Performance Management Chart Experiment

Performance Management Chart Experiment

Introduction

In this excellent presentation on Targetted TSS (Training Stress Score) for Ironman Training, Gordo Byrn and Alan Couzens of EnduranceCorner.com presented this interesting graph:

My initial reaction was “that shouldn’t work”. A 100 TSS bike workout for you is not the same as for me, unless we have the same FTP (Functional Threshold Power). For why see here. If our TSS’ are not equivalent then our Critical Training Load (CTL) can’t mean the same thing either. Consequently if you and I both reach a CTL peak of 120 in our IM builds, it is unlikely that our performances will be the same. That isn’t even taking into account variations due to weather, course difficulty and other performance affecting factors.

However, I was intrigued by the idea of planning my training to target a particular peak CTL.  This peak CTL was combined swim, bike and run as I understood the presentation to use that approach.

For interest I also went back to my 2013/14 training cycle and estimated my peak CTL prior to my Almere iron distance race last year.

Method

Byrn and Couzens use the Training Peaks (TP) software to manage their athletes. TP seems to be more targeted at coach/athlete combination and provides a lot of functionality that I don’t really need and is consequently seems quite expensive as a single self coached middle of the packer. Unfortunately Strava doesn’t allow estimation of TSS. I don’t have a power meter on my race bike and I wanted to combine swim/bike/run TSS. So Strava’s Performance Management Chart (PMC) was a non-starter.

I created an excel sheet to keep track of Fitness (CTL), Fatigue (ATL) and Form (TSB). If anyone is interested I can provide more information on how that sheet works.

For each workout I manually entered the TSS into my spreadsheet. Training Peaks free edition provides Swim, Bike and Run workout TSS. When I didn’t have my power meter, I estimated using the approach from my previous post.

I plan my seasons using the usual periodized approach described by Joe Friel amongst others. Detailed weekly planning I do about 2-3 weeks ahead. This season I added a new step of estimating the likely training load of each workout for the next week so that I could see the ramp rate for my Fitness (CTL) score. I aimed for a CTL ramp rate (increase in Fitness per week) of less than 3-5 per week. Taken together with a recovery every 4^th week, monthly ramp rate was around 10. These metrics were based on recommendations from Byrn and Couzen’s presentation.

For interest I looked back at Challenge Almere from 2014. My estimated peak Fitness score was 97 two weeks out from the race.

Conclusion

My peak CTL was 123 on 19 August. My A race was Challenge Weymouth iron distance on 13 September. That seemed about right with a 3 week peak and taper period. Due to my taper, the day before my race CTL had dropped to about 100. In hindsight, I think it would have been better to hold that drop to less than 10-15%.

On the day I finished in 13:26. It was a tough day with some strong winds and rain. I also had a mechanical problem and some illness. Even with that, there is absolutely no chance I could have gone sub 12 never mind sub 11 hours as suggested by the chart. So, as suspected (in my case) I couldn’t see a direct linkage between peak CTL and race time.

However I could see a linkage between increase in peak CTL and performance.

At Almere I finished in 12:46. However that was a much faster course with much less elevation, a lake rather than a sea swim and no rain. There were also no mechanical or illness problems. Taking out 10 minutes for these last two exceptions and looking at my time as a % of the age group winner there was an improvement:

Almere            126%

Weymouth       122%

So a 23 point increase in peak CTL got me a 4% improvement in comparison with the age group winner. I am however left with a number of questions regarding the validity and usefulness of all of this:

• ·      A sensible training plan would be based on something like a 1:1:2 swim:run:bike training time ratio and 3 sessions in each discipline per week split Tempo, Interval and Long. With that as a basis, weekly training time may be just as useful as CTL at my level.
• ·      Tracking CTL within a season doesn’t take account of the effect of multi year training. This was my 3^rd season of endurance training. Perhaps the gains could have been the same with less volume?

Last Word

Statistically, this experiment lacks validity. The sample size - one - is far too small.
For 2015/16, I will not be repeating the same CTL planning approach. I accept that for elite athletes it might be useful and it was certainly an interesting experiment for me. However, hours per week are probably good enough at my level for now.

Estimating Training Stress Score

Introduction

In order to maintain a meaningful Performance Management Chart it is necessary to measure Training Stress Score. My Powertap is on my training wheels. My race (TT) bike does not have the ability to properly measure power. In order to keep the PMC up to date I therefore need to estimate TSS.

Methods of TSS Estimation

Training Stress Score can be defined as:

(Intensity Factor)^2 x duration (hours)

I’ve tried several estimation methods:

·      Fixed multiple

·      Perceived Exertion

·      Energy Expenditure

·      Estimated Intensity

·      Heart Rate

Fixed Multiple

Based on the average TSS/hr achieved on a number of rides. Estimate is simply (Hours Ridden) x (average TSS/hr).

Perceived Exertion

Joe Friel on Training Peaks quotes the following table:

Energy Expenditure

Based on average TSS/Kj of work. Estimate is simply

(Total work done on ride)/(Average TSS/Kj of work)

Estimated Intensity

Based on personal assessment of intensity %. Estimate is:

(Intensity Factor)^2 x duration (hours)

Heart Rate

Strava and others propose using HR to estimate TSS. In principle this would be better. However I’ve not been comfortable with the results as the HR estimates appear to be much lower than power based measures for very similar rides.

Test Data

In order to compare these methods I took a selection of 6 rides of varying distance, intensity and terrain, where I used my Powertap:

Conclusion

Estimating TSS is clearly inaccurate. Using a Fixed Estimate is quite weak – especially when intensity varies. RPE estimates both too high and too low. Energy seems to be the most reliable of the methods. For me, dividing Energy by 8 gives an approximation of TSS. The estimate tends to be a little high on very low intensity rides.

Power Zones - % FTP or Peak Power?

Looking at this screenshot from Golden Cheetah got me thinking about power zones. It's in French, but it demonstrates the point better than the English screenshots I could find:

Golden Cheetah, like just about everything/one else, seems to take FTP and assign a % range for each of the zones. I noticed though, that the zones were quite closely aligned to key points in the Peak Power Curve: z3 between 1 and 2 hour peak power, z6 between 1 and 3 minutes.
Since riders are all different, it seems inaccurate to assign their training zones to standard % ranges, rather than to customise them to their specific abilities using the data available.
For example my own, not very wonderful, power curve looks like this:

Adopting peak power based, rather than % of FTP based zones would result in some differences:

Since my target race distance is 180 km, it is not surprising that zones 1 and 2 finish higher than a typical profile. It is perhaps surprising that zone 3 finishes lower. That is probably a personal weakness where I need to devote some time. Zones 4-7 are a bit higher than typical. Arguably, today I am pushing too hard when I'm aiming for z3, but not hard enough in all the other zones.

Another approach would be to assign different % ranges depending on generic rider types. Dr Coggan proposed several generic rider types here:

home.trainingpeaks.com/blog/article/power-profiling
He compares actual power outputs/kg to a standards table and looks at the resulting profile shape, categorising riders as:

• All Rounder
• Sprinter
• Time Triallist/Endurance
• Pursuiter

Perhaps the % ranges should be adjusted for each rider type? Endurance riders might see zone 6 starting a little lower - 110% for example. Zone 7 would similarly be lower. Sprinters might see Zone 3 start a little lower 75 or 80% perhaps.

The Effects of Age on Functional Threshold Power

Dr Coggan wrote a very good piece on how Functional Threshold Power declines with age here:
Successful Aging and Functional Threshold Power
The conclusion being that FTP w/kg declines on average at 0.04 w/kg/yr after 40 years old. After the FTP test at our club this week (see previous post) we were discussing how to compare the results of the younger guys with us old war horses.

It seemed to me that I might answer the question from another direction using the Veteran Time Trial Association Standards Tables and the well known equation for power (Martin et al 1998):

(g is Earth's gravity, m mass of rider+bike, Vg speed over ground, K1 is a constant representing rolling resistance, s slope, K2 a constant representing drag, Va is speed through the air)
I made some reasonable assumptions - flat course, still conditions, rider of 75kg and 8 kg bike, etc. and arrived back at exactly the same result for both men and women! I varied the rider and bike mass and the results did not vary to a significant degree. Whilst my results validated the average reduction per year, it was not linear. Up to age 65 average reduction per year was less than 0.04. After 65 it increased quite rapidly as shown in the graph below:

So at age 55, my FTP/kg of 2.85 is comparable to 3.45w/kg for a <40 year old. To arrive at your own age adjusted w/kg subtract 40 from your age and multiply by 0.04 then add to your FTP/kg. If you want to know how you compare to others you can then look yourself up in the table prepared again by Dr Coggan:

Maximal power output (in W/kg)
	Men				Women
	5 s	1 min	5 min	FT	5 s	1 min	5 min	FT
	24.04	11.50	7.60	6.40	19.42	9.29	6.61	5.69
	23.77	11.39	7.50	6.31	19.20	9.20	6.52	5.61
	23.50	11.27	7.39	6.22	18.99	9.11	6.42	5.53
	23.22	11.16	7.29	6.13	18.77	9.02	6.33	5.44
World class	22.95	11.04	7.19	6.04	18.56	8.93	6.24	5.36
(e.g., international pro)	22.68	10.93	7.08	5.96	18.34	8.84	6.15	5.28
	22.41	10.81	6.98	5.87	18.13	8.75	6.05	5.20
	22.14	10.70	6.88	5.78	17.91	8.66	5.96	5.12
	21.86	10.58	6.77	5.69	17.70	8.56	5.87	5.03
	21.59	10.47	6.67	5.60	17.48	8.47	5.78	4.95
Exceptional	21.32	10.35	6.57	5.51	17.26	8.38	5.68	4.87
(e.g., domestic pro)	21.05	10.24	6.46	5.42	17.05	8.29	5.59	4.79
	20.78	10.12	6.36	5.33	16.83	8.20	5.50	4.70
	20.51	10.01	6.26	5.24	16.62	8.11	5.41	4.62
	20.23	9.89	6.15	5.15	16.40	8.02	5.31	4.54
	19.96	9.78	6.05	5.07	16.19	7.93	5.22	4.46
Excellent	19.69	9.66	5.95	4.98	15.97	7.84	5.13	4.38
(e.g., cat. 1)	19.42	9.55	5.84	4.89	15.76	7.75	5.04	4.29
	19.15	9.43	5.74	4.80	15.54	7.66	4.94	4.21
	18.87	9.32	5.64	4.71	15.32	7.57	4.85	4.13
	18.60	9.20	5.53	4.62	15.11	7.48	4.76	4.05
	18.33	9.09	5.43	4.53	14.89	7.39	4.67	3.97
Very good	18.06	8.97	5.33	4.44	14.68	7.30	4.57	3.88
(e.g., cat. 2)	17.79	8.86	5.22	4.35	14.46	7.21	4.48	3.80
	17.51	8.74	5.12	4.27	14.25	7.11	4.39	3.72
	17.24	8.63	5.01	4.18	14.03	7.02	4.30	3.64
	16.97	8.51	4.91	4.09	13.82	6.93	4.20	3.55
	16.70	8.40	4.81	4.00	13.60	6.84	4.11	3.47
	16.43	8.28	4.70	3.91	13.39	6.75	4.02	3.39
Good	16.15	8.17	4.60	3.82	13.17	6.66	3.93	3.31
(e.g., cat. 3)	15.88	8.05	4.50	3.73	12.95	6.57	3.83	3.23
	15.61	7.94	4.39	3.64	12.74	6.48	3.74	3.14
	15.34	7.82	4.29	3.55	12.52	6.39	3.65	3.06
	15.07	7.71	4.19	3.47	12.31	6.30	3.56	2.98
	14.79	7.59	4.08	3.38	12.09	6.21	3.46	2.90
Moderate	14.52	7.48	3.98	3.29	11.88	6.12	3.37	2.82
(e.g., cat. 4)	14.25	7.36	3.88	3.20	11.66	6.03	3.28	2.73
	13.98	7.25	3.77	3.11	11.45	5.94	3.19	2.65
	13.71	7.13	3.67	3.02	11.23	5.85	3.09	2.57
	13.44	7.02	3.57	2.93	11.01	5.76	3.00	2.49
	13.16	6.90	3.46	2.84	10.80	5.66	2.91	2.40
Fair	12.89	6.79	3.36	2.75	10.58	5.57	2.82	2.32
(e.g., cat. 5)	12.62	6.67	3.26	2.66	10.37	5.48	2.72	2.24
	12.35	6.56	3.15	2.58	10.15	5.39	2.63	2.16
	12.08	6.44	3.05	2.49	9.94	5.30	2.54	2.08
	11.80	6.33	2.95	2.40	9.72	5.21	2.45	1.99
	11.53	6.21	2.84	2.31	9.51	5.12	2.35	1.91
Untrained	11.26	6.10	2.74	2.22	9.29	5.03	2.26	1.83
(e.g., non-racer)	10.99	5.99	2.64	2.13	9.07	4.94	2.17	1.75
	10.72	5.87	2.53	2.04	8.86	4.85	2.07	1.67
	10.44	5.76	2.43	1.95	8.64	4.76	1.98	1.58
	10.17	5.64	2.33	1.86	8.43	4.67	1.89	1.50

As you will see my 2.85 put me between Fair and Moderate. Age adjusted increases this to the lower end of Good. Given that I am finishing in the middle of my age group on ironman distance bike legs, that seems about right.