Normalized Power: Garmin vs Training Peaks

Feb 9, 21 12:39

Post #1 of 17 (2458 views)

Would appreciate some help understanding why the Normalized Power (NP) as reported by Garmin differs from Training Peaks (TP).

I was doing my training ride today and included in the workout were 8x1 min intervals @ 110 cadence and 2x5 min intervals @ 100 cadence. After completing the ride, I looked at the NP and other data for each of the intervals and noticed the the NP was different, and significantly so for the 1 min intervals. All other data, avg power, max power, HR, cadence, for each of the intervals was exactly the same. Why the difference in NP?

I find it interesting that the difference in NP for the 1 min intervals was higher in TP by anywhere from 11 to 22W, but for the 5 min intervals Garmin was higher, although only by a very small margin.

I should add that my FTP and training zones on my Garmin head unit and in TP are exactly the same, and for all of the intervals, there was no coasting (0 values) at all...it was constant pushing on the pedals.

For reference, here is the data:

Interval 1 (1 min) NP TP 299; NP Garmin 277
Interval 2 (1 min) NP TP 274; NP Garmin 262
Interval 3 (1 min) NP TP 298; NP Garmin 279
Interval 4 (1 min) NP TP 294; NP Garmin 277
Interval 5 (1 min) NP TP 288; NP Garmin 277
Interval 6 (1 min) NP TP 279; NP Garmin 258
Interval 7 (1 min) NP TP 298; NP Garmin 279
Interval 8 (1 min) NP TP 310; NP Garmin 290

Interval 1 (5 min) NP TP 256; NP Garmin 258
Interval 2 (5 min) NP TP 249; NP Garmin 250

Any insights would be greatly appreciated.

Jason

Re: Normalized Power: Garmin vs Training Peaks [JasonGeorge] [ In reply to ]

Feb 9, 21 12:46

Post #2 of 17 (2455 views)

Np on anything under 15 minutes is useless so ignore it.

Re: Normalized Power: Garmin vs Training Peaks [philg] [ In reply to ]

Feb 9, 21 12:52

Post #3 of 17 (2448 views)

philg wrote:

Np on anything under 15 minutes is useless so ignore it.

I don't disagree with your statement, but I'd still like to know why the difference exists. And, interestingly, why the difference seems to be greater for shorter intervals (this would suggest that it is not just a difference in the way NP is calculated).

Re: Normalized Power: Garmin vs Training Peaks [JasonGeorge] [ In reply to ]

Feb 9, 21 13:24

Post #4 of 17 (2426 views)

JasonGeorge wrote:

philg wrote:

Np on anything under 15 minutes is useless so ignore it.

I don't disagree with your statement, but I'd still like to know why the difference exists. And, interestingly, why the difference seems to be greater for shorter intervals (this would suggest that it is not just a difference in the way NP is calculated).

Because your interval is so short, any fluctuation is going to be magnified ie the start is going to be a big part of it, and did you wind your wattage up, or overshoot and ease off?

It really is a pointless exercise trying to analyze it, but read this and perhaps you will see why. Maybe Garmin calculates it differently or picks different start and end points

https://help.trainingpeaks.com/...804-Normalized-Power.

Re: Normalized Power: Garmin vs Training Peaks [JasonGeorge] [ In reply to ]

Feb 9, 21 13:33

Post #5 of 17 (2410 views)

In my opinion Garmins calculation of NP is incorrect for intervals. NP is based on a 30sec rolling average but Garmin includes wattage from your break between intervals in the calculation of NP for your interval.

Maybe TP is doing the calculation correctly using wattage from the interval only?

Last edited by: jth: Feb 9, 21 13:34

Re: Normalized Power: Garmin vs Training Peaks [jth] [ In reply to ]

Feb 9, 21 13:48

Post #6 of 17 (2389 views)

I haven't done the math but I'm pretty sure you're right.

I also feel much under 15 min has value when you're training 3 min power such as pursuit training.

Spiking the start of a 5 min interval has consequences on the rest of the interval and it'll be reflected in NP.

Re: Normalized Power: Garmin vs Training Peaks [jth] [ In reply to ]

Feb 9, 21 18:24

Post #7 of 17 (2329 views)

I just checked on TP for my short 40 second intervals from my race last Saturday, Normalized power is greater than average power indicating that TP does not use the 30 second rolling average in NP Calculation for short intervals.

Re: Normalized Power: Garmin vs Training Peaks [philg] [ In reply to ]

Feb 9, 21 19:06

Post #8 of 17 (2308 views)

Dr Coggan says under 5 minutes

Re: Normalized Power: Garmin vs Training Peaks [jaretj] [ In reply to ]

Feb 9, 21 23:25

Post #9 of 17 (2263 views)

jaretj wrote:

I just checked on TP for my short 40 second intervals from my race last Saturday, Normalized power is greater than average power indicating that TP does not use the 30 second rolling average in NP Calculation for short intervals.

Calculated by what unit and show in which web app?

Re: Normalized Power: Garmin vs Training Peaks [jth] [ In reply to ]

Feb 10, 21 3:23

Post #10 of 17 (2193 views)

Not sure what you mean by unit but the ap is app.trainingpeaks.com

Re: Normalized Power: Garmin vs Training Peaks [jaretj] [ In reply to ]

Feb 10, 21 4:01

Post #11 of 17 (2175 views)

The point is that TP appears to recalculate NP correctly for intervals (IMHO).

So you might see 250W NP for an interval on your Garmin head unit and in Garmin Connect, but will see a recalculated value of, say, 270W in TP

Re: Normalized Power: Garmin vs Training Peaks [jth] [ In reply to ]

Feb 10, 21 4:24

Post #12 of 17 (2167 views)

Yes, exactly, which doesn't make sense because TP says they do it one way and actually do it in their app a different way.

Personally I like the way TP does it in their app right now vs the way Garmin Connect does it.

Sorry if that sounds confusing.

Re: Normalized Power: Garmin vs Training Peaks [JasonGeorge] [ In reply to ]

Feb 10, 21 23:25

Post #13 of 17 (2084 views)

JasonGeorge wrote:

Interval 1 (1 min) NP TP 299; NP Garmin 277
Interval 2 (1 min) NP TP 274; NP Garmin 262
Interval 3 (1 min) NP TP 298; NP Garmin 279
Interval 4 (1 min) NP TP 294; NP Garmin 277
Interval 5 (1 min) NP TP 288; NP Garmin 277
Interval 6 (1 min) NP TP 279; NP Garmin 258
Interval 7 (1 min) NP TP 298; NP Garmin 279
Interval 8 (1 min) NP TP 310; NP Garmin 290

Interval 1 (5 min) NP TP 256; NP Garmin 258
Interval 2 (5 min) NP TP 249; NP Garmin 250

NP is defined to be calculated over a period of 30s. The general formula is, when I remember well, you get the 4th power of every thirty power-value of the last 30s, you calculate the average of that, and then calculate the 4th root of that average.
The problem is now, what do you do when the 30s not have been reached. One programmer will calculate NP there for example over the period which has already been passed, the other will for example just say NP=0 for the first 30s.
This will explain the huge difference here above in NP for 1 minute and the smaller difference for 5 minutes: the longer the total period lasts, the smaller the difference of the first 30s will influence NP. For a very long period, there should be no difference anymore, provided the formula used by the programmers is identical.

Re: Normalized Power: Garmin vs Training Peaks [JasonGeorge] [ In reply to ]

Feb 11, 21 0:07

Post #14 of 17 (2073 views)

Thanks everyone, your input on this has been very helpful. Just to be sure that I understand correctly, would the following be an accurate explanation of the difference between Garmin's calculation of Normalized Power (NP) and the way that Training Peaks (TP) calculates it?

First, it is important to note that Andrew Coggan says that NP should not be considered anytime an interval is 5 mins or less; instead, he says to look at average power for short intervals. Training Peaks says not to consider NP for any interval less than 10 mins. Now the question, and what we really want to know, is why.

TP calculates NP using a 30 second rolling average of power. This means that TP can't calculate a NP until you have 30 seconds of data from an interval and then the calculation would be only one data point. So for a 2 min interval, the rider effectively has an NP that is based upon 1 min 30 seconds of effort....you lose 25% of interval for the calculation and therefore, the physiological effect is not completely accurate. After 5 mins, the effect of the loss of 30 seconds at the beginning is largely eliminated.

With Garmin, it also uses a 30 second rolling average of power, but at the start of the interval, it will use 30 second from the previous (recovery) interval (I am not certain that this is in fact the way that Garmin calculates NP). One would expect that during the recovery power would be low, so in the end, Garmin's NP will be lower than TP. As mentioned before, with a longer interval, the fact that this is (in my opinion) wrong, gets washed out. And with a 5 min or longer interval we can see that Garmin NP and TP NP are indeed quite close.

Re: Normalized Power: Garmin vs Training Peaks [JasonGeorge] [ In reply to ]

Feb 11, 21 7:34

Post #15 of 17 (2016 views)

Just download the raw data from your head unit and calc NP yourself. You'll be able to figure out what TP is doing and what Garmin is doing.

Re: Normalized Power: Garmin vs Training Peaks [RChung] [ In reply to ]

Feb 11, 21 8:08

Post #16 of 17 (1999 views)

one other thought is that Garmin devices can apply smart recording meaning that the data is not evenly spaced. You can over ride this as 1 second recording. Maybe Garmin accounts for this in the calculation and TP does not? Robert's suggestion might shed light on this.

Re: Normalized Power: Garmin vs Training Peaks [RChung] [ In reply to ]

Feb 12, 21 5:18

Post #17 of 17 (1933 views)

RChung wrote:

Just download the raw data from your head unit and calc NP yourself. You'll be able to figure out what TP is doing and what Garmin is doing.

I like that idea. Before I get lost in figuring out how to download the raw data, any tips?

Our Partners

transition-DOT-area-AT-slowtwitch-DOT-com

Advertise with us

©1999-2024 Slowtwitch, Inc., and Slowtwitch.com
Reproduction of material from any Slowtwitch.com page
without written permission is strictly prohibited.