A study of MPG's predictive and prescriptive accuracy
Since 2008, MPG has been in a continual state of progress. The goal has been to create a performance-based system which can accurately prescribe endurance training specifics. In 2008 we started experimenting and documenting with swimmers, cyclists, runners and triathletes of all levels. As we progressed, we tried many different performance test strategies; some of which have been in use in coaching circles for many years. The one problem we found with every one of these performance tests, was that they were only accurate for a relatively small proportion of the athlete population. We found that a single 5km running test, or a single 20 minute time trial did not give us enough information to accurately predict an athlete’s performance consistently across the board. When one reviews the literature over the past 17 or so years, it becomes very clear that heart rate zones, even when determined using stringent heart rate testing protocols such as the Conconi method, do not offer a very accurate correlation to lactate physiology14,15, whereas performance measured in speed and power have shown to be far more accurate.
At the end of the day, if you cannot accurately predict how fast your athletes will go over a variety of distances, you cannot prescribe accurate and highly specific training zone intensities for that athlete. As a coach, you need to know where your athlete’s strengths and weaknesses are. Although most coaches claim to know this information about their athletes, when you put the athlete’s performance numbers to the test, coaches, myself included, have a hit-or-miss rate of accuracy. You need a scientific approach, that is evidence-based and also once that tests outcomes. In the case of coaching, we need accurate information that tells us about each athlete's changing physiology. Because there are a number of calculations that need to be made, one also needs a substantial amount of mathematics to get this right.
We designed a 2-part performance test and formulated an algorithm which could predict an athlete’s Performance/Duration curve; performances from 30s to over 5 hours. Using the data from the athletes who have been on the MPG system, we have been able to refine and perfect these algorithms, making sure the data fits each athlete. In this way, the system remains highly accurate for every type of athlete, from beginners to professionals.
The algorithm is basically an hypothesis that calculates the following: If the performance over a time of 4 minutes is X, and the performance over a time of 20 minutes is Y, then performance over 60 minutes is Z. Each of the 14 points (time/distance) listed on the Performance/Duration curve has its own algorithm. These algorithms have been refined using the data from thousands of athletes.
We have been very transparent with our methodology from the beginning. This has resulted in coaches and sports scientists testing our methods independently, in the sports science lab, by performing lactate studies on MPG athletes. The way this has been done is by creating the classic Lactate/Performance curve. These values are then compared to the MPG training numbers. Every time, we have received positive feedback and have been praised for developing a system so accurate. Every time, we had no idea that the tests were even being done, and we were only informed after the fact. The classic lactate curve, however, is not widely considered the gold standard in terms of determining the lactate turn-point. The current gold-standard is widely attributed to the Maximal Lactate Steady State (MLSS) in sports science circles. In this study we will test whether MPG can accurately calculate the athlete’s MLSS in runners by using the 2-part performance test.
Pilot Study objectives:
1) We aimed to determine whether the MPG algorithm, which is derived from a 2-part performance test, is able to determine the pace at Maximal Lactate Steady State (MLSS) with a high degree of accuracy, on a varying range of athlete abilities.
2) We aimed to determine whether MPG was able to accurately predict the race pace of a number of individuals, over a variety of distances.
3) Create an ongoing experiment to compare self-paced workouts to MPG prescribed workouts.
Part 1: Testing Maximal Lactate Steady State
It has been demonstrated that Running Speed at a RER of 1.00 correlates closely with the MLSS1. Also, it has been shown that performance at MLSS is a very good predictor of racing performance2. Furthermore MLSS has been shown to be valuable in creating accurate training intensities3, and shows an excellent correlation to the marathon pace6. Therefore if the MPG algorithm can accurately determine MLSS using the MPG-predicted marathon pace in a wide range of individuals, it will confirm the accuracy of the MPG algorithm.
Self-conducted performance tests are highly reproducible, more cost effective and more user friendly in comparison to a lab-based test. Also, tests conducted in real-world conditions have shown to be more accurate in predicting real-world performance4,5. We used the MPG self-conducted performance test which was observed and supervised. Instructions given were identical to those provided in the MPG mobile app.
It has been shown that intense training causes an increase in the circulating stress hormones, whereas low intensity training has the effect of lowering the circulating stress hormones7. It is therefore important to have the correct balance of easy and hard training. Perceived effort is also very important. Hard workouts need to be appropriately hard and easy workouts need to feel easy. When athletes fail to meet their workout targets, they feel a sense of failure. In the same way, if the hard workouts feel too easy, the athlete is left feeling that they are not being challenged sufficiently. In the same way, if the easy training sessions feel too hard, it will impact the recovery leading up to the harder workouts7.
In summary, it is critical that each training session is at the perfect intensity so that the goals of that particular training can be achieved. The construction of training programs needs to follow the principles of scientific exercise prescription, where intensity, frequency, distance and time are prescribed optimally for every athlete. The final part of this study aims to determine whether MPG meet these criteria.
Materials and Methods
8 Runners (3 male, 5 female) completed the study. All runners could run for at least 40 minutes continuously. The athletes were classified according to 4 groups: Professional (1), Elite Amateur (2), Amateur (3) and Novice (2).
Athletes were given verbal instructions of how to conduct the test. The information was identical to the instructions provided in the MPG Mobile App: Athletes performed a 2-part performance test individually, under observation: First, a 1000m time trial on an athletics track, followed by a full recovery. When the athletes were ready, they ran a 5km time trial, also on the track. The athletes were instructed to aim for a negative split, pacing themselves in such a way as to finish as fast as possible.
The times achieved for the 1km and 5km were then entered into the MPG Mobile App, where the Performance/Distance curve was then produced. The MPG-predicted marathon pace, shown in the app, was then used as the pacing guidelines for the experiment.
3 to 7 days later, the athletes ran for 30 minutes at the MPG-determined MLSSv which was determined by using the MPG-predicted Marathon pace for each individual. Blood lactate measurements were taken at 10, 20 and 30 minutes. Thereafter, the athletes aimed to run for an additional 10 minutes at 5% faster than the MLSSv. The following criteria were used to assess the level of correlation of the MPG predicted marathon pace, with the MLSS:
• Did the blood Lactate rise by 1,0 mmol/L or less from 10 to 30 minutes? If so, the pace would be considered to be within or at the MLSSv.10
• Was the Blood Lactate concentration at least 50% higher at 40 minutes, than at 30 minutes, despite being only 5% faster? This would suggest an exponential rise in lactate, characteristic of an effort at or above the lactate threshold.
• Did the athlete increase their pace between 30 and 40 minutes by less than 10%?
• Did the perceived effort of the 30 minute steady state effort feel comfortable?
• Did the marginally increased pace between 30 and 40 minutes feel significantly harder, despite being only marginally faster?
Recent studies have shown that LT can vary greatly between athletes and the range has been demonstrated to be anywhere between 2 and 8mmol/L. Therefore, we chose not to use the classic 4mmol/L measurement as an indicator. Athletes were then given a score of 1 to 5.
Lactate Measurements and Pacing
|(La) 10'||(La)20'||(La)30'||Pace 30'||Diff.(La)||Pace 40'||(La)40'||% Increase (La)||% Increase (Pace)|
• 4 (50%) of the athletes were able to meet all 5 criteria. The MPG pacing was highly accurate. These athletes all ran for 30 minutes at a pace that was certainly below lactate threshold. All these athletes could exceed the lactate turn-point successfully, by increasing their pace by a mere 6% on average.
• 1 Athlete met 4 of the criteria.
• 3 Athletes met 3 of the criteria.
The difficulty of an experiment like this is in getting the athletes to run at a consistent pace, without being paced automatically as one would on a treadmill. However, we chose to do the study like this because it is more realistic to the manner in which athletes would conduct themselves in real world conditions.
It is interesting that athletes 2, 5 and 7 had a higher blood lactate at 10 minutes, than at the end of 30 minutes. It is possible that the warm up, which culminated in some short hard run-throughs, elevated blood lactate prior to the start of the steady state run, and the test was started before the lactate normalised. In future tests, we will allow more time between the warm up and the actual test.
The benefit of doing a study with a relatively small sample size, is that we can look at individuals more closely and objectively. The difficulty is getting athletes to run at a consistent pace close to the lactate threshold, without ever exceeding it. My concern was that the micro accelerations and decelerations would push the lactate over threshold prematurely. Fortunately, this was not the case. Even in the athlete who had an increase of 1.2 mmol/L; that difference is certainly not a rise indicative of crossing the lactate threshold, but it is high enough to suggest that the MLSS pace is slightly slower.
If we look more closely at that athlete: Their lactate increased by 1.2 mmol/L from 10 to 30 Minutes. However, if we analyse the athlete’s pacing, they were marginally faster than prescribed. Presumably, running slightly slower at the prescribed pace would have resulted in an increase of less than 1,2mmol/L and likely less than 1mmol/L. The fact that the athlete’s lactate increased by a mere 0,5mmol/L from 10 to 20 Minutes suggests that this is a likelihood.
An athlete who scored 3 (Athlete 8) completed the performance test at temperatures approaching 30 degrees Celsius. High temperatures would slow the athlete down considerably, independent of lactate production. The performance test results were therefore slower than they would have been in cooler temperatures. In doing this pilot study, meeting the athletes for the tests had to be planned in advance and therefore we couldn’t predict the temperature, even though we scheduled the test for the early evening. The athlete’s subsequent lactate measurements were therefore predictably all low, and we were not able to demonstrate the MLSS, and it also predictably underestimated the athlete’s running speed in her race-pace prediction.
Interestingly, the 2 remaining athletes who scored 3/5 both completed the performance tests feeling like they could have gone faster and this clearly had a conservative effect on their Marathon-Pace prediction. With experience in repeating these tests, athletes get to improve their ability to judge their pacing. On the other hand, all 4 of the athletes who were able to produce a full 5 point score have done these performance tests regularly. Another factor that played a role was the fact that the performance tests of the slower athletes took longer than the faster athletes. The longer the test lasts, the more mentally challenging it is to repeat these tests on a regular basis. In future, we will get the slower athletes to do the shorter 2-part performance test prescribed in the MPG App (600m/3000m).
Part 2: An analysis of the accuracy the MPG race-pace predictions
5 of the 8 Athletes tested in this pilot study had a key race scheduled during the Training block of our study. The race pace guidelines covered races from 10,5km to the Marathon. As you can see from the table below, most athletes raced exactly as predicted. Two athletes managed to run slightly faster than predicted, but that was expected because:
• Athlete 8 did the performance test in the extreme heat, making her predicted times slower.
• Athlete 6 felt like that their 5km portion of the performance test could have been run a bit less conservatively. This was confirmed by their relatively low blood lactate, during the first 30 minutes of her MLSS test.
Pilot study race results
Getting the most out of every workout is very important in the long term. Every workout is an opportunity to become faster. Pacing oneself perfectly during these workouts consistently, makes a big difference to the athlete’s performance. Furthermore, the specific purpose of the workout needs to be achieved. For the purposes of this experiment we are testing a Tempo workout, where the purpose is to exceed the lactate threshold marginally in the intervals, but with sufficient recovery in place so that continued repetitions can be completed as fast as possible. The faster the entire workout can be completed, the more effective the workout.
Start the workout too fast, and the recovery segments become too slow, or the pace cannot be sustained in subsequent intervals. The goal is to make the hard parts as fast as possible, on average, but also to complete the entire workout as fast as possible. Run the recovery segments too fast, and the pace is not maintained in the subsequent intervals either. For this test, we use a Tempo Interval workout consisting of 5 intervals of 5 minutes of hard running, with 1 minute recovery jogging in between, to test the prescribed pacing. We exclude the warm up and cool down from the analysis. The duration of the hard part is almost 30 minutes which makes accurate pacing very important.
Part 1: The athletes are first instructed to run the workout as fast as possible, running on “feel”. They are instructed to go on “feel” to get the fastest possible average pace for their intervals, but also the fastest possible pace for the workout as a whole. Part 2: 2-3 days later the athletes are instructed to do the same workout, but are given the pacing from the MPG app, derived from their performance assessments. This includes pacing for both the Interval and Recovery segments. The two workouts are then compared.
We want to see which strategy produces:
1) The fastest average Interval pace.
2) The fastest overall average pace, including both Interval and Recovery segments.
In every case so far, the MPG-prescribed workout produced the faster average interval pace, and also the fastest overall workout pace. In athletes where the fastest interval exceeded the MPG pacing, the average interval pace achieved was always slower than the pace achieved when following the prescribed MPG interval pace. Interestingly, athletes who manage to self-pace themselves evenly almost always pace themselves too conservatively and they run faster when following the MPG guidelines, whereas athletes who start faster than the MPG pacing are almost never able to maintain their interval pace in the subsequent intervals. When they do, they have to recover for longer than 1 minute and break the protocol, resulting in a slower overall time.
This pilot study has not been published in any journal. We are preparing a more formal study conducted through a leading university, involving a larger sample size and more precise methodology. Stay connected with us for more details on that coming soon!
1) Leti T, Fourier J, Mendelsohn M, Laplaud D, Flore P. Prediction of maximal lactate steady state in runners with an incremental test on the field. J of Sports Sciences 30(6):609-16
2) Van Schuylenbergh R, Vanden Eynde B, Hespel, P. Prediction of sprint triathlon performance from laboratory tests. European Journal of Applied Physiology, volume 91, pages94–99 (2004)
3) Llodio I , Gorostiaga E.M , Garcia-Tabar I, C Granados 2, Sánchez-Medina L. Estimation of the Maximal Lactate Steady State in Endurance Runners. Int J sports Med, 2016 Jun;37(7):539-46.
4) Di Michele R, Di Renzo AM, Ammazzalorso S, Merni F. Comparison of physiological responses to an incremental running test on treadmill, natural grass, and synthetic turf in young soccer players. J Strength Cond Res. 2009;23(3):939–945.
5) Koklu Y, Alemdaroqlu U, Demirhan R, Arslan Y. A comparison of incremental Running Field and Treadmill tests in Young Soccer Players. J Hum Kinet. 2020 Jul:193-201.
6) Billat V, Sirvent P.,Lepretre PM, Koralsztein JP (2004). Trainning effect on performance, substrate balance and blood lactate concentration at MLSS in master endurance-runners. Pflugers Arch-Eur J Physiology 447: 875-883.
7) Hill EE, Zack E, Battaglini C, Viru M, Viru A, Hackney AC. Exercise and circulating cortisol levels: the intensity threshold effect. J Endocrinol Invest. 2008 Jul;31(7):587-91. doi: 10.1007/BF03345606. PMID: 18787373.
8) Blood lactate concentration at the maximal lactate steady state is not dependent on endurance capacity in healthy recreationally trained individuals. Smekal G, von Duvillard SP, Pokan R, Hofmann P, Braun WA, Arciero PJ, Tschan H, Wonisch M, Baron R, Bachl N. Eur J Appl Physiol. 2012 Aug;112(8):3079-86. doi: 10.1007/s00421-011-2283-7. Epub 2011 Dec 23. PMID: 22194004
9) Effect of aerobic training status on both maximal lactate steady state and critical power. Greco CC, Caritá RA, Dekerle J, Denadai BS. Appl Physiol Nutr Metab. 2012 Aug;37(4):736-43. doi: 10.1139/h2012-047. Epub 2012 Jun 8. PMID: 22680338
10) Beneke R. Methodological aspects of maximal lactate steady state-implications for performance testing. Eur J Appl Physiol. 2003 Mar;89(1):95-9. doi: 10.1007/s00421-002-0783-1. Epub 2003 Jan 21. PMID: 12627312 Clinical Trial.
11) Faude O, Kindermann W, Meyer T. Lactate threshold concepts: how valid are they? Sports Med. 2009;39(6):469-90. doi: 10.2165/00007256-200939060-00003. PMID: 19453206 Review.
12) Beneke R. Maximal lactate steady state concentration (MLSS): experimental and modelling approaches.Eur J Appl Physiol. 2003 Jan;88(4-5):361-9. doi: 10.1007/s00421-002-0713-2. Epub 2002 Oct 30. PMID: 12527964 Review.
13) Ralph Beneke, Renate M Leithäuser, Oliver Ochentel.Blood lactate diagnostics in exercise testing and training. Int J Sports Physiol Perform. 2011 Mar;6(1):8-24. doi: 10.1123/ijspp.6.1.8. Review
14) Bourgois, Jan; Coorevits, Pascal; Danneels, Lieven; Witvrouw, Erik; Cambier, Dirk; Vrijens, Jacques Validity of the Heart Rate Deflection Point As a Predictor of Lactate Threshold Concepts During Cycling. Journal of Strength and Conditioning Research: August 2004 - Volume 18 - Issue 3 - p 498-503
15) Jones, A.M., and J.H. Doust. The Conconi test is not valid for estimation of the lactate turnpoint in runners. J. Sports Sci. 15:385-394. 1997.