triathlon calculator question

Champion

7233

Subject: RE: triathlon calculator question

did you forget some things like training volume?

My Race Log

23

Subject: RE: triathlon calculator question

oops sorry i left out that i put in 100 miles per week training volume and race distance is 56 miles. I've only trained for 2 weeks and that's my average mileage so far. The results are still telling me to race at ~14mph and do my training from 7-12 mph.

Changing my training volume increases the estimated race pace for the 70.3 but training paces are always 7-12mph as long as I leave the 20 min Time Trial Speed at 21. Maybe 21 is a really really bad speed -- I'll try to do a real 20 min fast ride to see if I can actually top that. Just changing it from 21 to 23 makes a HUGE difference, it changes race pace estimate to 18mph and training paces to 10-16 mph which seems more reasonable.

Champion

19812

MA

Subject: RE: triathlon calculator question

Read all the notes/instructions on the bottom..it is usually a input error or misunderstanding of the format of how to input the data.

If you still don't get it email Jesse as his email address is on the bottom.

My Race Log

Expert

721

Chenequa WI

Subject: RE: triathlon calculator question

lawrence131 - 2009-09-18 12:16 AM oops sorry i left out that i put in 100 miles per week training volume and race distance is 56 miles. I've only trained for 2 weeks and that's my average mileage so far. The results are still telling me to race at ~14mph and do my training from 7-12 mph.

Changing my training volume increases the estimated race pace for the 70.3 but training paces are always 7-12mph as long as I leave the 20 min Time Trial Speed at 21. Maybe 21 is a really really bad speed -- I'll try to do a real 20 min fast ride to see if I can actually top that. Just changing it from 21 to 23 makes a HUGE difference, it changes race pace estimate to 18mph and training paces to 10-16 mph which seems more reasonable.

Yes. Don't guess the TT number as it is important you get real world time trial results for you. Ride all out on the TT on a flat course and know that it will hurt - a lot. If you feel that you have nothing left at the end, you have done it right. Your TT may surprise you. Note: 21mph isn't inherently bad. Its just a number based on your abilities at the moment. The reason that the race pace estimates are low is that many athletes fry themselves on the bike, only to fail on the run.

My Race Log

Elite

3088

Austin, TX

Gold member

Subject: RE: triathlon calculator question

Hmm...hadn't seen this before. Pretty cool.

I punched in my numbers and the race times for the swim and bike are very close and the power output for the bike is exactly what I did at Longhorn 70.3 last year.

The run, however, is way off for me. With a 21:00 5k time and 40 miles per week of run volume and 140 mpw in bike volume, it's predicting 8:25 per mile on the run. On less than 30 mpw average run and 110 mpw average bike, I ran 8:09 at Longhorn last year and 7:49 at Lonestar in the spring. It's strange to me that it predicts almost 1 minute per mile slower than the open 1/2 mary prediction it provides.

My Race Log

Champion

7233

Subject: RE: triathlon calculator question

years experiance will change the open and tri runs a bit closer.

overall though i find it pretty accurate for me (from tt speed/weekly volume it hit my last two HIM bike legs on power output within 4 watts).

My Race Log

Elite

3088

Austin, TX

Gold member

Subject: RE: triathlon calculator question

I had 5 years in the experience field. Even upping it to 10 years, 150 mpw bike, and 50 mpw run, it still comes up with 52 seconds slower than the open time. I just don't find that realistic and it's not because the open time is overly optimistic. That's actually quite accurate for me.

Like I said, the bike was VERY accurate. I had 245 as my 20-minute power and it spit out 200 watts for a 1/2 IM. I rode 199 NP at Longhorn last year.

My Race Log

Veteran

378

The Cold North

Subject: RE: triathlon calculator question

I've played around with this quite a bit. It's interesting to hear that some of you are pretty close to what it predicts.

It predicts that I could do a sprint tri 14 minutes faster than I've done them all summer. I'd love to shave off 14 minutes, but I just don't know how. I guess I'm just not pushing hard enough when I put all 3 sports together.

Champion

11641

Fairport, NY

Subject: RE: triathlon calculator question

I had an idea on a ride a couple months ago:

I had spent the morning programming and on my ride that afternoon my mind was wandering over various technical details about our database. I realized that we have one of, if not the, largest datastores of detailed triathlon training data in the world. We also have what I am confident is the largest detailed triathlon race results datastore in the world. And the kicker is: it's FOR THE SAME PEOPLE AND WE CAN CORRELATE THE DATA.

I started thinking that we could possibly use the training data in the weeks leading up to the races to find those datapoints that will consistently predict race outcomes. I'm not interested in predicting times because that's too dependent on external environment variables, but I'm sure we could predict Age Group and Overall placement with relatively good accuracy. What's also cool about this is that I realized we can set this to automatically update the dependent variables as more training and race data is entered. We can make the algorithms self-refining. How cool would THAT be?

There are a couple of large projects in the queue before it, but I hope to get started on this in the spring.

Although, this could also be a great dissertation project. We'd be willing to talk to any interested PhD candidates out there...

My Race Log

Elite

3088

Austin, TX

Gold member

Subject: RE: triathlon calculator question

I'm still waiting to be able to put power in my log.... :p

More seriously, that would be a pretty amazing idea. I imagine you're right that you could come up with a far more accurate algorithm than McMillan et. al, have. Although you discount it, I'd be you could be pretty accurate in predicting times for those races that you have a significant previous sampling for. Clearly wind and changes in course layout would have a measurable effect, along with temps, but still, I'd bet it would be pretty close. In some cases, it may be better than AG placement predictions since that depends very much on who shows up to race that day.

My Race Log

Resident Curmudgeon

25290

The Road Back

Gold member

Subject: RE: triathlon calculator question

marmadaddy - 2009-09-18 2:21 PM I had an idea on a ride a couple months ago:

I had spent the morning programming and on my ride that afternoon my mind was wandering over various technical details about our database. I realized that we have one of, if not the, largest datastores of detailed triathlon training data in the world. We also have what I am confident is the largest detailed triathlon race results datastore in the world. And the kicker is: it's FOR THE SAME PEOPLE AND WE CAN CORRELATE THE DATA.

I started thinking that we could possibly use the training data in the weeks leading up to the races to find those datapoints that will consistently predict race outcomes. I'm not interested in predicting times because that's too dependent on external environment variables, but I'm sure we could predict Age Group and Overall placement with relatively good accuracy. What's also cool about this is that I realized we can set this to automatically update the dependent variables as more training and race data is entered. We can make the algorithms self-refining. How cool would THAT be?

There are a couple of large projects in the queue before it, but I hope to get started on this in the spring.

Although, this could also be a great dissertation project. We'd be willing to talk to any interested PhD candidates out there...

Seems to me that your data leaves out the all-important genetic aspect as well. No training in the world is going to get me close to some of the more gifted athletes on this site.

My Race Log

Champion

11641

Fairport, NY

Subject: RE: triathlon calculator question

the bear - 2009-09-18 3:53 PM

marmadaddy - 2009-09-18 2:21 PM I had an idea on a ride a couple months ago:

I had spent the morning programming and on my ride that afternoon my mind was wandering over various technical details about our database. I realized that we have one of, if not the, largest datastores of detailed triathlon training data in the world. We also have what I am confident is the largest detailed triathlon race results datastore in the world. And the kicker is: it's FOR THE SAME PEOPLE AND WE CAN CORRELATE THE DATA.

I started thinking that we could possibly use the training data in the weeks leading up to the races to find those datapoints that will consistently predict race outcomes. I'm not interested in predicting times because that's too dependent on external environment variables, but I'm sure we could predict Age Group and Overall placement with relatively good accuracy. What's also cool about this is that I realized we can set this to automatically update the dependent variables as more training and race data is entered. We can make the algorithms self-refining. How cool would THAT be?

There are a couple of large projects in the queue before it, but I hope to get started on this in the spring.

Although, this could also be a great dissertation project. We'd be willing to talk to any interested PhD candidates out there...

Seems to me that your data leaves out the all-important genetic aspect as well. No training in the world is going to get me close to some of the more gifted athletes on this site.

No, but by comparing your training volumes, paces etc in the weeks leading up to a race as well as other things like years of experience, against a large sample of other people we very possibly determine how you will do compared to the other people in that race.

My Race Log

Resident Curmudgeon

25290

The Road Back

Gold member

Subject: RE: triathlon calculator question

marmadaddy - 2009-09-18 3:56 PM

the bear - 2009-09-18 3:53 PM

marmadaddy - 2009-09-18 2:21 PM I had an idea on a ride a couple months ago:

I had spent the morning programming and on my ride that afternoon my mind was wandering over various technical details about our database. I realized that we have one of, if not the, largest datastores of detailed triathlon training data in the world. We also have what I am confident is the largest detailed triathlon race results datastore in the world. And the kicker is: it's FOR THE SAME PEOPLE AND WE CAN CORRELATE THE DATA.

I started thinking that we could possibly use the training data in the weeks leading up to the races to find those datapoints that will consistently predict race outcomes. I'm not interested in predicting times because that's too dependent on external environment variables, but I'm sure we could predict Age Group and Overall placement with relatively good accuracy. What's also cool about this is that I realized we can set this to automatically update the dependent variables as more training and race data is entered. We can make the algorithms self-refining. How cool would THAT be?

There are a couple of large projects in the queue before it, but I hope to get started on this in the spring.

Although, this could also be a great dissertation project. We'd be willing to talk to any interested PhD candidates out there...

Seems to me that your data leaves out the all-important genetic aspect as well. No training in the world is going to get me close to some of the more gifted athletes on this site.

No, but by comparing your training volumes, paces etc in the weeks leading up to a race as well as other things like years of experience, against a large sample of other people we very possibly determine how you will do compared to the other people in that race.

So you're saying that if my "training volumes, paces etc in the weeks leading up to a race" is the same as, say, bryancd's, then our projected results will be the same?

My Race Log

Elite

3088

Austin, TX

Gold member

Subject: RE: triathlon calculator question

I would imagine he's including prior race results and possibly training paces in the "etc" and "other things" catch-alls.

ETA: Duh....he clearly says training paces. So add races and I'd be surprised if he's not right. And yes, if you're able to hold similar training paces AND volumes as Bryan, I'd say that's a good indicator of your potential performance.

Edited by dgunthert 2009-09-18 5:40 PM

My Race Log

triathlon calculator question

REGISTER TODAY