PDA

View Full Version : Percentage drop offs



Caladonian Boy
February 18th, 2013, 07:27 AM
Hi All,

As a 53 male what would be the expected % drop off for the following distances.

starting with 25m single length as the base

then 50m
then 100m
then 200m

Thank you

rtodd
February 18th, 2013, 07:53 AM
It depends on the level of swimmer. For a very good swimmer the drop off is less. Check out who the top ten swimmers are in the 50-54 age group and then look at their individual swims to get an idea. The nationals qualifying times can give you an idea of this spread as well.

Chris Stevenson
February 18th, 2013, 09:04 AM
You can use the Virginia Ratings Calculator to answer this question:

http://www.vaswim.org/cgi-bin/rcalc.cgi

Pick an event and course (there are no 25s since those aren't official events) then switch from "time" to "rating." A rating of 100 corresponds to a record-caliber swim, so use smaller numbers if you like. Then calculate the time for different ages. The percent increase tends to increase with age.

I have also created an Excel file to do these calculations.

http://www.vaswim.org/RatingCalculator.xlsx

Use the second worksheet (the tab is labeled "Swim Converter"). The converter is not the usual sort since it can convert between ages, which is exactly what you want. The answers from the Excel file won't be exactly the same as the CGI page since I updated the curve coefficients for the spreadsheet but haven't done so for the website.

jim thornton
February 18th, 2013, 08:43 PM
You can use the Virginia Ratings Calculator to answer this question:

http://www.vaswim.org/cgi-bin/rcalc.cgi

Pick an event and course (there are no 25s since those aren't official events) then switch from "time" to "rating." A rating of 100 corresponds to a record-caliber swim, so use smaller numbers if you like. Then calculate the time for different ages. The percent increase tends to increase with age.

I have also created an Excel file to do these calculations.

http://www.vaswim.org/RatingCalculator.xlsx

Use the second worksheet (the tab is labeled "Swim Converter"). The converter is not the usual sort since it can convert between ages, which is exactly what you want. The answers from the Excel file won't be exactly the same as the CGI page since I updated the curve coefficients for the spreadsheet but haven't done so for the website.

Chris, thanks a lot! This is brilliant. Is there anyway to import our meet results into your Rating Calculator so that it would automatically crank out a rating for all the different swims without us having to do manual entry for each individual swim? A way of automating things, in other words?

Could you explain some of the features? What, for instance, are FINA points?

Did you ever come up with a way to adjust for the suit change? Times done in SCY in 2010, for instance, tend to be a lot faster for many of us thanks to the pinnacle of the floatie suits. Since then, times have generally gotten slower even if the swimmers haven't necessarily become worse. This (with apologies to Leslie) seems to have affected men more than women. '

Anyhow, I would love to be able to do an apples to apples comparison. I am at the age where the curve seems to be bending northwards (an omen?) at an ever accelerating pace. I would like to know how much of my swimming performances are the result of aging, and how much are the result of the suit changes, and how much are a synergy of both?

Chris Stevenson
February 19th, 2013, 08:34 AM
Is there anyway to import our meet results into your Rating Calculator so that it would automatically crank out a rating for all the different swims without us having to do manual entry for each individual swim? A way of automating things, in other words?

One of the sheets allows calculation of up to 100 ratings at a time. The reason I limited to that number is that the calculations become unacceptably slow when you have too many rows to calculate, probably due to the over-use of lookup functions (which makes for clearer but less efficient code).

I'll work on this off and on, by next fall I want to have something set up for a dual meet we have here.

Another problem is unrelated to ratings: it is hard to get Meet Manager to export results in an Excel-friendly format. So a better approach may be to forego Excel altogether and write a script that works with a MM export file. That would take me a little longer to do since I am much less adept in traditional scripting languages than in Excel and MATLAB (which is the program I use to do the record fitting).


Could you explain some of the features? What, for instance, are FINA points?

FINA points are used in international elite swimming, it is a 1000-point scale based on this formula:

P = 1000 * (B/T)^3

where B is the "base time" and T is the time that is being rated. By contrast, the VA swim rating calculator uses a 100-point scale based on this formula:

R = 100 * (B/T)

FINA uses the "real" overall world record as its base time while the VA calculator uses an age-dependent curve fit of world or national records in masters age groups, meaning the base time B depends on the age of the swimmer.

So in my spreadsheet the FINA points is basically using the same B term as in the 100-pt scale but in the FINA formula. It doesn't give any new information -- you can calculate one rating if you know the other -- but it may be of interest to people who are used to using FINA points to assess the quality of a performance.


Did you ever come up with a way to adjust for the suit change? Times done in SCY in 2010, for instance, tend to be a lot faster for many of us thanks to the pinnacle of the floatie suits. Since then, times have generally gotten slower even if the swimmers haven't necessarily become worse. This (with apologies to Leslie) seems to have affected men more than women. '

A way to do this might be to designate two years, one that is mostly free of the effects of suits and one that is not. Then you calculate record-curve coefficients for each year; this allows one to calculate a rating using one set of coefficients and then use that rating to calculate a time with the other set. In other words, you have a time conversion.

A complicating factor is that records tend to get faster anyway even with no suit improvements -- by definition they can't get slower, right? -- so ideally you'd want to factor that in. If you choose two years that aren't that far apart (say, 2007 and 2010) then perhaps that factor wouldn't be too confounding and you could get a ballpark estimate.

To get more accuracy one could do some sort of time-series analysis on the progression of records and explicitly model that. Or you could use Top 10 times instead of records (but possibly the effects of noise would increase). Either way that's probably more effort than I want to invest in the suit question because I don't think it would result in anything different than the answer I've come up with for myself: an effect of roughly 1 sec per 100.

(Related to this: I'm at a bit of a loss on how to work in the updating of record-curve coefficients as new records are set. FINA just updates its base times every year and users must accept the fact that a given swim done in one year won't be rated the same in a later year if the world record in that event changes. I guess that's what we must do too, it just bothers me a little.)

Chris Stevenson
February 19th, 2013, 08:49 AM
As a follow up: if I were really interested -- as a researcher -- in the question of suits I wouldn't use record curves or other age progression curves. I would probably use the top 10 times for the year before and the year after the ban and compare the effects of the ban on the times of people who appear in both lists. You could get a lot of relevant factors that way: age, event, course (short-course vs long course) and then you could create a model with whichever parameters are found to be significant.

One problem I see is that you would be missing one possibly important factor: body type or weight, since that information of course isn't available in the TT lists.

Anyway, it isn't hard to get the TT information for a given year: go to

http://www.usms.org/comp/tt/toptenlmsc.php

and start downloading. It is a little tedious since the biggest "chunk" of data one can get is by zone/course/year, so a given year will result in 24 different CSV files to be concatenated (I do wish one could download all the TT times for a given year/course in one CSV file rather than eight...).

sunruh
February 19th, 2013, 09:01 AM
the only problem is guys like David Sims and Keith Switzer that kinda blow a big hole in the slowing down part.
hopefully chris can move up to the 50 age group soon! :D

smontanaro
February 19th, 2013, 10:14 AM
Another problem is unrelated to ratings: it is hard to get Meet Manager to export results in an Excel-friendly format.

The plain text meet results are actually fairly cleanly formatted, e.g.:

http://www.usms.org/comp/meets/meetsearch.php?club=1776&MeetID=20130208SCYCMPY

I don't know how to get the entire meet displayed in one go. You should be able to just toss out anything outside the <PRE> tags, then throw away any tags left (the <strong> tags come to mind), a restrict your attention to the first line of each swim:


1 Kata, Justin M 28 CASC 27.99 27.47 9.00

Skip

Chris Stevenson
February 19th, 2013, 11:34 AM
The plain text meet results are actually fairly cleanly formatted, e.g.:

http://www.usms.org/comp/meets/meetsearch.php?club=1776&MeetID=20130208SCYCMPY

I don't know how to get the entire meet displayed in one go. You should be able to just toss out anything outside the <PRE> tags, then throw away any tags left (the <strong> tags come to mind), a restrict your attention to the first line of each swim:


1 Kata, Justin M 28 CASC 27.99 27.47 9.00

Skip

Yes I'm aware of the HTML results format and Excel imports them just fine, but they are still a problem since the event is not replicated on each line (ie, each line is not a complete record of a swim). The file imports into Excel but then additional work is necessary to get the information into a good "one line per record" format with the parameters needed to calculate the swim ratings.

MM does offer SDIF-compliant (http://www.usms.org/admin/sdifv3f.txt) export formats and I will probably use them, rather than HTML results, as the input into a script or Excel, to get something set up to calculate ratings.

ande
February 19th, 2013, 11:34 AM
Hi All, As a 53 male what would be the expected % drop off for the following distances.
starting with 25m single length as the base then 50m then 100m then 200m Thank you

50/100/200 times / ratios greatly vary from one swimmer to the next. Race them and find out.
Ideally you want to do each race fresh and rested, if you do them on the same day with not enough rest between each, you're not as likely to do your best times on your 2nd & 3rd swims.
Also correctly split your 100 & especially your 200

I've written about 100 vs 200 comparison Ratios in Swim Faster Faster (http://forums.usms.org/showthread.php?4229-Ande-s-Swimming-Tips-Swimming-Faster-Faster&p=276466&viewfull=1#post276466)

For many well conditoned young swimmers, you take their 100 time double it and add 10 seconds
so if you go 1:00 in your 100 fr you're likely to be 2:10 in your 200 Free

Here's some guidelines:

4 - 7 second drop offs: distance swimmers and freaks of nature
ie a friend of mine's daughter went 57.4 in her 100 LCM fr and 1:59.2 in her 200,
she has the remarkable ability to take her 200 free out just 1 second slower than her 100 time,
yet still keep her 3rd & 4th 50's together

7 - 10 second drop offs: middle distance swimmers

10 - 13: most sprinters

13 + drop dead sprinters, poorly conditioned swimmers, & swimmers who made bad split choices in their 200's

Your ratios can change over time based on the type of training you do, but usually people are what they are and changing ratios is difficult and slow.

smontanaro
February 19th, 2013, 11:55 AM
Yes I'm aware of the HTML results format and Excel imports them just fine, but they are still a problem since the event is not replicated on each line (ie, each line is not a complete record of a swim).

I could write you a small Python script which would extract the swim info as I suggested and present you with an Excel or CSV file. Would that help?

Chris Stevenson
February 19th, 2013, 01:50 PM
I could write you a small Python script which would extract the swim info as I suggested and present you with an Excel or CSV file. Would that help?

Sure, send it to me at vawebmaster AT usms DOT org.

Caladonian Boy
February 20th, 2013, 07:32 AM
Hi Chris,

Thank you, that's a great little tool much appreciated.

In fact the times that it generated were pretty much inline with my current swim times, what did surprise me was the % drop off was about 10%-->12% which I thought was more than expected.

I am certainly not fit, but even then as I say the actual times and the projected times do match.

I did notice that inputing my PB's , that I am dropping off more as the distances increase to 200 and 400.

O_DeeDee
February 22nd, 2013, 01:45 PM
You can use the Virginia Ratings Calculator to answer this question:

http://www.vaswim.org/cgi-bin/rcalc.cgi

Pick an event and course (there are no 25s since those aren't official events) then switch from "time" to "rating." A rating of 100 corresponds to a record-caliber swim, so use smaller numbers if you like. Then calculate the time for different ages. The percent increase tends to increase with age.

I have also created an Excel file to do these calculations.

http://www.vaswim.org/RatingCalculator.xlsx

Use the second worksheet (the tab is labeled "Swim Converter"). The converter is not the usual sort since it can convert between ages, which is exactly what you want. The answers from the Excel file won't be exactly the same as the CGI page since I updated the curve coefficients for the spreadsheet but haven't done so for the website.

Hello and thanks for this information. I would like to start using age graded results, but was unable to pull up this second excel file. Maybe because I have a Mac and only have "numbers" not excel.

Chris Stevenson
February 23rd, 2013, 01:55 PM
Hello and thanks for this information. I would like to start using age graded results, but was unable to pull up this second excel file. Maybe because I have a Mac and only have "numbers" not excel.

I have a Mac too but yes you'll need Excel to use the file in the 2nd link. The file was tested on Excel for Mac 2011; I don't know if it would work on earlier version. I was a long-time user of Windows Excel and see no reason it wouldn't work on that platform (other than backward file compatibility) but I didn't test it.

If you don't have Excel you'll need to use the first link.