Running Functional Threshold Power - A Primer

By Steve Palladino

coach and consultant, Palladino Power Project

October 31, 2017

Functional threshold power (FTP) is a metric, developed by Andrew Coggan, PhD,  that has been in use in the cycling world for more than 15 years.  However, its use in the context of running is perhaps less than 2 years old.  For runners that have not been following discussions on FTP over the years, or the nuances of estimating it, this article is intended to deepen the reader’s understanding of this important metric, and refine its use within the context of a run training program.

What is FTP?

Why monitor FTP?

it enables a <runner> to define and measure intensities of <running> (or power levels) relative to their own current level of fitness, expressed in a manner that relates to the primary physiological adaptation that occurs at each intensity (power) level. This is very useful for guiding training and making sure that the mix of intensity and duration during a workout or training cycle is appropriate for gaining the specific fitness required for a <runner’s> target events.”  <runner/running substituted for rider/riding>  

Although training in supra-FTP zones is enhanced by following an individual’s power-duration curve at those intensities, FTP nevertheless, retains importance in providing a descriptive framework for those higher intensities.  Further, since most runners execute races from 5K to marathon at FTP +/- 10%, FTP guided training retains significance.

Figure 1.  Example of tracking FTP.

Power and Chronic Training Load on y-axis, date on x-axis.  White line is WKO4 modelled FTP.   Blue line is WKO4 Chronic Training Load (CTL).  The dotted vertical line on 6/10/2017 coincided with the runner’s 10,000m track PR (a 6.8% improvement in time versus prior to the season start).  Also note that the FTP peak and PR coincided with a short taper of the runner’s CTL high.

Errors of assumption regarding FTP....

Requisites for accurate FTP estimation

  1. A useful model of FTP estimation
  1. Within the model, the number of parameters used may be a factor
  1. Maximal or very near maximal efforts for each parameter used in the model


Methods of estimating FTP from running powermeter field data

How many parameters does your estimation model use?

Two parameter models work best with a shorter duration maximal effort, and a longer duration maximal effort.  The short duration effort should be at least a minute long.  The larger the difference in duration between the short duration test and long duration test (with the longer duration extending to as much as 20 minutes), the more reliable the estimate (as long as both tests are maximal).  To independently estimate CP outside of Powercenter, there is at least one online calculator that allows the user to use either two or three parameters.

Protocols - the WKO4 power-duration curve model

When using the WKO4 power-duration curve (PDC) model for FTP determination (referred to as mFTP), your maximal efforts are taken into consideration along the breadth of your own mean maximal power-duration curve. (Figure 2)  The model samples at least 3, but probably not more than 10 locations on your mean maximal curve - likely between the far left of the curve (Pmax) through at least the point where the flatness of your mean maximal curve begins to drop (approximately Time to Exhaustion).

Figure 2. Meanmaximal power curve (MMP curve) and resultant power-duration curve (PDC) modelling, including associated PDC metrics.

Meanmaximal power curve (runner’s actual maximal efforts for last 90 days) = yellow line

WKO4 PDC model (based on the runner’s MMP curve) = red line

 

From the PDC model, mFTP (among other PDC metrics) is estimated. (Figure 3)  The accuracy of the mFTP estimate is most dependent on maximal or near-maximal efforts represented across the parameter sampling area of the MMP curve.  Note that in Figure 3, the MMP curve (yellow line) sags below the PDC model prediction at perhaps 35s through 1m30s, which may be an area of the MMP curve that is underrepresented with true maximal or near-maximal efforts, and consequently, tethering the model down in this case.  In addition, note the flatness of the MMP curve (yellow line) from 1 second through 30 seconds.  This likely represents the lack of a true maximal effort in the realm of these durations, which may also impact the model.  (Note: there is a short time lag in Stryd recording from a standing start - it is best to test the far left of the athlete’s power-duration curve with flying start efforts.)  To help assess the accuracy of the FTP estimate, the standard error of the estimate and coefficient of variation can be calculated and reported. (Figure 3)


Figure 3.  Meanmaximal power curve (MMP curve) and resultant power-duration curve (PDC) modelling, including associated PDC metrics and S.E.E and CV of mFTP

Meanmaximal power curve (runner’s actual maximal efforts for last 90 days) = yellow line

WKO4 PDC model (based on the runner’s MMP curve) = red line

In this example, mFTP is estimated to be 233W with a S.E.E. of 2.8W and CV of 1.2%.

Practical guidelines for using the WKO4 model for FTP estimate (mFTP)

*(FRC = PDC metric Function Reserve Capacity - similar to anaerobic capacity - or the extent to which, in kJ, one can exert above mFTP before fatiguing)

Protocols - the Critical Power model

Stryd Powercenter critical power prototcols

The Stryd Powercenter critical power model uses two parameters.  Testing involves performing two maximal effort tests in one test session.  Resultant values from each of the two tests performed in the session are then input into the respective Stryd Powercenter critical power calculator to arrive at a critical power estimation.

   

Option 1:  3 lap and 6 lap tests performed in one test session on a 400m track

1. Warm up for 5 minutes. Do two to three 100-meter strides at approximately 80% maximum effort during warm up to enhance the blood circulation and ready your muscle for intense use.

2. 800 meters Easy-pace run. Two laps on a 400-meter track, please use the innermost lane. Run at an easy pace, such that you can comfortably maintain conversation.

3. Warm up for another 5 minutes.

4. 1200 meters Maximum-effort run. Run at a consistent pace throughout the test, but so that you are nearly exhausted at the end of the test.

5. Recovery for 30 minutes. Throughout the 30-minute recovery period, the runner should walk or jog slowly.

6. 2400 meters Maximum-effort run. As was the case for the three-lap run, it is important to maintain a consistent pace during this run instead of dramatically changing pace (and effort) during the run.

7. Cool down.

Option 2: 3 minute and 9 minute tests performed in one test session

1. Warm up for 10 minutes. Do five 100-meter strides at approximately 80% maximum effort during warm up to enhance the blood circulation and ready your muscle for intense use.

2. Maximum distance run for 3 minutes. Run at a consistent pace throughout the three-minute test, but to be nearly exhausted at the end of the test.

3. Recovery for 30 minutes. Throughout the 30-minute recovery period, the runner should walk or jog slowly.

4. Maximum distance run for 9 minutes. Again, maintain a consistent pace during this run instead of dramatically changing pace (and effort) during the run.

5. Cool down.

Independent critical power protocols

One can create their own critical power protocol, independent of the Stryd critical power model.  It may be, for example, that an ultrarunner is not able to execute solid maximal efforts at 3 and 6 laps or 3 and 9 minutes.  Perhaps the runner would like to use maximal effort tests of 2 minutes and 20 minutes.  Also, the athlete may want to test on similar terrain to which they train or race (for example, trail runners).  In these cases, the runner could design their own protocol and input the values into an online critical power calculator.  The CP Calculator - Monod Critical Power Calculator from Cycling Power Lab . com will take 2, or even 3, test results, ranging from 1 to 60 minutes.  Alternatively, the CP Calculator - Cycling Zone Calculator from BeginnerTriathlete.com will take 2 test results, with the shorter test being 3m00s-5m59s, and the longer test being 12m00s-20m59s.  Keep in mind, when designing your own protocol, the larger the difference in duration between the short duration test and the longer duration test, the more reliable the estimate (as long as both tests are maximal).

Practical guidelines for using the Critical Power model for FTP estimate

Protocols - estimating FTP using a single parameter model (using modified Riegel formula)

It is possible to arrive at a reasonable FTP estimate from a single parameter input model.  However, as stated previously the potential for error may be higher than with properly used multiple parameter input models.

The Riegel formula serves as a useful model for estimation of FTP.  Modification of the formula yields the following equation:

estimated FTP = (known power) * (predicted duration/known duration)^exponent 

The issues with this formula for FTP estimation are:

  1. The “predicted duration” function in the equation.
  2. The actual “exponent” to be used for the given athlete.

For the “predicted duration” function, I somewhat arbitrarily recommend using 50 minutes. Fifty minutes aligns fairly well with what the Stryd folks say is the approximate TTE (time to exhaustion) for their CP value. Fifty minutes also lies within the ranges cited for TTE for both FTP and CP. Lastly, it likely lies much closer to most runners’ TTE for FTP than does 60 minutes. That said, in practice, I typically will bracket my estimates for FTP using this method by using “predicted duration” values of 45 and 50 minutes. Admittedly, using a standard “predicted duration” of 50 minutes as TTE is akin to a broken clock being right twice a day.  However, it is likely that this “predicted duration” produces a reasonable FTP estimation under the circumstances.

The exponent for the equation is tied to what Riegel referred to as the fatigue factor.  The exponent used in the equation may vary, depending on; a) aerobic fitness level and stamina, b) gender, c) age, and d) perhaps even anaerobic capacity.  The actual “exponent” to be used is likely unique to the athlete for any given point in their training cycle.

Examples

Practical guidelines for using the modified Riegel model for FTP estimate

Other FTP Estimation Protocols

There are other FTP estimation protocols that have been proposed for runners.  For completeness, I’ll cover them here.  For various reasons, I do not use any of them.

Run with Power 30 minute time trial method 

(Jim Vance, Run with Power, p60-61, 2016)

Note: This methodology, on the surface, seems that it could overestimate FTP - particularly in faster runners.

Run with Power 3+9 minute test method 

(Jim Vance, Run with Power, p57-60, 2016)

Note: For a number of reasons, this methodology is likely more prone to a larger margin of error than any of the primary three methods previously discussed.

The Secret of Running 10 minute time trial method 

(Hans Van Dijk and Ron Van Megen, The Secret of Running, p118, 2017)

Note: This methodology assumes a) that FTP is equal to 60 minute power, and b) a Riegel exponent of -0.07.  These assumptions may lead to an FTP estimate that errs low.

The Secret of Running 20 minute time trial method 

(Hans Van Dijk and Ron Van Megen, The Secret of Running, p118, 2017)

Note: This methodology assumes a) that FTP is equal to 60 minute power, and b) a Riegel exponent of -0.07.  These assumptions may lead to an FTP estimate that errs low.

Stryd Power Center Race Time conversion to FTP 

Note: Speed = Power * Running Effectiveness.  The calculator only calls for entry of a race distance and time (thus speed is the basis of computations). Since this method does not call for the runner to input their Running Effectiveness, the calculator, no doubt, assumes a standard Running Effectiveness (RE).  The calculator then assumes that all runners with the same race speed (or distance/time) must produce the same power (in W/kg) to produce that speed.  A runner with a 5k time of 20:00 and RE of 1.04 is likely to be computed to have the same race power, and consequently, the same FTP as a runner with a 5k time of 20:00 and RE of 0.99.   This assumption means that the calculator result for CP errs high for more “effective” (higher RE) runners and errs low for less “effective” runners (low RE).  The percentage of error can be as much as 5% in some runners at the extremes of RE.  Further, it is unknown whether the calculator converts to FTP as 60 minute power, or some other duration for FTP.

The Secret of Running Race Time conversion to FTP 

Note: The Secret of Running uses the metric Energy Cost of Running (ECOR) rather than Running Effectiveness (RE).  ECOR is the inverse of RE.  Thus  Speed = Power / ECOR.  The Secret or Running uses a default ECOR of 0.98 in their calculations (the same as a RE of 1.02).  The calculator only calls for entry of a race distance and time (thus speed is the basis of computations). Since this method does not call for the runner to input their own ECOR, the calculator, no doubt, assumes a standard ECOR of 0.98.  Further, based on early conversions reported on page 118 in the book, it is likely that the conversions are to an assumed 60 minute FTP.  The percentage of error can be as much as 5% in some runners at the extremes of ECOR.  

The Secret of Running alternatives do allow for the input the time result of a greater variety of races distances than does the Stryd Powercenter calculator, which allows only 5K and 10K inputs at this time.

Matt Fitzgerald 30 minute time trial

(from Intensity Guidelines for 80/20 Running, by Matt Fitzgerald and David Warden) 

Note: This method will likely overestimate your FTP.  It would be better to apply the Riegel single parameter model for less potential error.

Matt Fitzgerald 20 minute time trial

(from Intensity Guidelines for 80/20 Running, by Matt Fitzgerald and David Warden)

Note: This is one of the most common mis-applied assumptions in estimating FTP.  It will be spot on for a few runners, close for more runners, but associated with too much margin of error for many.  Again, it would be better to apply the Riegel single parameter model for less potential error.

Other tidbits from Intensity Guidelines for 80/20 Running, by Matt Fitzgerald and David Warden 

Note: I am unsure whether this is the same formula that the Stryd folks have built into their Powercenter CP calculator.

l actate threshold testing with concurrent powermeter use

How do I know my FTP is reasonably estimated?

Summary

FTP is one of the key metrics in running with power.  It has the significant advantage of being something that every runner with a running powermeter can estimate from field data.  The estimate represents the highest power that a runner can maintain in a quasi-steady state without fatiguing, where the duration may range from 30-70 minutes, depending on the individual.  The FTP estimate correlates with power at MLSS.  

Most runners execute races from 5K to marathon at FTP +/- 10%.  FTP is one of the most foundational metrics in running performance.  Monitoring FTP over time can be as important as an indicator of metabolic fitness as race results themselves.

Besides enabling the monitoring of metabolic fitness, FTP can be used for training guidance and establishment of personalized training zones.  Further, FTP can be used in race planning and “pacing”, intensity scaling, and training load modeling.

In using FTP, it is important to understand and correctly apply the various models and protocols for estimating FTP.  These have been presented here for reader reference.  Avoiding the pitfalls and misconceptions of FTP estimation is essential to arriving at and maintaining good FTP estimations over time.

Suggested Reading / References

Athletic Records and Human Performance, Peter S. Riegel, American Scientist, 69:285-290, 1981.

The critical power concept. A review., by DW Hill, 1993.

Power vs. duration: the "critical power" concept, by Andy Coggan, PhD, 2010 (originally posted to the internet in 2002)

Lactate threshold: its significance and determination via field-test - by Charles Howe, in Velodynamics, circa 2005

The Seven Deadly Sins - from Alex Simmons'  Alex's Cycle Blog, 2008

Sins of  Sins (Testing FTP #2) - from Alex Simmons'  Alex's Cycle Blog, 2009

What is FTP? - by Hunter Allen in Hunter Allen Power Blog, 2013

What is Functional Threshold Power? - by Andrew coggan, PhD, in TrainingPeaks.com, 2016

Introduction of the New Time to Exhaustion Metric in WKO4 - by Tim Cusick, in TrainingPeaks.com, 2016

Running with Power: How to Find Your Run FTP - by Jim Vance in TrainingPeaks.com, 2016

Your most important number: how to find your running functional threshold power - from Jim Vance's runwithpower.net, 2016

New Fatigue Factors for Running : Men's Road and Track Racing, by Ron George, 2017.

New Fatigue Factors for Running : Women's Road and Track Racing, by Ron George, 2017.

Intensity Guidelines for 80/20 Running, by Matt Fitzgerald and David Warden, at MattFitzgerald.org

CP Calculator - Monod Critical Power Calculator from Cycling Power Lab . com

CP Calculator - Cycling Zone Calculator from BeginnerTriathlete.com

Marathon Time Prediction, by Ian Williams at fetcheveryone.com

The Science of Running calculator