In the most basic paradigm of quantification of runner training loads, coaches and athletes have long been monitoring volume -- typically weekly miles or kilometers, or total duration in h:m:s. If we’re going to be accurate, however, a runner’s training load is actually a function of BOTH volume and intensity. Without considering intensity in training load quantification, relying simply on volume can be misleading.
For example, the training load carried in an 80-mile week of running at easy to moderate sustainable paces (base miles) can be significantly different from an 80-mile week that includes substantial durations of tempo running and high-intensity interval training.
As another example, let’s say we have several 8-mile workouts: 1) a hard, hilly 8 miler at 6:45/mile pace, 2) an easy 8 miles on relatively flat terrain, and 3) a 1-mile warm-up, 6x1 mile at 6:00/mile pace on the track and a 1-mile cool-down. How do we assess the relative training load of each of these runs? To calculate the true training stress, we need to incorporate both volume and relative intensity into the quantification of training load.
Defining training load as a function of duration and relative intensity
This paradigm -- including both volume and relative intensity into training load quantification -- has evolved into practical use thanks to a number of scientists, following the introduction of the impulse-response model of Bannister et al in 1975. In 2003, based on cycling power meter data, Dr. Andy Coggan introduced Training Stress Score (TSS), wherein TSS = exercise duration x average power x an intensity-dependent weighting factor. Dr Coggan then presented The Science of the Performance Manager Chart in October of 2008. In turn, in December of 2008, Dr. Stephen McGregor presented running Training Stress Score (rTSS) based on normalized graded pace. Phil Skiba, DO introduced BikeScore, also based on cycling powermeter data, in 2008. Dr Skiba’s application of his models to running has received recent attention for their use in Nike’s Breaking2 project. The Stryd team rolled out RSS (Running Stress Score) in 2017 with their running power meter.
As we look forward, I am currently working on a model for running training load quantification based on duration, power, relative intensity, and variable “efficiency” in running. Perhaps even more enticing might be what Dr Coggan may be working on, suggested by his lecture in the 2016 Innovations in Training with Power for Endurance Sports Conference titled “Future View: The Evolution of Measuring Training Stress: Adaptation Scoring.”
Experiences with running Training Stress Score (rTSS) as a quantification of training load
I am most familiar with training load quantification using the WKO Performance Manager Chartbased on rTSS, so I will present my experiences using it as a method of monitoring (and acting upon) training load.
Here is the essence of rTSS, as presented by Dr McGregor (the image is from the linked article):
Distilled down, rTSS uses Dr Coggan’s model but substitutes Normalized Graded Pace for Normalized Power, and Functional Threshold Pace for Functional Threshold Power. It still uses duration and relative intensity in arriving at training load, or rTSS.
Next steps - representation of training load over time
Now that we have a way of quantifying the training load of a given workout as a function of duration and relative intensity, the next step is to represent the accumulation of that training load.
CTL - Chronic Training Load - 42-day period
Chronic Training Load (CTL) is the exponentially-weighted average of daily TSS from the last 42 days, figuratively representing the training load to which an athlete has adapted over the prior approximately six weeks. In other words, a runner’s CTL represents his/her relative running fitness.
ATL - Acute Training Load - 7-day period
Acute Training Load (ATL) is the exponentially-weighted average of daily TSS from the past seven days. ATL represents the most recent training load, new stimulus.
CTL and ATL can be charted in what is called a Performance Manager Chart (PMC). Below is one runner’s (Runner 1) PMC from January 1, 2017, through his final race of his outdoor track season on May 20, 2017. CTL is represented by the thicker, light blue line, and ATL is represented by the thinner, pink line. (More on the yellow bars and broken thin blue line later.) The yellow arrows mark the days on which this runner raced a 1600-meter race, and there were other races and distances through the season.
If an athlete’s ATL value is greater than his/her CTL value, it means two things: 1) the CTL will tend to grow, representing chronic training load progression, and 2) the athlete will be less fresh and more fatigued. On the far left of the chart above, we can see ATL exceeding CTL, and the slope of the CTL line is upward, meaning that the chronic training load is progressing. On the far right, we see the opposite: ATL is lower than CTL, resulting in a declining CTL, or a diminishing chronic training load.
TSB - Training Stress Balance
The relationship between chronic training load (CTL, or relative fitness/adaptation) and acute training load (ATL, or relatively new stimulus) is an important one to follow. The second point in the preceding paragraph (that the athlete will be less fresh and more fatigued if ATL is greater than CTL) can be quantified and depicted. Training Stress Balance (TSB) is computed by the simple formula CTL - ATL = TSB. More negative TSB equals more residual fatigue. More positive TSB equals greater relative freshness.
Looking at the chart above, TSB is depicted by the yellow bars. At the far left, where ATL is much higher than CTL, we can see the TSB bars well below zero, reflecting greater fatigue. On the far right, we can see the TSB bars in the positive range as ATL falls below CTL, reflecting greater freshness.
Here’s another example. Below is another runner’s (Runner 2) Performance Manager Chart over an approximately two-month-long build period. The blue line represents CTL, the pink line ATL, and the yellow bars TSB.
We can see that CTL builds gradually during this build period. We can also see in the ATL line the effect of harder days and easier days. TSB is negative each day, some days more than others, depending on how hard the prior day was; TSB for any given day represents the CTL minus ATL balance of the previous day. Note the example at the sixth TSB bar from the far right, where the more heavily negative TSB bar followed the big spike in ATL the day before.
Depending on the training and/or racing period, an important objective may be progression of chronic training load, during which time ATL is often higher than CTL, and the TSB is negative to some degree. Alternatively, when racing, the ATL may be allowed to drift lower, which means the CTL will fall and TSB will rise. In fact, a coach may actively manage ATL and CTL such that the athlete has a positive TSB on the day of a high-priority race.
Below is Runner 2’s PMC as he enters a racing period. The arrows represent race days.
We can see ATL tapering, CTL flattening, and TSB tending to be less negative. In fact, TSB was actively managed to be progressively less negative for each subsequent race, with a +1 value on the third race day (which coincided with a 5000-meter track PR for this runner).
Even though the basic metrics of the Performance Manager Chart (CTL, ATL, and TSB) enhance our ability to monitor runners’ training loads considerably by reflecting both volume and relative intensity, they do not necessarily convey composition of the training load. For example, a 100-TSS workout is treated the same, whether that 100 TSS came by a longer, lower-intensity run or by a run with higher-intensity intervals and a shorter total duration.
CIL - Chronic Intensity Load
CIL was originally developed to track the impact of shorter, intense mountain bike loads. It may also be useful for runners, particularly those racing 800-10,000 meters. CIL essentially tracks intensity load, allowing some interpretation of chronic training load composition.
Let’s take a look at Runner 2’s Performance Manager Chart again, this time with CIL added (below) as the dashed thin blue line. This PMC includes both build and racing periods; the white arrows, again, are race days. Note how the CIL line (representing intensity load) starts to rise as the build period progresses. Also note that once the racing period begins, ATL tapers, CTL begins to flatten, and TSB becomes less negative. During the same period, CIL is maintained at a relatively high level even while ATL tapers. This is indicative of shorter, more intense workouts during this period.
Thirteen days later (the far right of the chart below), while continuing to lighten the ATL and keep CIL sharp after the 5000-meter PR (the first arrow below), TSB is allowed to rise to a value of +13 for a 10,000-meter track race (the season’s “A” race), resulting in another PR.
Below is another example from Runner 1, showing again CTL as the thicker, light blue line; ATL as the thinner, pink line; TSB as the yellow bars; and CIL as the thin, dashed blue line.
At the far left, we can see that this runner’s base building progressed quite well, with a nice progression of CTL (the thicker, light blue line), until he caught a cold that impaired his running in late January (the red box in the chart below). During the period of the cold, we can see that missed runs resulted in a loss of CTL (fitness) and a rise in TSB (the yellow bars).
Following the cold, Runner 1 worked on rebuilding fitness as he joined formal practices with his school team (shown in the red box below). CTL was once again rising at a steady rate, while his TSB hovered in a sustainably negative zone. As a result of the greater focus on intensity, we can see CIL (the broken, thin blue line) rise just a bit faster than CTL.
In early March, this runner had weekend races on three successive weekends (shown in the red box below). The race days can be identified by TSB (the yellow bars) being purposely much less negative than on surrounding days. The runner raced 1600-meter races on two of the three weekends, noted by the arrows. Both of these races resulted in PRs. However, as a consequence of tapering, ATL (the pink line) and CTL flatten out. CIL continued to rise, due to the intensity of the races combined with the nature of the workouts during this period.
However, the flattening of the CTL trend and the two PRs raised a bit of a red flag. With the season only 25% complete, was the runner sacrificing too much CTL (fitness) too early?
In the chart below, the red line depicts a hypothetical projection of CTL (relative fitness) if he kept on the same course. The purple line at the far right depicts where his CTL (relative fitness) might have ended up with additional taper for the big meets at season’s end.
The runner was assigned a one-week block (the red box below) of added volume with added tempo running, coinciding with spring break, designed to raise CTL again and forestall a premature peak.
After spring break and the short CTL-build block, the runner resumed racing (shown in the red box below). During this period, he raced three more 1600-meter races (identified with arrows). The first of these was run on his most negative TSB of the season, followed by a bit less negative for the next one and then a slightly-positive TSB on the third, which resulted in another PR at the third race. Note that during this period, his CTL continued to rise at a gradual rate, as did his CIL.
The final period of his season (the red box below) was marked by a tapering period into the big races of the season: his league championships 1600m and the sectional championships 1600m (identified with arrows). We can see that during this time, CTL is exchanged for a positive TSB and fresh legs. In fact, during this period we manipulated training load based on projections made in the Performance Manager Chart to actually arrive on race days at a targeted positive TSB.
Also note that while overall training load subsided, relative intensity load, as depicted by CIL, remained up. The first arrow during this period marks the league championships, where wind conditions narrowly kept this runner from a PR. The second arrow marks the sectional championships, where he set yet another PR, ending a very successful season with nearly 4% improvement in average power over 1600 meters within the season, and nearly 5% improvement in time from his 2016 PR.
Miles vs rTSS
Let’s contrast the paradigm of monitoring training load by miles (volume) to rTSS (duration and relative intensity). Below is Runner 1’s season depicted on a Performance Manager Chart.
Now have a look at his training load in terms of miles vs rTSS in the following table.
|
|
Duration |
Miles |
TSS |
Work (kJ) |
|
5/15/2017 |
10:10:14 |
30 |
411 |
2,889 |
|
5/8/2017 |
10:25:06 |
28 |
398 |
2,749 |
|
5/1/2017 |
11:17:20 |
26 |
474 |
2,531 |
|
4/24/2017 |
13:18:20 |
42 |
677 |
4,184 |
|
4/17/2017 |
15:14:56 |
32 |
558 |
3,168 |
|
4/10/2017 |
10:38:32 |
41 |
569 |
3,882 |
|
4/3/2017 |
11:33:24 |
36 |
505 |
3,575 |
|
3/27/2017 |
12:33:46 |
35 |
446 |
3,471 |
|
3/20/2017 |
7:12:02 |
40 |
718 |
4,018 |
|
3/13/2017 |
14:53:34 |
35 |
540 |
3,407 |
|
3/6/2017 |
15:38:00 |
43 |
424 |
3,515 |
|
2/27/2017 |
14:59:46 |
38 |
433 |
3,522 |
|
2/20/2017 |
12:46:48 |
42 |
470 |
4,189 |
|
2/13/2017 |
11:21:16 |
37 |
456 |
3,639 |
|
2/6/2017 |
10:21:48 |
41 |
450 |
2,808 |
|
1/30/2017 |
5:34:38 |
18 |
214 |
1,882 |
|
1/23/2017 |
8:11:44 |
31 |
410 |
3,201 |
|
1/16/2017 |
15:13:54 |
34 |
437 |
3,451 |
|
1/9/2017 |
11:03:22 |
53 |
632 |
5,517 |
|
1/2/2017 |
9:44:28 |
26 |
314 |
2,478 |
The Performance Manager Chart depiction of Runner 1’s training load reflects progression of the training load far better than miles alone. The Performance Manager Chart adds the benefit of visualizing relative freshness/fatigue (TSB) and intensity load (CIL).
Applications of Performance Manager Chart Metrics
1. Prescribing and monitoring adequate progression of training load in base and build periods.
As demonstrated above, it is desirable that CTL progress during base and build periods. The key question is: what is the sweetspot for ramping training load? The answer is unique to each runner, based on age, years of running, recent training history, and injury history. For most runners, however, the optimal ramp rate is likely in the ballpark of building CTL within the Performance Manager Chart at a rate of 6 +/-2 TSS/day more each week. A rate higher than this is typically not sustainable, while a rate lower than this may leave some progression of training load on the table, so to speak. If CTL is not progressing more than perhaps 1-2 TSS/day for 3-4 weeks and there haven’t been other changes (such as a change in composition or increase in CIL), it’s likely that training stagnation is occurring.
2. Monitoring intensity within the training load
CIL gives some insight into the composition of the training load. This may help monitor progression of intensity during the build period and into the race period. Further, there is potential benefit in monitoring CIL during periods where the CTL ramp rate is flattening or declining due to volume taper while intensity is still being maintained or advanced. An example of this would be the classic peaking scenario, in which volume is reduced while intensity is maintained to some degree, as depicted in the red box in the figure below.
3. Avoiding training errors and their consequences
The most common cause of injury and illness in runners is training errors. One of the more common training errors is excessive ATL relative to CTL (in other words, too rapid of a training ramp). Another common error is inadequate recovery. These types of errors can be mitigated by a) avoiding a CTL ramp rate greater than 8 TSS/day week after week, and b) making sure that TSB does not stay deeply negative for sustained periods.
4) Managing tapering
Training load is typically reduced leading up to a race, and TSB will rise. A coach or runner can actually enter the estimated TSS of future planned workouts leading up to a race and evaluate their impact on estimated race-day TSB. For “C” or “B” races, we might plan to allow TSB to come to up a less negative level near zero, while planning a positive TSB for “A” races. It has been suggested that shorter, more intense races might be optimally performed at more positive TSB than longer races. Certainly the optimal value for an “A” race will be positive, but how positive will be dependent on the individual and the race duration/intensity.
A Final Example
Below is Runner 2’s Performance Manager chart from 1/1/2017 through 6/10/2017.
The red box below reflects the runner’s base and build periods, throughout which we can see a maximally-sustainable CTL ramp. In the build period, progression of the CIL is notable.
The second section of the season (the red box below) reflects the runner’s racing period, and this period is the focus of all following images. In this phase, the CTL progression rate is much flatter, while CIL continues to progress.
I’ve taken the area within the red box above and zoomed into it for the entire image below, using arrows to indicate races and race distances.
This runner opened her season with the primary target of a 10,000-meter race on April 13 (the fourth arrow from the left above). Her first race was a 5000-meter track race; it was run at -22 TSB and produced a time of 17:26 (her PR prior to 2017 was 17:17). Her next race was a week later, a 1500-meter track race run at -7 TSB that resulted in a 1500-meter PR. With further progression of CIL and management of the TSB to +1, she ran her second 5000-meter race of the season thirteen days later in 17:07, a new PR. Another thirteen days later, she ran her target 10,000-meter track race of the season after managing TSB so that she raced at +13 TSB on race day; she achieved a 1m45s PR for 10,000 meters that day (35:39). The remaining two races -- a 5000-meter track race and a 1500-meter track race -- were “run out the season” races. In the 5000-meter race, unprescribed weight training two days prior to the race and raceday temperatures conspired to result in a 17:26 time. Her TSB was +4, but did not (could not) reflect the added fatigue of the weight training session. The last 1500-meter race, run at +17 TSB, also failed to produce a new PR, primarily due to race tactics and competition that resulted in a slow early pace.
This runner’s 10,000-meter performance allowed her to gain entrance into a more elite track race at the Portland Track Festival in June, thus extending her track season. Consequently, a four-week block (the red box in the figure below) was designed to raise her CTL back up, focusing on 10,000-meter specific training.
At the end of this block, the runner ran a road 5K as a training race tune-up. The race is not depicted with an arrow in the figure above, but it occurred on the last day of the time period inside the red box, with a TSB of -10. She won the race while running at less than full effort in a time of 17:11 (better than her track PR prior to 2017). She was ready to pop a good time in her 10,000-meter race at Portland. All that was left to do was manage her TSB into a nice positive number while keeping sharp.
In “A” races, the fine tuning of TSB is possible by entering the anticipated rTSS of planned workouts leading into the race. By reviewing actual rTSS values of various pre-race week workouts from performances prior to other races, or by estimating rTSS, we can enter these as expected values and then adjust them higher or lower (by modifying volume or intensity load) accordingly. For example, the workout list below reflects the planning for the final week going into the Portland Track Festival 10,000-meter race. The red box contains the planned/expected rTSS values for the various workouts.
The resultant changes in the Performance Manager Chart (shown in the red box in the figure below) were then checked to determine that the predicted TSB on race day was sufficiently positive. The prediction was for a TSB of +16 on race day. (Note: since no intensity values were entered, CIL was not used as a predictive metric.)
The chart below is the runner’s Performance Manager Chart with the actual workouts loaded after the race. Her rTSS for each actual workout was very slightly lower than each day’s predicted rTSS. Consequently, her actual TSB on race day was +19. She ran 34:51 (another PR), knocking another 48 seconds off of her previous best from April 13 and 2m33s off her PR prior to 2017!
Caveats
- The Performance Manager Chart was originally designed for cycling. Its efficacy for runners has not been widely reported, but in my experience, it appears to be as useful for runners as it is for cyclists.
- The Performance Manager Chart requires the input of all running training data for greatest utility. Failing to load some runs will adversely impact its utility.
- The Performance Manager Chart requires entry of a reasonably-accurate and up-to-date functional threshold pace.
- The Performance Manager Chart metrics of CTL, ATL, and TSB, as presented here, are based on normalized graded pace. Now that running power meters are available (all of the runners I work with use a Stryd power meter), it may be that basing the Performance Manager Chart metrics on power may enhance its utility for runners who use a power meter.
- The Performance Manager Chart is based on input running data. To the extent that confounding variables may be present (like sleep deprivation, dehydration, poor nutritional support, mental readiness, and, as presented in the case above, doing weights two days before a race, among others), TSB may be impacted.
- As George Box said, "The most that can be expected from any model is that it can supply a useful approximation to reality: All models are wrong; some models are useful." The Performance Manager Chart is a model. It is useful. It is not an infallible reality. Guidelines, not rules.
Summary
The quantification of a runner’s training load is a prudent aspect of his/her training. Attempting to quantify training load by volume alone can be misleading; instead, quantification of the training load by incorporating both volume and relative intensity is superior.
Stepping beyond merely quantifying training load using both volume and relative intensity, we can employ the Performance Manager Chart and related metrics. We can then monitor and even actively manage training to optimize training load progression (including intensity load), taper/sharpen a runner into optimal race-day preparation, and perhaps even contribute to injury avoidance.
In the end, the Performance Manager Chart and its related metrics are not without limitations. Nevertheless, the PMC is a useful model that takes runners and coaches to a new level of objectivity in training load management, which in turn can be used to help runners shine.
This article was written by Steve Palladino. Steve Palladino is a middle- and long- distance running coach and consultant for Palladino Power Project. He holds a 2:16 marathon PR, 54:27 40K Cycling ITT, qualified for the 1980 US Olympic Trials, and is the winner of the 1978 San Francisco Marathon. He is a retired DPM foot and ankle surgeon, and has been training and racing with a power meter and WKO user since 2003.