Many times runners and coaches have asked us how it is possible that the Garmin Race Predictor predicts way too fast times, which are absolutely impossible for many runners. There is also a lot of criticism on the internet from disappointed runners, who are demotivated by the predictions of the Race Predictor. That is all the more unfortunate as most runners hope and expect that their watch can make a good prediction of race times based on the training data. That is an important reason to buy such a watch!

In an earlier article we had already figured this out. As a result of that article, we received many responses from runners who shared their experiences with us. The most interesting response for us came from a runner, who suggested that the cause perhaps was that Garmin uses the formula of Jack Daniels (J. Daniels, Daniels Running Formula, 2005, Human Kinetics, Leeds, UK) in predicting race times based on the VO_{2} max. In our books (Hardlopen met Power! and The Secret of Running) we have already compared Jack Daniels’ formula to our running model and we concluded that his formula does indeed lead to faster times.

In this article we will discuss Daniels’ formula and analyze the differences with our universal model. Finally, we will show that the differences can be explained and that our model gives the best results for most runners. To prevent readers from having to scroll back to our first article, we start with a summary of this.

**What is the basis of the**** Garmin Race Predictor ?**

Your Garmin first calculates your oxygen uptake capacity VO_{2} max based on your training data and in particular the data of your speed and your heart rate. The training data is then extrapolated to your maximum heart rate HR max. They do this based on research by Firstbeat Technologies (Firstbeat Technologies white paper, Automated fitness level (VO_{2} max) estimation with heart rate and speed data)

According to this article, the method is able to determine the VO_{2}max with an accuracy of 95%, provided that the HRmax is correctly entered.This is a very nice result!

Next, the Race Predictor predicts your race times based on this VO_{2} max. How does that work and how does it work in practice? And why is it now that the predicted times are too fast? As always, we also tested the results on the basis of our own results with Garmin watches; Hans the Forerunner 935 and Ron the Fenix 6X.

*Hans and Ron studying the results of their watches*

**How does it work? VO _{2} max**

One of the nice aspects of a running watch is that you get an update of your fitness every day in the form of your VO_{2} max. The watch calculates this from your training data and in particular the data of your speed and your heart rate. As we previously showed in our articles and books, the VO_{2} max is a very useful measure of fitness: the higher your VO_{2} max, the better your fitness and the faster you can run.

The table below shows the standard classification of VO_{2 }max values of men and women as a function of their fitness (which is also used by Garmin):

**How fast can you run with your VO _{2} max?**

In principle, your running speed is directly proportional to your VO_{2} max. In practice, the weather conditions and the course and distance will of course also have an impact. All these factors are covered in our books. You can also calculate them yourself with our calculators. The race time at various classic distances is shown in the table below (which is based on our universal running model and applies to ideal conditions).

**How does it work? Garmin’s Race Predictor**

Garmin’s Race Predictor predicts race times at 5K, 10K, half marathon and full marathon based on your VO_{2} max. How this works exactly is a Garmin trade secret. In principle Garmin takes into account the personal information (especially age and gender), the training history and training structure and the data of the training (especially heart rate and speed, divided into different segments).

On the internet we found a useful and interesting overview of the relationship between Garmin’s VO_{2 }max and the predicted times of the Race Predictor.

The first 2 columns of the table below are taken from this overview and show the relationship between the Garmin VO_{2} max and the predicted 5K time. Next we took that predicted 5K time as a starting point and calculated with our universal running model (as explained in The Secret of Running) which “real” VO_{2} max corresponds to that time. The result is in column 3 and appears to be no less than 3-5 units higher than the VO_{2} max reported by Garmin. The difference between the “real” VO_{2} max and Garmin’s is in the 4th column.

**How it is this possible? An analysis.**

The results of the table correspond to the experiences of many runners and coaches and can also be seen in numerous posts on the internet. Usually it is reported that you only get realistic times if you subtract 3-4 units from the Garmin VO_{2} max. Various running friends have reported to us that they also get realistic times if they take the Garmin VO_{2} max and use our calculator (and so not the Race Predictor). Coach Sander Schreurs even sent us an anonymized Excel with the VO_{2} max and Race Predictor values of his athletes running with a Garmin.

We can illustrate this with the results of author Hans (marked in yellow in the table). Hans has his VO_{2} max regularly determined at the physiological lab of SMA Midden Nederland in Amersfoort, the Netherlands. Therefore he knows that his VO_{2} max is in the order of 60 ml/kg/min (depending on the shape of the day of course).

His Garmin Forerunner 935 watch neatly shows the same value (depending on the shape of the day it varies slightly between 59 and 61). Hans runs times of around 18:30 at the 5K, which corresponds well with the result of our calculator with a VO_{2} max of 60. However, the Garmin Race Predictor predicts an unthinkable time of 17:01! Hans would need to subtract 4 units from his Garmin VO_{2} max of 60 to get a realistic time of 18:30.

As an aside, we remind you that Hans is currently injured and has therefore started cycling. Remarkably, since he started cycling six months ago, his Garmin watch has steadily increased his cycling VO_{2} max to a value of no less than 65! This is strange, because Hans has been running for 40 years and has only been cycling for six months, so why would his cycling VO_{2 }max be so much higher? But even stranger is the fact that the Garmin has simultaneously increased its running VO_{2 }max to 65! And the Garmin Race Predictor now predicts a 5K time of 15:53! That is completely surreal, Hans ran this time 35 years ago when he was around 30. For the age class M65, this would mean a new world record!

All this has made Hans curious whether his cycling VO_{2 }max is really so high or whether Garmin’s cycling algorithm is incorrect. He therefore plans to do a bicycle test at the SMA to determine what his cycling VO_{2} max really is. So to be continued.

** ****How is this possible? A possible explanation.**

Without information on Garmin’s exact algorithm, it is of course very difficult to explain what is wrong with Garmin’s Race Predictor. On the internet, however, we found the original patent of FirstBeat Technologies (Garmin’s partner). It states that the oxygen consumption during running is calculated with the formula: VO_{2} max = 3.5*v, where v is the speed in km/h.

With our universal running model, however, we have deduced that the correct formula for oxygen consumption (without air resistance) is: VO_{2 }max = 3.77*v. This corresponds to a difference of over 7%. Consequently, at the same value of the VO_{2} max, Garmin calculates a considerably higher speed v and thus a faster time. In addition, the air resistance, which is greater for faster runners, must also be taken into account. We have not been able to find whether and how Garmin calculates this.

**What about Jack Daniels’ formula, could that be the explanation?**

The aforementioned white paper by Firstbeat Technologies refers to a table from Daniels’ book for the relationship between VO_{2} max and race times. Daniels expresses the power of the human engine as the VDOT/VO_{.} (in ml/kg/min) as a function the speed v in meters per minute. He has empirically derived the following relationship:

*VO _{2} =-4.80+0.182258v+0.000104v^{2} *

The disadvantage of Daniels’ formula, of course, is that it has no physical basis. For example, he does not take into account the fact that the drag (in his formula v^{2}, in reality v^{3}) must be inversely proportional to the weight! In our books we have already compared the results of the Daniels formula with our model and we noticed that with the same VO_{2} max the calculated speed is a lot higher. We explained that by assuming that Daniels derived his formula for runners with very high running efficiency (Running Economy in ml O_{2}/kg/km), which means that these runners used less oxygen at the same speed.

To show the impact of this, we created the table below. The table shows for all speeds how high the VO_{2} max will be according to Daniels’ formula and according to our universal running model. It is also indicated how high the corresponding Functional Threshold Power FTP in Watt/kg is.

As an example, we use the data of author Hans again. He knows that he has a VO_{2} max of about 60 and he does indeed run about 15 km/h with an FTP of about 4.3 Watts/kg. This is all in line with our model. According to Daniels’ formula, however, with a VO_{2} max of 60 he should be able to run 18 km/h with an FTP of 5.2 Watts/kg! Hans would like to be able do that, but it is unfortunately absolutely unfeasible!

To illustrate the influence of running efficiency (RE), we also created the table below. This shows how high your VO_{2 }max must be to run a certain speed as a function of your running efficiency RE.

As an example, we use once more the data of author Hans. He has a VO_{2} max of about 60 and runs about 15 km/h. This corresponds to a running efficiency RE of approximately 210 ml O2/kg/km and that value has indeed been confirmed in treadmill research at Guido Vroemen’s SMA in Amersfoort, the Netherlands.

The table also shows that Hans could theoretically run 18 km/h with the same VO_{2}max of 60 if he were able to run super-efficiently with an RE of 80ml/kg/km. This is an extremely low value that is exceptionally mentioned for Kenyan and Ethiopian elite-athletes, who use very little oxygen at a certain speed.

Conversely, Hans could only run 13 km/h with the same VO_{2} max of 60 if he would run very inefficiently with an RE of 240 ml/kg/km. In our research with the Radboud University Nijmegen, we have seen this kind of values in untrained, inefficient runners.

**Conclusions**

Garmin’s Race Predictor systematically predicts way too fast times.

The explanation for this is probably an error in the algorithm that calculates the speed/time. It could be that this is because Garmin uses Daniels’ formula, which is derived for super-efficient elite runners with an RE of 180 ml/kg/min. We believe the VO_{2} max of Garmin itself is realistic.

There are 2 solutions to this problem:

- Use the VO
_{2 }max of Garmin (or that of a test at a physiological lab) in combination with our calculator (this gives the correct times) - Decrease Garmin’s VO
_{2}max by 3-5 units in combination with the Race Predictor (the order of magnitude will be correct depending on where you are in the table).