You are hereForums / Preparation / Geek Gear (GPS/Comps/Lights/etc.) / New Strava "Fitness and Freshness" graph

New Strava "Fitness and Freshness" graph


By Pete B - Posted on 17 November 2014

NB: Originally posted elsewhere on the Global Riders Network and appears via syndication.

I know most people on here use Strava and many of them are premium users.

For those that are, have you seen the new Fitness and Freshness graphs under the training tab? If so, what do you make of them - useful training aid or a load of old cobblers?

It seems to me that only the longer rides have much effect. I try and do at least 1 hour long interval session a week round Centennial Park but they have very little effect on the graph but definitely improve my fitness/strength.

Haven't had a good look, but it seems to use TRIMP, a training stress measurement based on HR.
http://www.trainingimpulse.com/banisters-trimp-0

It's handy because you can take your HR monitor to any bike and still log results.

HR based measurements are usually more accurate for longer rides/intervals. Doing 15 sec sprints you can generate massive power, but your HR might only reach 90% max. Do ~10 of these properly and you should almost die, but your HR graph won't look impressive at all.
Power is the best way to log data and most people use the TSS metric (training stress score).
Same idea as TRIMP, but power never lies.

I use both TRIMP and TSS to log my performance. I ride 4 bikes, but only two have power meters.

It is based on sound science. It will be based on measures like Chronic Training Load, and some others i can't remember off the top of my head. Fitness takes longer to build and lose (weeks), but fatigue/freshness runs on a cycle that is as short as a few days.

You actually lose a bit of fitness in a pre race taper, but because your fatigue levels are.low (if done properly) you are a lot fresher and your performance will be better.

I hope that makes sense.

You can get the same data for free on Training Peaks and Golden Cheetah if you know what measures to look at.

Well it should be based on sport science, not sound science.
Just my humble opinion.
Sticking out tongue

Yeah, regardless of which metric you use, they all try to use a combination of short term stress (STS), long term stress (LTS) and stress balance. They also like using different terms for the same thing just to keep you guessing. Acute, chronic, balance, form, fitness, freshness.

As alluded to, if you have some decent data behind you, (and you know how to use that data to train and taper etc), it can be an invaluable tool.

I find when my STS gets above a certain value I'm very likely to get sick.
Likewise, there is no point doing a tough ride when that STS is high. Especially for high power intervals its best to be well rested.

Short answer: it is accurate enough to use for Amateur and semi-pro.

The L O N G answer:
Strava calculates Fitness, Fatigue, and Form based on but modifies an impulse-response model.

The impulse-response model Strava uses is a reasonable basic model for the amateur/semi-pro athlete. Use the Strava model as a guide, not as a 'bible'. The impulse-response model they (and others with variations) utilise is almost certainly the best (IMO) currently available to the amateur cyclist despite its shortcomings.

Lots of stuff around on the impulse-response model Strava's system is based on - so try this one http://home.trainingpeaks.com/blog/article/the-s... - (but maybe ignore the math). Caution - it justifies the model Training Peaks uses. It covers the robustness of the impulse-response model but also some of it's problems, such as:
*no upper limit to performance, the more training you do the more performance you will get (once fatigue reduces), which is unrealistic.
**More measurement may be required to make the model more accurate - suggested that maximal performance be tested every four days for optimal accuracy - impractical, so is the output of the model a guide? (see below).
***Data (of the variables measured) may contain too much 'noise' for the model to handle - and the model 'force fits' the data. So is the output 'model created'?

The above is a VERY crude representation of the issues covered by Training Peaks The usual issue of sample size and different disciplines (running, weight lifting, cycling, etc.) and other issues that plague training research applies.

There are many issues with programs measuring training and output. The simple Functional Threshold Power test that is used for example - athletes rarely use the one hour maximal effort required as it can take days (or weeks in my case!!) to recover from. They should also re-do this regularly (as Trainer Road advises). But we use the 20 minute test and take a 95% of that as functional threshold power. But such "treatments of the data convey a false sense of precision and may lead to inappropriate training decisions" (Dr. Philip Friere Skiba, PhysFarm Training Systems LLC, http://perfprostudio.com/WebHelp/Studio/scr/Bike...)

So in my opinion (worth every cent you paid for it!!) the Strava data is a guide/indicator and useful from that perspective.

For those interested some articles (NOTE depending on which browser used and how set up it may want to download the PDF docs.)
http://jap.physiology.org/content/92/2/572 - Effects of training frequency on the dynamics of performance response to a single training bout.

http://www.fmh.utl.pt/agon/cpfmh/docs/documentos... - The Quantification of Training Load, the Training Response and the Effect on Performance

http://www.fisioex.ufpr.br/resources/BE711/BE711... The Analysis and Utilization of Cycling Training Data (also makes some comments on power metres and issues)

http://perfprostudio.com/WebHelp/Studio/scr/Bike...

Note this is a book "Routledge Handbook of Ergonomics in Sport and Exercise" and the link should take you to the appropriate page - but you can search the book from the left panel.
http://books.google.com.au/books?hl=en&lr=&id=3U... (see the summary on page 261 for indication of what's coming Smiling.

HTH! :/

Long answer is it takes a couple of months riding to build a trend in the graph. For me, the fitness number needs to be high, the freshness low and the form number positive. Power meter users have been able to use the graph for over 12 months now and I've been able to tailor my training to increase the fitness number without getting too fatigued, or over trained. I know it works because my PR's and KOMs match up to the peaks in the graph.

Sounds like it's a good thing then. I'll keep an eye on it and see if I can raise the mumbers. ☺

i love oldernslower,s short answers!! SmilingSmilingSmiling

i love oldernslower,s short answers!! SmilingSmilingSmiling

this is a great thread but there's a little too much science for an amateur like me who rides for mainly for pleasure and where fitness benefits are largely secondary. Is there an idiots guide to getting started and applying some of this theory to build a an effective training regime to get maximum benefit from time in the saddle .. where to start?

Ok I'll try to keep it simple
Fitness number is a representation of your fitness/training level.
Freshness number is a representation of your fatigue level.
From is the difference between the two.
Your aim. Train your fitness number up without letting your freshness number get too high. This would give you a negative from number.
Or simply train than rest. It's a graphical way it plot and target events. Your freshness will drop quicker than your fitness.
Train hard before an event, than rest up with some quiet rides in the days just before to freshen up. Your form number will be in the positive and you should be ready to fire.
Ahh, that's as simple as I can put in.

Using Strava's terminology

Fitness = looks at your rides over a longer period (eg last 60 days)
Fatigue = looks at your rides over a shorter period (eg last 14 days)
Form = Fitness - Fatigue

The 60 and 14 days that I've quoted are just examples. Who knows how many days strava uses.

42 days for long term stress average.
7 for short term stress.

Training peaks, GC etc work off that as standard.

As you probably figured out, even the idiots guide can get complicated pretty quickly.

It's harder to explain how it works than just doing it. The best way to figure it out is by trial and error over a couple of months. You'll soon learn what the graph looks like when your going well and what it looks like when your not so well.

i presume you need to wear your HR strap on every ride to keep it relevant? i only wear mine on MTB rides - which have unfortunately been limited the last few months - so my graph appears rather wacky!

I wear mine every ride except for grocery runs to the neighbourhood shops. It's just become ingrained habit.

I also have PMs on the two bikes I ride most often. Now there's a tool for quantifying suffering, especially when you're trying to maintain a number at the back end of the interval.

I think its great strava have added it to look at HR or Power. I've used HR training load software for years as part or sportstracks (which I no longer use). If you are going to use it as a guide for preparing for races, remember, not everyone is the same so you might find you race better when the form rating is slightly below zero where others might be better when above zero.

There's also many other factors outside of the fitness/freshness graph. Quality of sleep, nutrition on and off the bike, life's stresses etc.

If your heart rate monitor is like mine and throwing random spikes over 200 then this is going to skew your data. My form currently looks bad but that's not the reality (I think). This might actually force me to change the battery Eye-wink. Overall it will be interesting to hopefully see some trends over time.

I was using the similar chart on training peaks for a couple of years. Generally it is pretty good but it seems to fall down when you start making intervals the basis of the training. Somehow it never seemed to get the training load right when you do the short sharp stuff.

Basically it wasn't able to predict when I was feeling like crap, and the only thing that seems to count as training load is hours on the bike. This was using a power meter - using HR will have even more limitations.

As far as being science - it is but it is a pretty mushy representation of training adaptation. Scientist would call this a "curve fit" this is where some arbitrary parameters are used to adjust the output until you end up with something that is pretty real.

The problem I found is that it doesn't measure training stress for our different energy systems. If you shift to VO2 max intervals then you are stressing that system, but the fitness/freshness doesn't reflect this.

Nonetheless it is well worth a look, and within its limitations it gives a pretty good idea of your training load.

What's the chart called on TP?

The performance management chart.

Thanks Alan

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Best Mountain Bike