Zone 2: Heart Rate Adaption Speeds and Running vs Cycling

There are further parameters which may impact the Z2 HR/Power relation e.g.

  • athlete age
  • fitness level
  • HR reserve capacity (HRmax-HRrest)
  • temperature
  • sleep
  • recovery from previous workout

I’m just wondering if it is possible to create an algorithm which will cover some of these aspects with adding some paramters in the athlete profile. Would it be possible with AI?

1 Like

AI is generally used for processing large amounts of data and coming up with answers humans can’t see. So the only people that likely have the list you have is TrainingPeaks (not sure if intervals has sleep for example). Secondly I am not sure what you mean that “affect HR/power relation” today’s temperature or what you ate before the ride affect HR/power relation. If you mean affect how quickly HR goes up and down that is some what genetic but also is based on fitness level. I am not sure how anything else affect those. I would assume heart rate recovery (HRR) speed is similar to initial HR reaction time but I have no idea.

I am curious what problem you are actually attempting to solve with these thought. We do have another user on this forum that is really an expert in this stuff. He commented on some other maffetone/z2 posts. He could likely provide the best answer here. I have not seen him in a few weeks.

Researching and understanding HRR could provide some answers to this.

Here whoop says how to improve HRR


With my list I just wanted to raise awareness on other potential factors which may impact HR, HR goes up/down and Power. Of course genetics belongs to it as well. Just an extension to app4g statement “per-person dependent situation due to how HR lags vs power”.

Unfortunately I haven’t found any posts regarding HR for cycling from Maffetone himself.

My two sons drive a small SW company which provides apps for hospitals. One son has a masters degree in Computer Science the other a master in Medicine. Since more than 15 years they are using AI , they have some expertise. E.g. they say that the application from TR could be easily done with algorithmic programming. I’m not sure if XERT uses AI - probably it’s just an adaptive algorithm and it works fine. Based on this I don’t think AI isn’t the way to go, as well for TODAY TD functionality. The current algorithmic implementation is sufficient from my point of view.

When I am re-thinking, probably a simple approach would be sufficient for TD. Just the two user-parameters for HR up and HR down: currently Prod is 60sec up and 30sec down (Beta 120sec and 60sec down?). The user can figure out what suites him best, regardless what his genetic, his age, his fitness level is, …

This are only my irrelevant thoughts. We may finish this discussion as I don’t want wasting your valuable time for TD development.


Not irrelevant at all. TD is a product of our customers. Very few ideas are my own original ideas. I am very impressed with myself if I ever do have an original idea :slight_smile: something is not just a slight variant from someone else’s idea. Anyway. I agree 100% with your sons. Nate has a video describing how they use AI, it’s machine learning. That is not a real time activity and it’s not fast. My guess is they use AI to get some ideas and everything is programmed with rules/algorithms. Xert is all rules as far as I know. I have given it a lot of consideration and testing and I don’t really see a clear place for AI in training other than translating human questions to answers but those answers still driven from rules or data. I am sure the day will come that AI can play a deeper role but we are not there today. Just ask chatGPT any question about something you are a true expert and dig in deeper. It’s just wrong 1/2 the time when it gets to the details.

So yes your thinking about 30/60 vs other is exactly what I was thinking but I just tested 120/120 with a user and it did not solve his problem so I think something else is going on. I do think this still might be the solution after I figure out his issue.

Please always contribute your thoughts. It really helps. We need to get your sons more excited about TrainerDay so they can help too :slight_smile:

Maffetone does not talk about cycling much at all but everything related to running should apply the same to cycling. But there could be some nuanced differences. He did train Mark Allen the triathlete. Mark wrote some books on this too.

The biggest difference is runners and triathletes fall apart quickly with too much duration + intensity. So there becomes no choice. Cyclists don’t have this problem usually but can still reap the same benefits. Is it possible AeT is different running and cycling.

If someone is motivated enough to read this, they might find the answers.

Direct link to PDF

I do see by reading the conclusion they point out that ventilatory thresholds can change based on different stimulus and time lines for cycling or running as well as VO2max being sport specific. This would indicate their is potential that cycling AeT is different than running and additional reading that AeT might increase faster for cyclists. Hence if you have been cycling a while then it could be above this 65% Maffetone limit. I would still use Mafftetone as the start point, see improvement and experiment.

A very plausible answer came up on the Intervals forum today, explaining why AeT can be different for multiple sports. Depending on which one you have more experience with (more hours done) it can be either running or cycling that is lower or higher:

‘more highly trained muscle can process lactate better’

If you consider AeT based on 2.0 mmol/l that argument would be very solid. I think this 2.0 figure is just a rough estimate for simplicity and it’s really the lactate deflection point. Talking to the guys at Moxy, that’s how they see it from an SMO2 perspective and this is clear in SMO2.

You research this stuff more than me so might know better. But that deflection point could be a different lactate levels also for different sports making it even more complex but regardless what you are saying makes sense and seems reasonable. I think of AeT being HR related (ventilatory seems related to HR…) but really I guess lactate is the more accurate definition?
Assuming it is, it seems logical that you could have different AeT for different sports. I just wonder how much? I would guess people that run a lot and ride a lot it’s likely closer than people that are much strong at one than the other. But it’s all just a guess. Maybe cycling is just higher than running.

Here he defines AeT as the deflection point in lactate and not strict 2.0 also.

I will ask Andrea. Since he spends many days just measuring athletes in a lab with a mask or his team does then he might have a better idea :slight_smile:

This brings another topic. So AeT=LT1=VT1, meaning VT1 is based on airflow. I would assume their is a direct relationship between air flow and heart rate (a total guess).

This guy is pointing out that in a well trained athlete that VT1 can be 10% less than LT2.

So when you hear San Milan or Seiler talking about figures that sound more like 80%+ of max HR then they could be speaking very directly about advanced athletes. Where Maffetone’s 180-age which is usually similar to 65% of max HR could be for average athletes.

So my guess is it is less sports related and more level of eliteness being the major factor. But I would also guess familiarity of a sport would make a big difference and finally another guess (my last and final guess… ) is that running vs cycling would be a smaller difference in an advanced triathlete for example.

Really to be clear we are talking about lactate levels vs HR in different cases. Can we use HR for a lactate proxy accurately? And secondly is how close is our AeT from an HR perspective to our LT2…

There are probably more factors but this seems relatvant

  1. Level of athlete
  2. Possibly even VO2max
  3. Familiarity or current training level with specific sport
  4. Sport specific muscle differences.

Andrea said he will give me his thoughts soon.

I get a big difference in HR between Maffetone and 65% HR. I’m 43 with no adjustments so my Maffetone HR is 137 and 65% of a 185 HR peak is only 120.

I’ve done the TD HR quick rides typically at 125 BPM. I did one last night for 90 minutes at 131 BPM. The 131 ride was moderately harder. I haven’t tried 120 BPM.

I did 131 BPM because typically TD HR rides reduce power when I go 5 BPM over setpoint so that keep me at the Maffetone limit.


I would go with Maffetone, not 65%. I made a mistake when I said most people should be similar to 65%. What I really mean it 65% should be a very safe number for almost everyone. I have been saying 65% because to the general public I always think saying 180-age would scare many as it seem similar to 220-age to estimate max HR which most people know can be highly inaccurate.

My personal experience and what I read from many in the past in Maffetone groups is going much lower still produces good results so 120 might not be a problem. If you are not improving HR to power than reducing HR is worth trying.

Not exactly the same, but they are so close that for practical use, you can consider them the same.

That’s the whole point of doing Z2 volume. If AeT is lower then 85-87% of AT, it will be more efficient (performance wise) to train your aerobic base. When you get up to 90%, VO2max training will be beneficial for performance. Then again, go back to aerobic.
Now, let’s be clear, this is a point that will be very hard to accomplish by the average recreational athlete. And you need a met-cart to measure both points with that precision. That’s more for the pro’s, but an interesting approach.

Sure I understand, but it seems the level of your aerobic base based on training is the main factor in % of HR vs lactate (Actual AeT) not the sport running vs cycling. So trying to say Maffetone is only good for runners and San Milan’s higher numbers make sense for cycling seems like it is not likely correct. It would seem more likely that the difference is based on the caliber of athlete rather than the sport. I know you are using Maffetone for cycling but this would be an argument as to why everyone should consider using these lower numbers for cycling.

I would guess there is some difference between running and cycling AeT but that is more for the advanced athlete that is really trying to optimize Z2 and doing very precise testing.

Do you agree with my thinking here?

I agree that this doesn’t sound correct. On the Intervals thread someone came up with the question, and the guy was a runner, just starting cycling to increase aerobic time in a safer way. He saw the inverse, his cycling RPE was way higher then what he normally feels during running at the same HR.
Maybe we should calculate MAF per sport?

Oh I see, yes that makes sense taking the -5 for a just getting back into it sport, that sounds reasonable even though you are 0 or +X for your main activity.

When it comes to determining ‘zones’, I really like this video:

It first of all makes clear that we should be careful when naming a zone without telling what system we mean.
And it actually builds up from physiological anchor points.

This one adds breathing markers to the mix.


Just wanted to show what an extended Z2 phase can do for performance below AeT. Here’s a chart for running performance over 17 weeks almost exclusively at AeT (122 bpm for me). Top chart RED is avg 1k running GAP and GREEN avg 5k GAP. Second chart is GREEN running efficiency in mtr/bpm, BLUE is running fitness watts/bpm and GREY is running technique in mtr/watt.
Third chart is BLUE/GAP, PINK/cadence and PURPLE/Power.

The first period (1), about 9 weeks, very little is happening to performance, but total distance per week is building up. Then, all over sudden, performance gains kick in and a steady raise is noted for the next 3/4 weeks. Got a light cold then (2) and stopped training for 4/5 days, that’s the short dip. And when I was healthy again, continued the normal training process and I’m very happy to see that the gains keep evolving.
Over the period of 17 weeks, my Z2 running GAP is about 1 min/km faster and the total distance per week of 60-70km can easily be handled. Even on longer runs around 15km, there is no significantly reduced speed at the end and decoupling for HR/Power is minimal. I do 2-3 MTB and/or indoor cycling sessions on top of the 6-7 running sessions per week. Coping very well with that load. That’s confirmed by HRV and HRrest, which remain almost stable.
I’m now focussing on increasing cadence some more, still within the MAF HR range. That should be beneficial to even more Z2 speed. I don’t plan to change a lot in my training regime as long as the gains keep coming at the same rate. During all those weeks, only 4 sessions have been done on higher intensity. One run, just because I felt like doing it that day. One MTB ride, plenty of mud, where it is basically impossible to stay in LIT and 2 indoor ‘tempo’ interval rides.