Friday, September 5, 2014

CTL as predictor of performance

A quick n=2 analysis of CTL vs kilojoules and other metrics for predicting performance in cycling.

Subject 1:

Elite female racer, (my wife) 12 months of data during the 2013-2014 bike racing season. All data collected with a powertap. Any workouts lacking power data have estimated TSS  entered by hand and kilojoules estimated from that TSS. PMC set with a 42 day time constant. For each month in the data set we plot mid month CTL value against the peak minute normalized power achieved in that month (within ~15 days of the CTL value).  45 minute normalized power was chosen as a measure of performance as it is known that the athlete in question would have all out efforts in that duration every month and it is a good proxy for general aerobic power (aka CP, or FTP)  Other performance metrics are shown in a chart of correlation coefficients. Including 1 minute power, 5 minute power, and powerfactor (which is just a weighted average of the other 3).  These same performance metrics were then compared to an exponentially weighted moving average of kilo-joules, using the same 42 day time constant as CTL, I call this ewaKJ.  This amounts to using the same approach as CTL but replacing TSS with kilo-joules

click to enlarge

click to enlarge

For this athlete, the two metrics (CTL and ewaKJ) do about as well as one another at predicting aerobic performance, and 1 and 5 minute power as well.

Subject 2:

Me, a cat 3 male bike racer, during the early 2013 season. Again all data collected by a powertap, same procedure as subject 1:

click to enlarge

This time the results are very different. My training period had a large shift in its makeup, from primarily long steady training rides to lots of racing and crits. ewaKJ was trending down during the season as CTL was trending up, and so was performance.  Being able to handle differences in training variability is exactly what is supposed to make CTL a better indicator of overall training load than simpler metrics. So this is a good sign.

I would love to do more sophisticated analysis like this, with more subjects, but data quality is paramount, as is a deep understand of the athletes training and racing history, so that performance metrics relevant to them can be devised.

Tuesday, June 10, 2014

Championship TT Build

by Jack Mott

Being former or current multi-sport athletes in many cases, the ATC Racing women pride themselves on their time trialing prowess and put up impressive results at this year's state TT. Katie Kantzes nailed a third place in the Eddy Merckx category, Marla Briley got second in category 3, and Kat Hunter took first place in Cat 1/2. Read Kat's race report on

On Sunday, the team of Missy Ruthven (ATC owner), Maggi Finley, Marla Briley, and Kat Hunter backed up those impressive results with a 56:37, fastest women's time, in the team time trial.

To give you an idea of the attention to detail that goes into a winning time trial, we will break down all the gear, aero data, and power data of Kat Hunter's winning ride.

Bike Build Specs

Power Meter Data
  • Goal Power -  250 watts
  • Actual Average Power - 247 watts
  • Variability Index - 1.0
  • First Half Power (Headwind) - 254 watts
  • Second Half Power (Tailwind) - 241watts
  • Avg Temp - 82 deg G
  • Avg Cadence - 92
  • Avg Speed - 26.6 mph
  • Distance - 40k
  • Time - 56:23

Power File

WKO+ Power File from the TT

Aero Data

Approximate yaw angle distribution from

Using BestBikeSplit, an excellent online aerodynamic and pacing tool, we can approximate that Kat had a CdA of approximately .20 at 0 yaw, and .19 at yaw. We can also see that about 70% of the race was between 0 and 5 degrees of yaw, with most of the rest of the race between 10 and 15 degrees of yaw.  Knowing what the angle of attack of the wind is during a race can help you make intelligent equipment choices. For instance, we can see from this data that a deeper front wheel would have been faster, as around 30% of the race had yaw angles where a Jet 9 is a bit faster than a Jet 6.  Over 70% of the race was at very low yaw angles, which means choosing a narrower tire was definitely a good call.

Close-up Bike Shots 

The venerable P2 Classic remains competitive

View Speed Skewers, Cervelo FK26 fork, and HED Jet 6 with Continental Attacks

HED Jet 9 with a disc cover taped on,  nice clean chain

TriRig brake, stem, and HED aerobars make for a clean front end

Sunday, June 1, 2014

The Mean Maximal Power Chart
All about the power - duration curve

by Jack Mott

In a previous post, Power Meters Explained, we went over some of the benefits and uses of a power meter, also providing a quick review of the popular models available. Now we will dive deeper and discuss the Mean Maximal Power (MMP) chart and its many uses.

The Power Duration Curve

The power duration curve represents the maximum power you can produce on the bike over each duration of time. This horizontal axis is logarithmic time, which allows you to see relevant changes in  power more clearly. Most cyclists will have a power duration curve shaped somewhat like the example above. For periods of about 1 to 10 seconds, you can produce a huge amount of power using primarily the phosphogen energy system, shown in red. Then there will be a steep drop-off in your sustainable power from there to about five minutes, when power production is dominated by your anaerobic energy system. Anaerobic capacity is a fixed amount of energy, lasting only a few minutes. Hence the steep decline in sustainable power in this region. As your sustainable power levels off, power production is dominated by the aerobic system, which is almost indefinitely sustainable, with a slow, gradual drop-off as the duration goes on for hours and hours. Knowing how these three energy systems interact, you can predict how much power you should produce at any duration, as long as you have enough data to have an idea of how your own power duration curve is shaped.

The MMP Chart

The MMP chart looks at the most power you have ever averaged for each given amount of time. It scans through all the training files you specify, finding your best ever one-second power, two-second power, and so on all the way out to your longest ride. This functionality is available in many power analysis programs, including Training Peaks, WKO, and Golden Cheetah. If you have done all-out efforts over many different time periods, your MMP chart will look very much like the example above. If your data is sparse you won't get a clear picture of how your power duration curve is shaped.  Below are two examples. One is an MMP chart with data from many rides; this data includes hard sprints, hard anaerobic efforts, and hard long-term efforts. Note the resemblance to the theoretical power duration curve. On the right is an MMP chart with sparse data. The rider has done no all-out sprints, so you don't see the sharp decline in the anaerobic zone. If you want to have a good idea of what your power duration curve is, you should periodically do all-out efforts in each zone that you are interested in.

Identifying Strengths and Weaknesses

Once you have a good set of data in the MMP chart you can use it to identify strengths and weaknesses. Sprinters will naturally have very high power in the 1- to 30-second range. Good lead-out men or pursuit and kilo riders will have a huge anaerobic capacity, while long-distance TT specialists and mountain climbers will have big power in the aerobic range. This is the same concept as power profiling but with more refined detail. In the chart below, you can see the difference in the shape of the power duration curve between someone who might be a good cat 2/3 sprinter, and someone who might be a good cat 2/3 time trialist or triathlete. A bike racer might use this data to decide he needs to work on his sprint, or he may decide his sprint is hopeless and focus his tactics on breakaways, or switch to triathlon!

Monitoring Progress

By using date filters you can overlay different sets of data onto one chart. If you load up the day's ride on top of all your previous rides of that season, you can see if you have broken any personal records at a glance. In this example below from WKO+, the dark yellow line represents an athlete's entire season of data, and the dotted line represents a single ride. With a quick look, the athlete can see that he set a new all-time best power in the 10-minute range, highlighted in red.

Click to Zoom
Here is another example, using the MMP functionality in Golden Cheetah. The selected ride's MMP is shown via the black line, while the colored line represents the entire season of data. Again, at a glance the athlete can tell that a new sprint power record was set in the 30-second range. The shape of the day's MMP chart can quickly tell you how hard or easy a ride was, and what the nature of the ride was like.

Click to Zoom
After a group ride or race, you can load up that day's effort and compare it to the rest of your season and see if you have set any new records. New records in the longer durations would suggest your FTP might have gone up.

Guessing at Your FTP (or Any Other Power/Duration )

Once you have a good amount of power data, you will be familiar with your own personal power duration curve and can often guess what your FTP (or  approximately 60-minute power) is by looking at your MMP chart.  For example, take the athlete's MMP data below. This athlete did an all-out time trial of about 17 minutes, but has no recent data for hard efforts longer than that. The sudden drop-off in power is circled in red. If she wanted to guess at her FTP, she could just eyeball the general curve out to the one-hour mark, or use the power duration model built into Golden Cheetah to predict it (dotted red line). Don't trust these kinds of models blindly, however, as they depend on sufficient data and are not perfect!

Another tool you can use to guess at your FTP, or sustainable power for any duration, is to use normalized power (NP).  Maybe you haven't done a steady, all-out one-hour effort yet this season, but you have done plenty of hard group rides or bike races of around an hour in duration. The stochastic nature of these efforts will not result in your best possible average power, but you can use NP to estimate what an equivalently hard steady state effort's power would have been. Most power analysis tools can display the MMP chart using average power or NP. Again, don't trust NP blindly, as it can sometimes overestimate your sustainable average power. By glancing at your MMP chart in both Average and NP form you can usually get a good idea of whether your aerobic power is on the rise. With experience, you will learn if NP tends to overestimate for you.

Extrapolating your power duration curve from your MMP charts allows you to set power goals for intervals or events of any duration and estimate your current FTP, even if you haven't formally tested it.

Friday, January 3, 2014

Kona Bike Stats

A common topic of debate is how much faster modern triathletes are today thanks to fancier bike equipment. Some claim that legends like Mark Allen and Dave Scott rode their round tube frames just as fast as today's pros ride their high-tech carbon, aero equipment. Comparing bike performance is a tricky business, as a host of factors make bike times very "noisy." The winds at Kona vary greatly, which can affect bike times by as much as 15 minutes or more. Tactics also affect times, as some years the contenders will all be together on the bike course with nobody pushing the pace. To try to answer the question and make sense of it all we have put together some interactive charts.

The slowtwitch kona archive provides a handy source of data on the top 10 finishers each year since the start of Kona. We chose to look at the time period from 1988 to 2013, as this represents a period when the depth of talent was solid, and the course was relatively constant. It also represents a time after the introduction of the aerobar, when professionals were already adopting bike positions similar to modern athletes. Some small course changes have occurred over these years, but the bulk of the bike course has remained the same. First up, we take a look at the average bike splits among the top 10 overall finishers. Hover over a year for more info, pictures, and links when available.
You can see that there is a clear downward trend in bike times. The linear trend shown in light blue suggests that bike times have improved by 13 minutes, or 4.5% over the time period. However, that isn't necessarily all a result of improved bike gear. Records have been dropping in all sports, even those like running, in which equipment plays almost no role. Since running isn't impacted much by technical advancement, it gives us a great point of comparison. We can compare the trends in the Kona run and bike and see if one has been improving at a faster rate than the other.

If the fitness and talent had been the only thing improving Kona performances, we should actually expect to see cycling improve at a slower rate than running, as the nature of aerodynamic resistance limits how much time is saved by a more powerful athlete. But we actually see that cycling is improving about 1% faster than running at Kona over the time period.

Another way to slice the data is to look at the fastest bike split each year. In this case we took the fastest bike split each year among the top 10 finishers. Anyone setting a fast time and then blowing up on the run is thus excluded. Hover over a point below to see who set the fast time that year.
Again we see a clear downward trend in bike times, almost the same trend as in the top 10 analysis in fact. One interesting property of both the top bike splits and the average bike splits is the consistently slow times between 1997 and 2005. Wind, tactics, drugs, and talent are possible explanations that come to mind, but we really don't know. If you have any ideas, drop us a comment and let us know.