Tapering your training is a key factor in recovering from heavy training and preparing for competition

Why fast, exponential decay beats step reduction when preparing for a peak performance

Almost all athletes and coaches agree that tapering - the reduction of training in a systematic way - is a good thing, because it ensures good recovery from heavy training(1) and is a key part of preparation for an important competition(2). Unfortunately, there is wide disagreement about how tapering periods should be constructed. These debates revolve around how long a tapering period should be, the extent to which training volume, intensity, and frequency should be reduced during a taper, and also - very importantly - the rate at which these variables should be reduced.

One dispute has centred on whether tapers should contain 'step reductions' in training or 'exponential decays'. In a step reduction, total training is reduced by a certain amount, and the new volume of training is sustained throughout the tapering period; in an exponential decay situation, the quantity of training decreases steadily over the course of the taper in a continuous slide, reaching bare-bones levels at the end of the tapering period. One popular step-down strategy is to clip training by 65-70%, then maintain the new, lower volume of work for 1-3 weeks. Traditionally, exponential decays have been linked with shorter durations of time, often just 4-8 days.

Until now, the relative merits of step reduction and exponential decay tapering have been poorly evaluated. Several years ago, the outstanding tapering theorist Joe Houmard asked 5k runners to cut training by 70% for three weeks (a step reduction). At the end of the 21-day period, the runners' 5k race times were not significantly better, nor did they exhibit greater muscular power(3). By contrast, a seven-day exponential decay in which training volume was reduced each day and overall weekly volume dropped by 85% produced dramatic improvements in 5k race times and muscular power(4).

These very different results have led some tapering theorists to argue that when training volume is reduced aggressively and progressively to an extremely low level, performance is improved to a greater extent than with a single step reduction - or even several such reductions - over a longer period. Some anti-step scientists even argue that step reductions usually maintain performance rather than enhancing it.

Such arguments are not completely fair, since step reduction tapering has been linked with fairly impressive gains in physical capacity. For example, in a classic study carried out by renowned exercise physiologist Dave Costill in his laboratory at Ball State University, collegiate swimmers reduced training volume from 10,000 to 3,200 yards per day over a 15-day period(5), after which their performance times improved by 3.6%, their arm strength and power increased by up to 25% and blood lactate levels were lower during 200-yard swimming 'sprints'. These impressive results led Costill to recommend (in his fine book Inside Running: Basics of Sports Physiology) tapering periods of approximately two-weeks' duration, with volume set at about one third of usual levels - a large step reduction.

Short step reduction tapers do work well - at least in some studies

In later work, Raymond Kenitzer and Catherine Jackson asked 15 female collegiate swimmers to pare training volume by about 60% over a four-week period(6). For the long-distance swimmers involved in the study, volume dropped from 8,000 yards per day to 3,500 yards. During this step reduction taper, blood lactate levels fell steadily for about two-and-a-half weeks, and performances increased progressively over the same time. After this period, however, lactate concentrations and performance times both began to worsen. Kenitzer and Jackson drew the obvious conclusion: 60% step-reduction tapers lasting up to 17-18 days are good things.

Step reductions can do more than maintain performance levels. However, the exponential cause was advanced pretty dramatically shortly after the publication of Kenitzer's work. Another scientist with a strong interest in tapering, Duncan MacDougall of McMaster University in Hamilton, Ontario, asked a group of well-conditioned runners, who were averaging 45-50 miles of running per week, to try out three different kinds of one-week tapering strategies, as follows:

(1) doing nothing at all during the week (a 100% step reduction)

(2) running about 18 miles during the week at a leisurely pace, with a complete rest day at the end of the week (a 64% step reduction)

(3) undertaking a drastic exponential decay in training over the week, with an emphasis on quality running. Using this strategy, the runners completed five hard 500m intervals on the first day, four 500m blasts on the second day, 3x500 on day three, 2x500 on day four, and a single 500m surge on day five.

After a rest on day six, they were ready to be tested on day seven (along with the runners in the other two groups). Crucially, each 500m interval was performed at about one-mile race pace, and since the runners warmed up with 500m of slow jogging before the quality intervals, the total training volume for the week was about 10k, or just over six miles. Thus, this decay involved an overall 87-88% reduction in training(7).

The performance test on day seven involved running as far as possible at one-mile race pace, and the 64% step reduction runners did fairly well, advancing endurance time at this speed by 6%. By contrast, the 100% reduction runners failed to improve at all. However, the exponential runners blew the roof off MacDougall's lab, raising endurance time at one-mile pace by a full 22%! The expo group also demonstrated enhanced leg muscle enzyme activity, augmented total blood volume, increased red blood cell density and greater muscle glycogen storage, by comparison with the step-reducing runners.
These results certainly made exponential decay tapering look better than step reduction plans, but a few comments are in order. First, note that MacDougall's decaying runners employed a relatively high quantity of quality running during their taper - about 7.5k out of a total volume of 10k (75%). It is possible that the 64% step reduction runners would have fared far better if they had been able to include quality work in their training as well.

In addition, the expo decay runners trained during their taper week at exactly the pace which was used for testing. Thus, their tapering period was highly 'neural', in that it 'tuned up' their nervous systems and prepared their neuromuscular systems for the exact intensities and most efficient patterns of coordination and overall movement which they would use in the test. So, as you can see, MacDougall's work did not really compare step reduction tapering with exponential decay cutbacks but instead merely contrasted two very different tapering plans.

Other studies offer support for the 'steep slide' course

None the less, MacDougall's unique exponential plan looked pretty good, and further work by Joe Houmard and his colleagues added weight to the idea that tapering should proceed along a 'steep slide' course. Inspired by MacDougall, and having used the Ontario taper to prepare very successfully for a marathon, Houmard asked eight experienced runners (six males and two females), who had been running about 43 miles per week, to abbreviate their weekly workout to 6.2 miles of interval training and seven miles of jogging(8). Almost all of the interval training consisted of high-intensity, 400m intervals at about 5k race pace, or slightly faster.

The exponential part of the plan was modelled along MacDougall lines, as follows:

  • 1 Day 1, 8x400m intervals
  • 1 Day 2, 5x400
  • 1 Day 3, 4x400
  • 1 Day 4, 3x400
  • 1 Days 5 & 6, 2x400
  • 1 Day 7, 1x400

During the workouts, recovery intervals (composed of walking or resting) lasted just long enough to let heart rates drop to 100-110 beats per minute, and an 800m easy jog served as both pre-workout warm-up and post-training cool-down, accounting for the seven miles of jogging for the week. A control group of eight runners maintained their usual training volume of 43 miles per week.

When a 5k race was held on the eighth day of the study (immediately following the one-week taper), the exponentially advantaged runners trimmed average 5k times by a statistically significant 29 seconds, from 17:16 to 16:47, with all eight runners able to improve their clockings. They also improved running economy by a rather dramatic 6%, while the control group improved neither economy nor 5k performance.

So, to settle the argument, what's the ideal tapering period?

Such investigations didn't settle the tapering controversy, however. For one thing, many athletes tried exponential decay tapers similar to those employed by MacDougall and Houmard, only to end up with very sore legs and quite modest performances. Furthermore, even if people accepted the idea that exponential decay tapers were best, questions remained about how long such tapers should last. Was one week really the optimum duration, or should the tapering period be longer - or even shorter?

Enter E W Banister, a kinesiologist at Simon Fraser University in British Columbia, Canada, who was known for his innovative, although sometimes eccentric, research on endurance training theory. Regular readers may recall Banister as the fellow who developed a unique system for determining how much 'training stimulus' an athlete gets from a particular workout.

To use the Banister system, you simply determine your average heart rate during a workout, then subtract your resting heart rate from that figure to obtain a number which we can call 'A'. You then take your maximum heart rate and subtract from it your resting heart rate to get a second number, 'B'. Finally, you simply divide A by B to determine the relative intensity of your workout and multiply the result by the length (in minutes) of your workout to obtain 'TRIMP', your 'training impulse' for the day.

For example, let's say that your max heart rate is 200 beats per minute and your resting heart rate is 50. On a particular day, you exercise for 30 minutes with an average heart rate of 150. Thus, A = (150-50) 100, and B = (200-50) 150. A/B = (100/150) 0.67, the relative intensity of the training session. Of course, TRIMP = 0.67x30 (the number of minutes in the workout) which comes to 20.

This is a logical way to determine the value of a workout: after all, the number A is simply a measure of how far you climb above your resting heart rate during a workout, while B is an assessment of how far above the resting rate you could go if your workout were truly maximal. That means that dividing A by B automatically calculates the intensity of your workout - how close you are to working maximally during your effort.

If A and B are identical, it means that you were at maximal heart rate throughout your session and have worked as hard as you possibly could. On the other hand, if you barely climb above resting heart rate during your training session, A will be a very small number, and the workout will have a low value (TRIMP) - unless, of course, you train for many hours. Multiplying A/B by the number of minutes in your workout simply allows you to reckon the overall impact of the session - and to compare one workout with another. For example, using the above figures (max heart rate of 200 and resting rate of 50) 23 minutes of exercise with a heart rate of 180 would have the same TRIMP as 30 minutes at a heart rate of 150.

Banister's system rewards intense training but ignores specificity

There are good and bad things about Banister's system. One good feature is that it tends to reward intense training. For example, still using the above max and resting heart rates, let's compare a workout at 150 beats per minute with an equal length session at only 75 beats per minute. If we just counted heart beats, then the 150 workout would be twice as good as the 75 session, but with the Banister scheme the 150 workout gets an intensity value of 0.67, since it is two-thirds of the way between resting and max heart rates, and the 75 effort gets a rating of only 0.17, since it is one-sixth of the way. Thus, Banister would assign a 150 workout 0.67/0.17 - four times the value of a 75 session. The Banister system penalises sauntering and rewards
full-tilt exertion, which makes good physiological sense.

Another good thing about the Banister scheme is that it allows you to attach a specific number to each workout and to a weekly training load (or some other training cycle if you do not like to use weeks) - a number which takes into account both the intensity and length of your training sessions. That adds a bit of precision to your training, allows you to determine how much your training is really increasing (or decreasing) when you make changes, and also enables you to compare various parts of your training history: for example, you might want to take a close look at your weekly TRIMP values for a significant personal best.

Unfortunately, though, there are several key problems with the Banister scheme: for one thing, it revolves around heart rate, which is sensitive to a multitude of factors and may not accurately reflect intensity on a particular day; for example, heart rate might be high on a hot, humid day, even though running or cycling speed is rather low. The Banister plan also fails to take into account the specificity of training needed for a particular event. For example, someone training 600 minutes per week at an average heart rate of 140 beats per minute would tend to end up with dramatically bigger TRIMP values than someone else working out for 240 minutes at 170 beats per minute. If both individuals were training for a 5k, however, and 25% of the latter runner's training was at 5k speed while only 5% of the former runner's workouts fell into this category, the runner with the lower TRIMP would actually be better prepared for the race.

When four tapering strategies were compared, here's what they found

We thank Banister for his efforts, though, and especially appreciate his new research on tapering(9), in which he attempts to predict the extent to which performance will be improved following different kinds of tapering periods. He uses a systems model of training(10) to predict performance by transforming daily TRIMP scores into separate daily tallies of fitness and fatigue, during both regular training periods and subsequent tapers. He then examines the reliability of these predictions in a real-life field trial with 11 triathletes, who train for 94 days, with an additional tapering period, in an effort to optimise performance. In this new work, Banister employs an 'intensity factor' called Y, which is based upon the rise in blood lactate levels during a particular workout, to give even more credit to intense training. In effect, he calculates TRIMP with the equation TRIMP = D (duration of workout in minutes) multiplied by A/B multiplied by Y. In this equation, D is the training volume component and (A/B) x Y is the intensity component.

The triathletes, all male, had an average age of 26 and mean body fat levels of 8.7%. During the study, the athletes employed a total of four different tapering programmes, with one group of five using two strategies and the other group of six trying another two, as follows:

  • a single-step reduction plan, maintaining 22% of regular training volume over a two-week period
  • a 'medium' exponential decay taper, cutting training volume to a modest extent each day over a two-week period, with total tapered training averaging about 30% of normal
  • a 'slow' exponential decay taper, trimming training volume by a very small amount per day over a two-week period, so that total tapered training volume was about 50% of normal
  • a 'fast' exponential decay tapering plan, slashing training volume more aggressively each day.

The overall protocol was this: all 11 athletes trained in a standard way for 31 days, then five athletes used taper-plan 1, while the other six employed plan

2. The two weeks of tapering were followed by a four-day break of unstructured training. Regular, strenuous training was then resumed for 33 days, the athlete cohort was reconstituted, and five athletes tried taper 3, while the other six used taper 4. Thus, the comparisons were between step reduction tapering and a medium exponential decay taper, and also between slow and fast exponential decay tapering.

Banister's model predicted that medium exponential decay tapering would beat the step-reduction stratagem, and indeed it did. The medium decay triathletes improved 5k running performance from 1149 seconds (19:09) to 1103 seconds (18:23), from just before tapering began to close to the end of the two-week tapering period. Meanwhile, the step reduction triathletes reduced 5k time by an average of 13 seconds, which was not statistically significant.
The athletes were also tested to exhaustion on a cycle ergometer. For this test, the triathletes warmed up for four minutes at a work rate of just 30 watts, then exercise intensity was increased by 30w per minute until an athlete was exhausted or unable to maintain a pedal rate above 80rpm. For this test, the exponential decay also proved to be superior: athletes using exponential decay advanced their maximal power from an average of 423 to 446w, while step reduction triathletes moved from 412 to 418w. Both changes were statistically significant, but the upswing in power was significantly greater for the exponential decayed athletes.

Decay tapering really does maximise physiological improvements

A beautiful feature of Banister's research was that the intensity component of training ((A/B) x Y) was similar for both plans and for the regular training which preceded the tapers. The exponential reduction in training was achieved solely by a reduction in the duration of training sessions or the frequency of sessions per week. Thus, improvements in performance for the exponential decay group could not be attributed to higher intensity of training within the tapering period, or even to the inclusion of slightly more volume within the tapering plan - a difference which was not statistically significant. Decay tapering really was better than stepping down training with a single blow.

Comparison of tapering plans 3 and 4 also produced very interesting results. Banister's modelling predicted a superiority of the latter over the former - and again he was on target. For the cycle ramp test, slow decay triathletes improved max power from 394 to just 409w - a 4% increase - while fast decay athletes jumped from an average of 433 all the way to 467 watts - a significantly better 8% climb.
The trend was in the right direction for the 5k runs, too, with 5k time improving from 1159 seconds (19:19) to 1131 seconds (18:51) for the slow decayers (a 28-second and 2% improvement), and from 1167 seconds (19:27) to 1093 seconds (18:13) in the fast decay group (a 74-second and 6% change). However, these differences were not statistically significant.

As Banister points out, athletes train to improve physical performance, but initially the opposite response occurs, in that a significant increase in training actually induces greater levels of fatigue and poorer performances. It is left for the taper period to reveal the extent of the actual physiological improvements induced by the hard work. The tapering period must be constructed to maximise these physiological improvements, and it appears that exponential decay tapers are able to do this.

What would such a taper look like? First, there would be no loss in intensity, compared with regular training: in fact, Houmard's and MacDougall's work indicate that average training intensity should increase; secondly, the taper would have a decay pattern rather than a step reduction look; finally, fast decays seem to be better than slow ones, based on Banister's experiments. This simply means that if you are going to taper for two weeks and you usually train eight miles per day, you are better off getting your daily mileage down to 4-5 miles very quickly rather than through a gradual process. This is because the quicker reduction seems to spur more rapid recovery from and response to the previous training; slower reductions are much more like routine training loads and are less like tapers. Once you have got your mileage down to 4-5 quickly, you can then ease off on the slashing, slowly descending to just two miles-or-so of training per day just before competition.

Banister offers one final suggestion: make sure your tapering periods include at least one day off per week, with no exercise at all. These days off enhance fitness and reduce fatigue in his complex mathematical models. And the same is true in real life!

Owen Anderson


1. International Journal of Sports Medicine, vol 15, pp 492-497, 1994

2. Journal of Applied Physiology, vol 72, pp 706-711, 1992

3. International Journal of Sports Medicine, vol 11, pp 41-45, 1990

4. Medicine and Science in Sports and Exercise, vol 26, pp 624-631, 1994

5. Physician and Sportsmedicine, vol. 13, pp 94-100, 1985

6. Medicine and Science in Sports and Exercise, vol 21(2), pS23, 1989

7. Medicine and Science in Sports and Exercise, vol. 22(2), Supplement, #801, 1990

8. Medicine and Science in Sports and Exercise, vol 26(5), pp624-631, 1994

9. European Journal of Applied Physiology, vol 79, pp 182-191, 1999

10. Journal Therm Bid, vol 18, pp 587-597, 1993

Get on the road to gold-medal form and smash your competition.
Try Peak Performance today for just $1.97.

Tagged in Recovery & Training
Privacy Policy [opens in new window]
Please Login or Register to post a reply here.