras beyond the TID. Said another way, once we're consistent with our training and do enough overall volume with most of it at easy-to-moderate intensity, the distribution of remaining percentages between a bit above or below, roughly, 10K pace may not have a significant impact on ultra-marathon performance, at least for recreational athletes.
In 2020, Festa and colleagues studied 38 recreational runners who trained for 8 weeks in a POL or THR program (77, 3, 20% or 40, 50, 10%). Both groups improved vVO2max, RE, VT, RCT, and 2K running time. No differences were found between the groups on any parameter investigated but it's important to note that the THR group trained 17% less time. If you're unfit and you have limited time, it might be just as useful to split your time between easy/moderate and moderate/somewhat-hard intensities, with just a bit of fast-for-you running according to this study.
Also in 2020, Kenneally et al. published a TID analysis of 7 world-class middle-distance (800-1500 m) and long-distance (5,000 m–10,000 m) runners over 50 weeks using 2 different approaches to organize TID zones: (1) based on individual specific race pace and (2) based on physiological parameters. Training volumes for the 3 domains (rounded) were 89, 7, 4% for race-pace based approach, and 87, 6, 7% for the physiological approach. They wrote "The approach based on race-pace zones produced pyramidal distributions in both middle- and long-distance runners across all phases of the season. The physiological approach produced polarized and pyramidal distributions depending on the phase of the season in the middle-distance runners, and pyramidal type TID across all phases of the season in the long-distance runners." Specifically, in the short 'competitive' microcycles of the year, middle-distance runners adopted a slightly POL TID (by less than 1%). Otherwise, all training was PYR and all training at all times was PYR for the long-distance runners.
It's routinely reported that the fastest athletes run the most mileage. From the 2001 paper by Billat to a more recent big data analysis in Nature Communications, faster athletes run more per week than slower runners. The differences seem to be caused by something more than just the obvious fact that faster runners can cover more ground than slower runners in the same amount of time. And it's reasonable to question whether this trend manifests from other aspects of physiology. For example, the best endurance athletes will tend to be more slow-twitch and they can likely handle lots of low intensity mileage. They may even need it for optimal training stimulus.
The effect of different TID programs aimed at enhancing endurance capacities is also complicated by the starting glycogen levels. The external stress may be the same for two identical runs but the internal strain, and its impact on adaptive response (e.g., mitochondrial adaptations) may differ depending upon the finishing glycogen concentration. This was the main topic we covered in episode 119 with Drs. Morton and Louis titled Finish Low.
Elite and pro runners - who train for distances up to the marathon - train up to 13 times per week, with two sessions on most days of the week. They have sessions that are essentially all intervals. For example, they don't run for 90 minutes with 4 x 5 min hard in the middle. Instead, they'll do the interval session and then they'll run easy miles in a different session. Most amateur athletes run only once per day and often blend harder efforts within the context of running slower miles. The total time or distance in each intensity zone is what we should be looking at, not the number of sessions binned by the highest intensity of those sessions. It takes two weeks for the recreational athlete to get in as many training sessions as the elite or professional athlete does in one week. Thus, the 'session goal approach' used by Seiler and colleagues in 2006 (and since), which has formed the foundation of the current craze in polarized training, may not be relevant for the once-per-day athlete. Moreover, the reality that POL training approaches are only supported when using the session goal approach, has gone under-appreciated. Even the highly referenced 2006 study by Seiler and Kjerland found a non-POL time distribution.
There are at least a dozen studies in sports other than running.
In 1999, Lucia et al. reported a pyramidal TID (70/23/7%) during the Tour de France based on the heart rate time-in-zone method over 22 competition days.
Five years later, Fiskerstrand and Seiler published data from a survey of Norwegian International-level rowers across three decades, 1970-2001. The data clearly shows a PYR TID in all decades, except in the 1970s where there was more overspeed training but still not a POL TID.
In 2009, a study of world-class rowers reported a 95, 2, 3% TID...which is mathematically POL but the actionable message is nearly all training, 95%, easy or moderate and about 5% distributed among higher outputs.
In 2013, Neal et al. found that 6 weeks of POL training was more effective than a THR TID in 12 well-trained cyclists using a randomized, cross-over design where athletes go through both programs and, therefore, serve as their own statistical references. There was no PYR group.
In 2014, several studies shed more light on the topic of TID. Orie et al. published results from an analysis of 38 years of TID in Olympic speed skaters. The TID of successful male Dutch Olympic speed skaters in four Olympic seasons (1972–2010; assessed by interviewing the coaches and athletes) was based on THR in 1972 (40/40/20%) but transitioned to a PYR approach by 2010 (~80/~12/~8%).
A THR TID approach was also supported by a study in 2019 by Selles-Perez et al., who reported that training time in zone 2 was related with better performance on a half-Ironman race in amateur triathletes.
So, while we might begin to theorize that a THR TID can be useful in recreational athletes and possibly even more effective than a POL approach in some well-trained athletes, top performers have transitioned to a PYR approach in more recent years.
Muñoz et al. (2014) quantified the training distribution of nine recreational Ironman triathletes; they found a greater percentage of training done in the middle zone was strongly correlated with worse performance while more training at a low intensity was correlated with better Ironman performance.
Clearly, not all studies will agree, which is likely due to more subtle factors in the study design. On aggregate, however, it is a consistent finding that more time spent at easy and moderate intensities consistently correlate with the best performances. Moreover, programs with 90-95% of training time in those zones are practiced most of the year by most of the best endurance athletes across numerous disciplines.
Stöggl and Sperlich (2014) studied the training of 48 healthy competitive endurance athletes from Austrian national teams who participated in cross-country skiing, cycling, triathlon, or running. Unfortunately, this time, they did not provide time or distance in zone data but only the session-maximum, which, unsurprisingly, was polarized. The study methods stated that athletes wore heart-rate monitors throughout training sessions, so they should have all the data to report time or distance at various intensities, to the extent that heart rate reflects relative intensity. So, I emailed Dr. Stöggl. He told me that they did record the data but did not store it and could not make those determinations though it was likely the athletes did train in a PYR distribution by time or distance. He followed up by confirming that the discrepancy in TIDs by the two methods is, as he put it, "huge", and that the majority of research demonstrates that top and successful endurance athletes by-and-large employ a PYR TID, not POL, when assessed by time or distance of training at intensity.
In 2017, Solli et al. published an interesting study of the training characteristics of the world's most successful female cross-country skier, Marit Bjørgen. Her training was >90% easy-to-moderate. The remainder was PYR early and only became more severe in the final preparation phase before competition, where severe intensity bursts would be required. Indeed, when we see a POL TID in endurance athletes’ regimens based on time or distance, it seems to be only in the final preparation phases before competitions and where there will be a requirement for those brief but very high intensity bursts. For example, in 2020, Myakinchenko et al. found that international-level cross-country skiers and biathletes increased the percentage of time at race pace and intensities as they neared competitions but that "all teams employed the pyramid model of intensity distribution."
Also in 2020, Rohrken et al. conducted an intervention study in 15 moderately trained triathletes assigned to POL or a MOD TID (which had no severe intensity training), concluding "polarized TID in moderately trained triathletes did not prove to be superior compared to a more moderate TID."
My overall take from all of this research is that a) polarized training is uncommon and only employed for brief periods and in specific phases for specific events, b) most amateur athletes will benefit similarly from any TID as long as total volume is sufficient and the emphasis is placed on easy-to-moderate intensities, and c) some time or distance spent at efforts at half-marathon pace or faster are likely beneficial for ultra-marathon runners but need not be more than about 5-8% of training time.
I've reviewed studies on running and studies on non-running sports. The next and last section is of review papers that synthesize multiple studies.
The first was in 2008 by Seiler and Tønnessen and, along with another review by Seiler in 2010, explains that, across endurance sports, successful top athletes tend to have a pyramidal training distribution by time or distance but have a polarized distribution in terms of training sessions. By this point, you understand this to be true. But, even then, there is an especially important wrinkle to the fabric of this story; the severe domain intervals are not all-out but rather low zone 4 of a 5 zone model - or just barely into Z3 of a 3-zone model. Again, this is somewhere close to 10K pace, on average. I think most runners have gotten the idea that they should be running their hard sessions or intervals harder than is true. The consensus view from the data is that extremely little running needs to be much faster than 10K race pace when training for ultra-marathons.
In 2018, Kenneally et al. published a paper titled, The Effect Of Periodization And Training Intensity Distribution On Middle- And Long-Distance Running Performance: A Systematic Review. They wrote, "According to the results of this analysis, pyramidal and polarized training are more effective than threshold training, although the latest is used by some of the best marathon runners in the world. Despite this apparent contradictory finding, this review presents evidence for the organization of training into zones based on a percentage of goal race pace, which allows for different periodization types to be compatible." The conclusions are consistent with the observations that top athletes may adopt a POL TID as they get into competition phase and when the competitions contain high intensity output requirements (like rowing that may only last 7 minutes or XC skiing that has short uphills with downhills that are effectively high intensity interval performances). Applying the consensus observations of this systematic review to training for ultra-marathons indicates a PYR distribution is recommended.
In 2019, Rosenblat et al. published Polarized vs. Threshold Training Intensity Distribution On Endurance Sport Performance: A Systematic Review And Meta-Analysis Of Randomized Controlled Trials, where they wrote, "The pooled results demonstrate a moderate effect favoring the POL group over the THR group. These results suggest that POL may lead to a greater improvement in endurance sport performance than THR." Only 3-4 studies made it into quantitative and qualitative analyses and PYR TID was not part of this review.
Finally, in 2019, a very thoughtful narrative by Bourgois et al. titled Perspectives And Determinants For Training-Intensity Distribution In Elite Endurance Athletes includes an evolutionary perspective that suggests humans are likely most sensitive to training that mimics how we lived during our evolution as hunter-gatherers, which is mostly easy with some brief spurts of high intensity and little or no middle. They write, "we speculate that type II-oriented endurance athletes (eg, 65% type 1 and 35% type II fibers, and thus more anaerobically oriented) will have a faster force development pattern and a higher movement velocity as compared with type I-oriented endurance athletes (e.g., 85% type I and 15% type II fibers, and thus aerobically oriented) and will induce more adaptive benefits from PYR. On the other hand, type I-oriented endurance athletes might benefit more from POL. Since a higher volume of ThT in PYR (vs POL) can have a negative impact on the autonomous nervous system as well as on glycogen homeostasis as compared with POL, we also assume that total training volume in POL, used by type I-oriented endurance athletes, can be higher than in PYR (used by type II oriented athletes). With regard to the biological and training age of an individual, which will also have an impact on the planning and periodization strategy, the lower total training volume of young athletes will allow a higher contribution of Z2 and Z3. As athletes develop toward elite level, the training volume will increase, and thus, the contribution of Z1 will be more pronounced…Up to now, we can only speculate that a large volume of LIT (> 70%) combined with a low proportion of ThT and HIT (POL or PYR) is paramount."
It's worth wrapping up by exploring a 2015 review. In 2015, Stöggl and Sperlich published The Training Intensity Distribution Among Well-trained And Elite Endurance Athletes. This is the same Stöggl who previously published with Seiler, who reports session-goal rather than time-at-intensity, and now we get to see Stöggl's take on the field without that constraint. The findings of this review indicate that elite endurance athletes spend a high percentage of their TID in a pyramid shape—that is, great portions of HVLIT with 84–95% in zone 1 (moderate), 2–11% in zone 2 (heavy), and 2–9% in zone 3 (severe). Depending on the competition calendar, the TID during the pre-competition phase may vary between endurance disciplines. The shifts were toward a larger emphasis on race paces, which resulted in PYR or POL depending on the events. They also wrote, "Overdoing THR by >20% through reducing HVLIT may exert a negative impact on the autonomic nervous system with no further adaptation. In fact, THR training places greater demands on carbohydrate fueling, leading to restricted training time due to limited glycogen storing. However, THR may be more applicable for untrained and/or recreational individuals" and, I'll add, for the race preparation phase if the performance is at THR paces, such as in the marathon. So, I think this is a useful assessment and summary. Nearly all of your training should be easy-to-moderate with 0-15% of your time or distance spent running faster than marathon pace, with generally less of that at progressively faster paces - i.e., a PYR TID. You might spend 5-10 percent of that time in higher intensity training as you get closer to race-day only if you're doing a short event that will require those red-lining efforts.
I'll finish by repeating the take-home message. Based on 15+ years of research studies, the TID that seems to produce great endurance performances, based on time or distance, is a PYR approach, where the clear majority of running (often more than 85 or 90%) is moderate - slower than marathon pace - most of the rest is in the heavy domain - roughly up to 10K pace - and only a few percent is above 10K pace. If you graph your time or distance spent at all intensities over a week or more of training, by heart rate or power for example, this will look like a line or curve where more time is spent at any lower value when compared to any higher value.