Better than a dartboard? (probably)

Not as accurate as a standard 2x20 minute or 5min + 1x20 test?

Does it just take 95% of the best 20minutes it finds?

I think the Strava FTB calculator has little to no connection to actual FTP. The Strava FTB is based upon experience over time related to power levels over time, which is materially different than setting up a test that tells you how high you can keep your power for one hour. For me, the FTP actual test vs Strava is off by 40+ watts and because most of my rides are long, the Strava number isn't based upon any ride that is intended to be reflective of an FTP test. At IMTX the first 42 miles were at power of 249 with normalized at 255 and Strava has my FTP at 249, and I can assure you I wasn't at FTP for the first 42 miles of 112. I would ignore it.

My Strava estimate looks to be pretty damn close to what I use as a working FTP right now.

So what you are saying is that the strava FTP calculator is not accurate if it has incomplete information. Well, I think that is obvious. If it had all of your rides, I would bet that it would be much closer.

I wouldn't use Strava Power Analysis ever.... TrainingPeaks / Golden Cheetah are the two devices you should use.

---

Carson Christen
Sport Scientist / Coach
Torden Multisport

As far as i know, its just taking 95% of your best 20 minute effort.

I don't know what it's doing, but I can say for certain it's not doing that. My reported FTP is higher than my max 20 min power. I have a race effort of about 15 min that it must be queuing off of in some way.

---

Powered by Dalzell Coaching

Mine is exactly 100% of my 20 minute power (but I don't think that's how it's calculating it). It's surely 10% higher than my true FTP. Their algorithm is crap.

I just did Sufferfest - Rubber Glove the other day and looked at my Strava now... my estimated FTP from Strava is ~96.6% of my 20 minute effort in that workout (normally we use 95%, of course). So mine looks ok to use as a starting point. Certainly a bit better than "in the ballpark."

They just use the critical power model and your power curve. That's known to over-estimate FTP. I wouldn't call it "crap." It's just the same model that just about everyone else uses to estimate FTP from a curve.

Why not? It's not rocket science. Strava has by far the cleanest presentation and ease-of-use for the basics of power.

I use Golden Cheetah on a ~weekly basis, but Strava is just so dead-simple that it's a great daily tool. And I've never noticed any "errors" in the data. Power data is power data.

Its usually within 10 watts for me; I'm not sure what formula they use but I don't think its a percentage of 20 minutes.

Considering everything but a pure hour test is guesswork, I'd say its good enough. Most of the time I adjust my FTP best on RPE of my set zones.

I was more in the terms of Power Prediction Analysis, sorry! When I come back from a session and it's said I've AVG 450W, when it was actually 300W. That's a problem :D

---

Carson Christen
Sport Scientist / Coach
Torden Multisport

Well the quality of the prediction is highly dependant on the inputs. If you do not good short power intervals, it will really overestimate your FTP. If you only do short power intervals and 3 hour steady state rides it will probably underestimate your FTP. Or as the poster in this thread showed that only long uploading long rides results in a very underestimated FTP. If you fill in the curve well it can be more accurate than the 95% of the 20 minute test, because some people may be 93% of 20 min and others may be 97%.

And WKO 4.

WKO 3+ is good right now, we are beta-testing WKO4+ and it's gonna be great too. Andy C is going all in with this one. It's really neat to see the features and what will be available!

---

Carson Christen
Sport Scientist / Coach
Torden Multisport

Same

I don't use the Strava number, but every now and then I look at it and it is very close to what I use in WKO and GC.

The Strava Estimated FTP is close for me too.

It's probably close if you've put good data in - eg some shorter and longer all out efforts. Without those, it just assumes some are all out and it can lead to bad results.

Strava screws up other power analysis royally though - they recalculate all old rides if you update your FTP, so all your old rides' analyses are wrong. This is especially bad in the fitness/freshness model - it's complete garbage the way strava does it.

---

Powered by Dalzell Coaching

That depends upon the duration of the tests used to calculate critical power.

If you use tests of longer duration (as originally recommended in the scientific literature), then critical power is equivalent to FTP (and MLSS, the respiratory compensation point, the NIRS breakpoint, iEMG threshold, etc.).

OTOH, if you uses tests of shorter duration (e.g., GoldenCheetah almost invariably picks points at ~1 and ~10 min), then the calculated critical power will be higher than FTP (and hence also higher than MLSS, etc.).

You ain't seen nothin' yet. ;)

Important clarification, yes.

I only just discovered this, but also that you can set the short and long "search window".

Based on your comments, would a 5min short be better than a 3 min short?

Thanks for your time.

Yeah, that's the issue. We ain't seen shit of WKO4 yet despite plenty of assurances to the contrary. ;)