MAP vs CP thresholds

I’ve commented previously on the seemingly high MAP thresholds for the new system, but just realised that this analysis makes the point very clearly.

This table is percentiles from the intervals.icu database, showing power levels vs time (for men, this table). Ignore the green line, that’s just my recent data.

Compare the 5 min to eFTP values. What they call eFTP is actually 3-parameter CP here. If Zwift is defining MAP as 5-8 min power, it’s clear that the thresholds they are using are absurdly high. At the A/B border, 4.0 CP is about 74%ile, However 5.4 even at 5 min is well over the 90%ile (and it would be about 95%ile for 8 min).

Is there a single rider who exceeds the MAP threshold but not also the CP threshold? If not, the entire MAP calculation is a complete waste of time. Did no-one at zwift even check whether the thresholds were set at a sensible level?

In order to be meaningful, the MAP thresholds should be reduced to a level that has some relevance. I suggest using this table, or something similar, as a guide.

1 Like

I think you are on to something here. It would be very interesting to know how many are in their category because of MAP vs CP. If the components are weighted equally, there should be a 50/50 distribution.

By using this table, splitting categories in 25% percentiles and averaging 3, 5, and 10 min power, new boundary values could be changed like:

A/B: 5.4 => 4.7
C/B: 4.1 => 4.1
D/C: 3.9 => 3.6

Have they actually said they are calculating MAP as 5-8min power? If so, that’s incorrect - MAP is closer to best 3min power. 5-8min pace is VO2max territory.

Thank you for pointing me towards intervals. Have only been looking at for past hour so still do not fully understand. I wanted to check what eFTP was - my gut feel was that it is an estimate of your FTP. Most information I have read suggests CP is a little above FTP but probably by such a small %age that it is not worth a jot.

I found two items in their forum posts which might be interesting to others. These were both posted by David (intervals.icu):

“Sure. You need a max effort of long enough duration. This is used to place you one one of many pre-defined power curves modelled from real world data. So if you can do 330w for 12m that will put you on a model curve that has 330w at 12m. The 1h point on that curve is your eFTP. It is essentially a generalisation of the “95% of 20m power” rule of thumb.”

I take this to probably mean that they take your strongest point on your own power curve and then overlay their own power curve to arrive at your potential best eFTP.

If you want to identify your eCP then David’s answer to someone else may help.

"On the /power page you can also change to “eCP” instead of “eFTP”. This uses the same curves as eFTP but reports the model CP value instead of FTP.

Screen Shot 2022-02-18 at 21.42.33908×870 84.2 KB

If you want really accurate results you need to do max 2-3m and 10-12m efforts and capture those with “Define Curve” and choose MS 2P or Morton 3P:"

I don’t however see that changing to eCP actually changes the table away from eFTP.
EDIT : Just realised that changing to any one of the Power Models doesn’t change the Power Profile chart. I suppose what I’m trying to ask - Is it ok to say eFTP is 3-Parameter CP here?

i absolutely agree. Going from using one power metric to using 2 is pointless if, using the ‘OR’ logical operator, one of the power metrics is not broadly in line with the other.

I fully understand Zwift’s reasons for not showing us exactly how CP and MAP are being calculated. However it makes it impossible to check whether everything is correct without testing the precise workings. Difficult to report fixes if you cant test whether a fix is required.

I think I have said it elsewhere but this is now a more appropriate thread so I will say again. I don’t fully understand Power curves and appreciate they are not always linear but mathematically the relationship between the three category boundaries just do not look right to me -

4.0 cp OR 5.4 map … 5.4/4.0 = 1.35
3.2 cp OR 4.1 map … 4.1/3.2 = 1.28
2.5 cp OR 3.9 map … 3.9/2.5 = 1.56

3.9 map is relatively much closer to 4.1 than 2.5 is to 3.2.

I would suggest that MAP will never come into the equation for the C category calculations.

i do… 6.2wkg 5 min 345w, with 248cp at 55.5kg. if they went off my 3 min power of 7.3wkg my CP would be even lower

edit: i should add that there are probably only a couple of people on zwift with a power profile that looks like mine combined with my weight

Um…but your 248 is already 4.5W/kg, so you’re over that threshold for A anyway, it’s just the 250W cut-off that puts you in B. Sure your estimated MAP is a bit higher percentile-wise than your CP, so is mine according to the green line. But only by a modest margin.

the point is though that only the 5 min limit is what would stop me from joining B, so it does serve a purpose. also, if i was to lower my 5min to 5.4 by not doing any max testing on zwift, i guess my estimated CP would rise (according to my basic understanding of how it works) and i would end up in A anyway.

but i think there are a few light riders out there who are capable of a 5.3-5.4 5 minute while having a lower than 250 CP. cyrus, who posted in here the other day would probably be one of them, many japanese zwifters also.

Looking at this data from cycling analytics it look like a top B rider at 4w/kg FTP should be able to do 5.01w/kg for 5 min.

So I would think the 5min MAP should be above that to make room for the sprinter types.

95% of 20 min watts is often higher than eFTP. With eFTP more in line with 60 min FTP IMO. It is better to use eFTP as more real FTP estimate.
5 min W/kg of 4.74, as in the table of the OP is about right I think.
@_JamesA_ZSUNR You seem to be borderline B/A (75%) for the 3, 5, 10 min W/kg. Does that match your experience?

Remember that it’s your *modelled MAP values (based on your CP and W’) not your actual.

Parking that though, I agree I doubt there are many breaking the MAP limit. Zwift has access to everyone’s data and should be able to look at this very easily. They could very easily see what the impact would be of either lowering the threshold, or using a different power duration (personally I think 3m-5m would be far more applicable to Zwift racing, even if it doesn’t link as nicely to VO2 max (still not really sure why that link needs to be made).

@JamesBailey @xflintx any idea if this sort of modelling is happening? Are there many people upgrading from hitting the MAP threshold rather than CP?

2 Likes

Incorrect for how it is used in this guise. MAP is tied to a duration, it’s the maximal power output at VO2 max (maximally aerobic) and therefore has to be constrained to time (check out t-MAP).

They use MMP (mean maximal power, average) between 5m-8m currently to ascertain the MAP value, as this has close correlation with VO2 max values (I’m not sure I agree that there is much value tying it to VO2 max values, but that’s a different debate). I agree that MAP, if you just google it, tends to refer to a test result from a ramp test, but it is OK to say 5m-8m modelled MAP too.

I’m not sure where you get these numbers from.

My FTP is considerably higher than my CP (15w). They are measuring different things.

The assertion that you can estimate 5m power from FTP is also completely wrong, that’s kind of the point of moving to CP and W’.

High W’ can compress CP, and vice versa. So having an understanding of someone’s CP and W’ means you don’t just understand their critical power threshold (tied to significant physiological state change, but we don’t need to dive in to that) you also understand their ability to execute work above CP, which is a key differentiator in performance, especially in zwift races.

James raises a good point about whether many actually exceed the MAP value. However his chart shows eFTP not CP (you can’t get the chart to show CP unfortunately)

What we don’t get from this model is any idea of repeatability, but that hasn’t been well modelled anywhere from what I can tell.

Look up W’bal and have a look at your own W’bal from a Zwift race. Also look in to Xert if you really want to go down the rabbit hole. Interesting stuff

P. S. Can you tell I’m rationing my forum time :joy:

1 Like

@gloscherrybomb Sorry. I haven’t set aside time to dive into these concepts and should not comment on things I know little about :roll_eyes:

1 Like

It’s OK to comment, just not make assumptions on how it works. There’s a lot of confusion on here because of that. That’s why I tried to explain in simple terms on that other post. Guess I missed the mark a bit.

1 Like

Another issue with the chart posted - look at the different between eFTP and 60m power. It is just an average of actual performances over durations, not modelled maximal values, and not CP.

The caption for the table explicitly says:

“Estimated FTP is calculated using power curves from FastFitness.Tips and Morton’s 3 parameter critical power model.”

I don’t see how you can interpret that as “not CP”. It may not be using precisely the same method as Zwift, of course, and since they insist on keeping the details secret, there is no way we can ever know that.

Intervals.icu can show you eftp or cp depending on which you select, they are different values

1 Like

Yes, in the sense that I can hang around in ordinary/weak A cat races though didn’t have much fun this season in the ZRL which was all sprints, even more so than usual. In the mixed cat ZHR masters I usually beat about as many As as there are Bs that beat me :slight_smile: Haven’t done any for a while though.

Of course I could have a sprint (or not) irrespective of these numbers. It just happens that I don’t :slight_smile:

1 Like

Estimated FTP uses a number of different things, that why David, who created intervals, specifically differentiates the two, and you can select to view CP separately from eFTP. So yes, pretty easy to interpret it as ‘not CP’.

Edit - as Rich says above.