Cadence in the literature - what Jem Arnold is learning

If you are interested in the topic of cadence, I thought this was an interesting read:

https://twitter.com/jem_arnold/status/1671513623498457088

Well interesting if you really are not Coggan, struggle with keeping all the basic physiology concepts in our head, but enjoy getting some (any!) perspective on what has been looked at and studied.

Some may know of Jem from his blog.

And some forum posts under the @SpareCycles name.

10 Likes

Critical Power is basically FTP

This is the first study mentioned

Where “10 collegiate men athletes” were tested and average threshold was 9% higher at 60rpm than at 100rpm.

My first question, are those 10 collegiate men athletes representative of cyclists that train? I was unable to pull up the full text of the study.

2 Likes

My first thought is that 20% (or 2, so large margin for error) in the first post actually have higher CP at 100 rpm vs 60rpm. Any number of possibilities could explain the variation. Depending on how representative, the other subjects could not be highly trained and could have low self selected cadence–or a bunch of other possibilities.

My mind wandered a bit going through the rest of the thread, but I did notice the first three graphs in this post: https://twitter.com/jem_arnold/status/1671513638916734977 which I think is saying that the higher the wattage, then a higher cadence is better.

So, maybe, most of the ten athletes in the initial study are not representative. But I also do not have access to the full article.

1 Like

yes, that was a good graph and I generally agree that is true for myself.

However, one of my best all-time threshold pacing efforts was 67rpm on a 10% grade. I have an opportunity to ride that grade again in a couple weeks, so that got me to thinking :thinking: All my other short (20-min) and long (30-60 min) efforts are somewhat split between 80-82rpm and 88-90rpm.

Thanks for linking. Has been an interesting rabbit hole to dive down. I think the conclusions don’t change much on the applied side for our training/racing, but hopefully we can understand things to a deeper level.

I like the “ok, but why?” questions like this :grin:

Here is the full paper of the first study referenced
https://sci-hub.ru/10.1016/j.resp.2014.12.008

On the other hand, a 9% difference in threshold from only changing cadence??

Very true! Important observation. There is a lot of variability here, but this is not the only study to show these directional results. All the refs I’ve provided are representative, not unique.

Yeah, like most exercise physiology research, it’s done on untrained male uni students, not trained athletes. I do think it’s important to draw a distinction between ‘exercise physiology’ and ‘sport science’ in terms of research objectives. This study was evaluating this particular intervention to gain insight into human physiology, not necessarily represent elite sport. Sometimes we have to grab the table scraps from the literature and extrapolate to our all-things-considered very niche application.

That being said, I don’t have a citation on-hand looking at CP/thresholds in trained+ cyclists, but my general synthesis is that (1) lower cadence = more efficient holds true pretty broadly, but more importantly (2) there is a ton of variability and interaction with workload, intensity, absolute forces, body anthropometrics, cycling experience, biomechanics… etc.

And anyway, our brain has already worked out what our optimal cadence is: the cadence we use when we’re not thinking about cadence

Yeah this is super important. Energetically optimal cadence, and freely-chosen cadence - which are two different things - both scale linearly with increasing workload.

In fact, workload alone seems to explain >90% of variance in metabolic efficiency. Fibre contraction speed (which cadence is a large modifier to) is the next largest contributor, but it’s all marginal.

It’s almost like the deeper you read into the thread, the more pieces of the story will (hopefully) come together. Trees turn into forest :wink:

4 Likes

I’ll add here… wild speculation, 60% confidence, but I think the best application for deliberately adopting lower cadence to improve performance might be virtual racing.

Short cranks, open hip angle, front wheel raised, lower cadence, lower inertia (low gear)… and a few other sneaky optimisations I could think about :thinking:

2 Likes

:+1: thanks for dropping in! Instinctively I lower cadence on fast tailwind sections because even surfing above/below threshold it seems to lower breathing & heart rate, so I can push longer and farther. Same for steep pitches but I’m low W/kg and am forced to drop cadence.

1 Like

As well as the absolute Critical Power value, cadence must play a big role in how long that power can be sustained for.

I was interested to see this table of track hour records, with average cadences calculatable from distance & gear size

Source Technologies for speed in 20th century competitive cycling

4 Likes

Very nice table! Yeah I wonder what the decision making is going into choosing gearing? Maybe some trackies can comment. Would love to hear Dan Bigham’s thoughts on this. Interestingly, he’s commented about choosing a (marginally) lower cadence at 95 vs 105 rpm

Recent 2x20 workout: I guess I like 91rpm

Curious to try lower next time and see how it feels. I’m nearly certain it will feel much harder for me. I do use a lower cadence indoors, often standing, for short Zwift racing climbs. But nothing like the 60rpm or lower you see some guys using in their Zwift streams.

You can usually find anything at scihub with the DOI number.

1 Like

Flipping it around to low cadence, here is the WKO workout search term for “20-min power > 95% ftp, and average cadence < 70rpm”

(meanmax(power, 1200)>(0.95*bikeftp)) && (workoutrange(meanmaxfrom(power, 1200),meanmaxto(power, 1200),avg(cadence))<70)

Results:

and doing some searches for other cadences, here is my summary:

cadence number of 20+ min ~ftp efforts
less than 80rpm 13
80-90rpm 42
greater than 90rpm 17

Some outlier examples of lower cadence at threshold… All on grades except for the 62-min “tour-de-torque in flatland” :rofl:

May 2021: 17.5 minute climb up 10% grade:

average cadence, power, and torque in lower left.

That effort saw my lowest ever HR for a threshold effort. Normal HR-to-power on the ~6% 35-min climb that preceded it. :man_shrugging: The only thing unusual that day - my ride was delayed after a morning searching outside the hotel room for our “lost” cat. After 90 minutes of searching we went back to the room and opened the (unused) dresser drawer, and she was hiding behind the drawer :rofl: But it was a lot of stress :scream: Maybe I need that type of stress before the next field test or pacing effort!

April 2020: 62-min tour-de-torque in pancake flatland (perfectly flat except for a freeway overpass)

Nov 2018: ~30-min climb up 6.8% grade:

Jan 2018: ~20-min climb up 6.2% grade:

Nov 2017: ~20-min climb up 8.6% grade:

2 Likes

With that many efforts, can you plot cadence vs power of 20min efforts? Or, what I’d find even more interesting:
Av power of 20 min effort / FTP at that time
vs
Av cadence of effort / preferred cadence.

That is, normalise the power to your FTP to take out any shifts over time. And I’m wondering if you get a curve for cadence where being at preferred cadence results in max power.

Might also be interesting to see what it does to HR, to get an idea of the load on the body, but I’d worry HR is affected too much by temperature (time of the year), fatigue, etc.

1 Like
Date Temp FTP Avg Power NP IF Avg HR Avg Cadence Avg Torque Duration
March 19, 2017 68F 275W 284W 292W 1.06 158bpm 80rpm 34.6Nm 29:10
April 9, 2017 58F 275W 272W 276W 1.00 156bpm 83rpm 31.6Nm 53:18
May 7, 2021 71F 264W 249W 258W 0.98 145bpm 75rpm 33.1Nm 34:37
May 7, 2021 64F 264W 273W 274W 1.04 149bpm 67rpm 39.2Nm 17:37
Jan 22, 2022 52F 272W 276W 276W 1.01 157bpm 79rpm 33.8Nm 32:10
Jan 24, 2023 52F 272W 287W 288W 1.06 158bpm 79rpm 34.8Nm 20:34
May 4, 2023 58F 275W 301W 299W 1.09 154rpm 89rpm 32.4Nm 8:05

so pulling data from a table like that?

Interesting.

I was talking to my coach about this today… Was doing a SS/threshold Wo (93 to 100 ftp, 4x10) and I usually struggle when trying to keep the cadence above the 85 pushing to 90s… Today I decided to go by cadence feeling and it settled about 79 for the 100s, 82 for 95 and 85 for 93%… the best i have felt doing these kind of wo in forever

2 Likes

It’s wild to me to think of being able to sustain higher Watts at threshold or above at a lower cadence.

If I drop down below 90 I immediately start to feel the burn build quickly and chances of repeating intervals is pretty low. Whereas in the 95rpm range I can stay on top of things.

2 Likes

It’s wild indeed.

On running i am a very high cadence person… Usually hovering in the 190 to 200 and racing in the 210 to 220 steps per minute…

Biking is the opposite… I feel much much better at a lower cadence. I have 0 issues doing 20+ minutes at SS/threshold if I keep the cadence at 65 to 75

1 Like

I remember there was a trend for low cadence in IM tri for a while too.

It just seems to shift the load so much to the muscular side of things to me. I could turn the pedals over at 60-80rpm at 200W for a long time, but at 300+ I feel like I’m not “strong” enough and it just blows me up.

I found the same when I was a triathlete, I used to ride at a high cadence during the IM bike leg too to save my legs for the run. I just figured it was to do with build (not muscular enough)

Yes, I was rhinking you could plot cadence vs IF, for example. However I don’t think there is a trend observable in that dataset. It is probably not large enough, and the variability in cadence not that high.

If you look at HR vs IF or HR vs cadence, I guess the main point is that lower cadence leads to lower HR.

The last effort in that list seems to be some sort of outlier - IF is pretty high, cadence is high, but HR is relatively low. However this seems to be a short 8-min effort, which might not be comperable to the longer efforts

Yeah I’m working on automating data extraction and plotting. The last one was a short pacing effort and right now my HR takes a good 8-12 minutes before it starts ramping up towards LTHR.

3 Likes