Do The Numbers Lie?
Today I raced my last road race of the season--the Benton Park race in the Gateway Cup Omnium. The course was a pretty technical cockeyed figure-eight with 10 turns. I waffled between doing the Masters or the Cat 3 race and decided, after hearing about what happened in yesterday's Masters race at the Giro della Montagna (three wrecks) as well as consulting the oracle of the Internet (Facebook friends) that I would do the Cat 3 race.
View Benton Park Classic Option 1 in a larger map
I had previewed the course at least 4-5 laps, so I knew some of what to expect. Obviously a course looks different at speed and without traffic, but the general features remain the same. It's a course where, if a break can get away and gain enough time on the field, it could stay away. It also reminded me a lot of the old U-city Corners course where the race could be lost on the start line by getting bad starting position and having to spend the rest of the race moving up.
I lined up and got bad start position--2nd row from the last. Then, at the start I couldn't get clipped in nor could about five people in front of me. I ended up saying "screw it" about clipping in and pedaling weaving through the morass of riders still trying to find their pedals, and didn't actually clip in until after the first turn. By the time I'd gotten through turn #1, the lead rider was already diving into turn #2. And I thought, "this is going to be a long race."
Cutting to the chase, it was a long race, and I was quite pleased to finish 36th, right in with the main field and about halfway back in the group that actually finished. I sprinted my finish, and don't feel like I left much if anything out on the course.
By this time you might be wondering what the title of this post might mean, especially since the entry is sounding pretty much like a Woe-Is-Me race diary entry. Well, I got home later and, after showering, popped the PowerTap CPU into the download cradle and sucked in today's data, eager to see what the numbers tell me about how hard the race was. Here they are:
That was certainly not what I expected to see. Those numbers are low...really low. For comparison, look at any of the TNW entries and those numbers. Normalized Power at TNW is typically closer to 300, and here it's 266. Now, sometimes one can get unexpected or unusual readings from PowerTap that wasn't zeroed properly--which I didn't do--so I checked the file. There are several locations where I have zero cadence, zero power, zero torque and non-zero speed, so it should have zeroed itself and if it were off the negative side, I should see some negative torque or power values in the file.
Next I looked at the VI (Variability Index) which is a ratio of NP to AP. But, that doesn't look unusually high.
Why on earth did this race feel so hard?
Sometimes it helps to look at a picture, so I look at the power plot from the race. The part with the black background is the period of the race.
If you've never looked at a power file from a ride or a race, you might look at it and say, "huh?" or "so?" What struck me was the shark-tooth nature. The software I use for analysis, WKO+, has a nifty feature called "Fast Find" (FF), which I don't often use for some reason, but looking at the file got me wondering "how much time did I spend above threshold?" FF gives me a way to pick out excursions above some power value within a user-determined window of time.
This second screenshot shows the ride file, the FF dialog with its parameters, and the number of "finds". The power value I used is the most recent approximation of threshold power. So, I was wondering how many times did I wander above threshold for between 15-60 seconds. The answer, as shown on the graph, is three. That doesn't explain why this felt hard, because three isn't that many. Moreover, they're spread out all over the race, too, giving ample time for recovery. No, that's not it. But, I wonder if I shortened the lower end of the window. So, I did that, and shortened it to 10 seconds rather than 15. Here's what I got:
Whoa! Twenty-freaking-nine excursions! That's a whole heck of a lot! Hmmm...maybe that has something to do with why it felt hard? Then, as I scroll through the "finds" on the left side of the screen (which you can't do on the picture) I see that not only were these above my threshold, but they were also most above 500W with several in the neighborhood of 600W. That's a lot of time above threshold, and pretty far above threshold, too.
What I find really interesting, though, is that looking at the numbers all alone suggests that this should be a pretty easy race for a rider with a threshold of 290W. It was enough to make me wonder whether my threshold may have fallen, and consider re-testing. Now, testing probably is a good idea at this point for several reasons since I'm in between seasons, but every other indication is that the value is not overestimated.
My current take-away from today and analyzing the file are two things:
The solution to #1 is obvious: work it into my workout regime. Figuring out #2 first requires heading back to the data to see how much speed I was carrying into and through a turn so that I can see how much speed I "gave up" maybe to braking and such, then I need to work on that aspect of my bike handling. Good thing that cyclocross is about working on bike handling!
If you're interested, you can download the original WKO file (2.x format) and look at it. I'm also interested in any feedback on the analysis or the file in general.
View Benton Park Classic Option 1 in a larger map
I had previewed the course at least 4-5 laps, so I knew some of what to expect. Obviously a course looks different at speed and without traffic, but the general features remain the same. It's a course where, if a break can get away and gain enough time on the field, it could stay away. It also reminded me a lot of the old U-city Corners course where the race could be lost on the start line by getting bad starting position and having to spend the rest of the race moving up.
I lined up and got bad start position--2nd row from the last. Then, at the start I couldn't get clipped in nor could about five people in front of me. I ended up saying "screw it" about clipping in and pedaling weaving through the morass of riders still trying to find their pedals, and didn't actually clip in until after the first turn. By the time I'd gotten through turn #1, the lead rider was already diving into turn #2. And I thought, "this is going to be a long race."
Cutting to the chase, it was a long race, and I was quite pleased to finish 36th, right in with the main field and about halfway back in the group that actually finished. I sprinted my finish, and don't feel like I left much if anything out on the course.
By this time you might be wondering what the title of this post might mean, especially since the entry is sounding pretty much like a Woe-Is-Me race diary entry. Well, I got home later and, after showering, popped the PowerTap CPU into the download cradle and sucked in today's data, eager to see what the numbers tell me about how hard the race was. Here they are:
Cat 3: 40 + 3:
Duration: 43:22
Work: 615 kJ
TSS: 65.2 (intensity factor 0.95)
Norm Power: 266
VI: 1.12
Pw:HR: n/a
Pa:HR: n/a
Distance: 18.972 mi
Min Max Avg
Power: 0 1070 237 watts
Cadence: 35 174 94 rpm
Speed: 5.2 35.9 26.2 mph
Pace 1:40 11:30 2:17 min/mi
Hub Torque: 0 358 61 lb-in
Crank Torque: 0 1105 223 lb-in
That was certainly not what I expected to see. Those numbers are low...really low. For comparison, look at any of the TNW entries and those numbers. Normalized Power at TNW is typically closer to 300, and here it's 266. Now, sometimes one can get unexpected or unusual readings from PowerTap that wasn't zeroed properly--which I didn't do--so I checked the file. There are several locations where I have zero cadence, zero power, zero torque and non-zero speed, so it should have zeroed itself and if it were off the negative side, I should see some negative torque or power values in the file.
Next I looked at the VI (Variability Index) which is a ratio of NP to AP. But, that doesn't look unusually high.
Why on earth did this race feel so hard?
Sometimes it helps to look at a picture, so I look at the power plot from the race. The part with the black background is the period of the race.
If you've never looked at a power file from a ride or a race, you might look at it and say, "huh?" or "so?" What struck me was the shark-tooth nature. The software I use for analysis, WKO+, has a nifty feature called "Fast Find" (FF), which I don't often use for some reason, but looking at the file got me wondering "how much time did I spend above threshold?" FF gives me a way to pick out excursions above some power value within a user-determined window of time.
This second screenshot shows the ride file, the FF dialog with its parameters, and the number of "finds". The power value I used is the most recent approximation of threshold power. So, I was wondering how many times did I wander above threshold for between 15-60 seconds. The answer, as shown on the graph, is three. That doesn't explain why this felt hard, because three isn't that many. Moreover, they're spread out all over the race, too, giving ample time for recovery. No, that's not it. But, I wonder if I shortened the lower end of the window. So, I did that, and shortened it to 10 seconds rather than 15. Here's what I got:
Whoa! Twenty-freaking-nine excursions! That's a whole heck of a lot! Hmmm...maybe that has something to do with why it felt hard? Then, as I scroll through the "finds" on the left side of the screen (which you can't do on the picture) I see that not only were these above my threshold, but they were also most above 500W with several in the neighborhood of 600W. That's a lot of time above threshold, and pretty far above threshold, too.
What I find really interesting, though, is that looking at the numbers all alone suggests that this should be a pretty easy race for a rider with a threshold of 290W. It was enough to make me wonder whether my threshold may have fallen, and consider re-testing. Now, testing probably is a good idea at this point for several reasons since I'm in between seasons, but every other indication is that the value is not overestimated.
My current take-away from today and analyzing the file are two things:
- need more work on super-short neuromuscular intervals (15s and under duration intervals)
- perhaps I am not carrying enough speed through corners and the spikes are due to having to catch up to the wheel in front of me
The solution to #1 is obvious: work it into my workout regime. Figuring out #2 first requires heading back to the data to see how much speed I was carrying into and through a turn so that I can see how much speed I "gave up" maybe to braking and such, then I need to work on that aspect of my bike handling. Good thing that cyclocross is about working on bike handling!
If you're interested, you can download the original WKO file (2.x format) and look at it. I'm also interested in any feedback on the analysis or the file in general.
posted by cleeland at 6:59 PM
8 Comments:
This was a good read Chris.
It's also interesting that your prediction on how hard this race could be based on starting position was correct looking at the file post race.
I'd love to see the download of the guys who stayed in the top 15 the whole race-- I'm sure they didn't experience 29 excursions at those power levels you did!
This race is very unique situation. When I've been in these kinds of races before the only thought I have is "how can dig for a good 30 seconds to get out of this horrible accordian?" You were in a very bad place in the peloton and it's dang near no-win if you can't move up.
Ooops, that comment is listed under "Anna Z." but it's really me, Gary Z.
Two words: quadrant analysis. :-)
Gary/Anna: the only place that a 30 second dig could work for me was along the start/finish stretch. The only other long, straight stretches had "features" that made it difficult to move up without "cooperation" from the rest of the field, e.g., either a tailwind or a long downhill into an off-camber. The start/finish stretch was a short flat followed by a power incline (I won't call it a climb), and I was often able to regain several places on that stretch. Sadly it just wasn't enough.
@andy: that's another tool I rarely use, and in this case I'm pretty ignorant of it as well.
I think the thing that's most interesting is how the numbers don't jive with PE. If I hadn't had a PM on there, I'd swear it was one of the hardest races I'd ever done and would have assigned myself an IF > 1.0 (since the race was significantly less than an hour).
"@andy: that's another tool I rarely use, and in this case I'm pretty ignorant of it as well."
It's never too late to learn!
Well, it probably is for this year...but I'm sure you would benefit in future similar criteriums from what you would learn.
I guess I missed the boat on how this would have been useful looking at QA- I only learned enough to think it allowed analysis of power output v. crank velocity (cadence for dumb folk like me, right?) so you could pinpoint the demands of your event.
I was mostly using it to determine my cadence and associated power output in TT's where I rode the fastest so I could mimic that in training.
Gary, maybe you and I can attend "QA For Dummies" together.
Andy, will the QA webinar enlighten me as to what QA will tell me about this ride analysis?
The only other race that I do like this on the road is the Soulard crit, which has something on the order of 16 corners. It wasn't held this year, but it would be interesting to pull up last year's file on it and do the same analysis.
Post a Comment
<< Home