PDA

View Full Version : The Thornton Cup: Suit Change Math Formula Competition



jim thornton
March 29th, 2010, 12:00 PM
I would like to extend an open invitation to all my swimming comrades, with or without an interest in mathematical modeling, to enter the Thornton Cup. This competition is named after one of our swimmers who concedes he has benefited tremendously from so-called "cheating body suit" technology.

The competition seeks to find a formula that will prove the most accurate, in both the men's and women's divisions (conceivably, two formulae will be needed), in predicting the change in masters swimming times as we make the sadly disheartening (or, I suppose, to some, happily playing-field-leveling) transition back to the past (though not distant past when Johnny Weismuller and his Olympic peers were allowed, indeed, forced, to swim in body suits).

Normally, a simple and elegant formula along these lines might be considered best:

C (cheating suit time) = {a few coefficients and quadratic equations, possible some linear algebra thrown in for good measure} = P (back to the past suit time)

However, as much as I personally enjoy elegance, simplicity in and of itself will not net you the Thornton Cup. Accuracy, and accuracy only, will.

If you are able to triumph by use of a super computer running simulations 24/7 at the Jet Propulsion Laboratory nonstop between this year's SCY Nationals and the start of next year's event, more power to you!

There will be a trophy presented to the winner of the Thornton Cup, possibly showing said individual bowling or maybe playing softball (such trophies tend to be easier to find at yard sales; I will, however, keep my eyes open for a Mathlete Trophy if one should come onto the market).

But more than such an cherished bragging-rights keepsake, you--sir or madam--will have achieved something much finer indeed: kept the flagging hopes of Thorntons everywhere from collapsing into total despair as his or her once respectable swimming times plummet next year into the stuff of true Intimations of Mortality.

To jump start your own ideas, I have posted my initial ideas at the following:

http://forums.usms.org/blog.php?b=8806

Please be sure to look at the excellent comments so far posted--and use this particular vlog entry as a place to keep the officials at The Thornton Cup competition h.q. apprised of your progress.

Thanks ever so kindly!

Dolphin 2
March 29th, 2010, 03:35 PM
Actually, the change in times only (of swimming in briefs VS a tech suit) is a case of using simple arithmetic.

However, the cheating effect of using a tech suit is a case of simple physics and arithmetic.

Just have the swimmer wear plain briefs and have them float in a "water tunnel" tethered by a line attached to the end of the tank and measure the drag force with a accurately calibrated scale. Then repeat the experiment wearing a tech suit.

Therefore the "Cheat Factor" is just the percent of change in drag force. :agree:

D2

jim thornton
March 29th, 2010, 03:48 PM
Actually, the change in times only (of swimming in briefs VS a tech suit) is a case of using simple arithmetic.

However, the cheating effect of using a tech suit is a case of simple physics and arithmetic.

Just have the swimmer wear plain briefs and have them float in a "water tunnel" tethered by a line attached to the end of the tank and measure the drag force with a accurately calibrated scale. Then repeat the experiment wearing a tech suit.

Therefore the "Cheat Factor" is just the percent of change in drag force. :agree:

D2

A bit too simplistic as drag forces increase as a square of velocity (or something like this). In terms of sheer exhaustion, in other words, swimming fast, very fast, and as fast as you possibly can do not tire you out in a linear way but rather become acceleratingly exhausting as the speed increases.

Thus the effect of cheating suits on ones ability to sustain velocity different distances may or may not be the same.

So far, the very preliminary data based on the NCAA men's top 3 times in freestyle this year compared to last year shows that the longer the distance, the more percentage of impairment loss of body suits have. This may indicate that sustaining drag force reduction over longer period of times may have been more beneficial to distance vs. sprint swimmers.

If the formula were easy, D2, there would be no Thornton Cup necessary to inspire our best and brightest swimming-mathematician minds.

andrewkubik
March 29th, 2010, 04:06 PM
Dolphin2 - I would like to speak with you offline about your expert witness experience in the Calypso case. I'm an attorney representing plaintiffs in an front load mold case. You can email me at ajk@eppsteiner.com or call me at 858-350-1500. Alternatively, you can try to contact me through this forum.

Dolphin 2
March 29th, 2010, 04:48 PM
Except for the purported increase in speed, attempting to theoretically quantify the total effect of using a tech suit is a pretty complicated proposal.

It's about as perplexing as measuring how bad a tornado is by factoring in all sorts of extraneous factors such as the pattern of debris scattering, body count, broken bone injuries, emotional distress, etc.

As I've previously stated, the simplest way to get all the worms back in the can is just to forget there ever was such a thing as a tech suit and that minimizes the chances for cheating in the first place. :2cents:

D2

james lucas
March 29th, 2010, 05:03 PM
So far, the very preliminary data based on the NCAA men's top 3 times in freestyle this year compared to last year shows that the longer the distance, the more percentage of impairment loss of body suits have...
The results from last year's NCAA Division I meet were colored by the "Olympics effect" - that is, the results were skewed by swimmers like Lockte who were nearing their Olympic peaks.

In the meantime, when one compares this year's results to last year's for the Division III meet, one sees a very different result. In the 40-or-so men's and women's events that were contested, swimmers set Division III records this year in 12 events.

In several events, swimmers in 2010 broke Division III records that had stood for more than a decade. Plus, the new records were set in both sprint and distance events (both 200 medley relays, the men's 200 free relay and the women's 100 free - as well as the women's 500 and 1650 free as well as the ultimate distance event: the 200 fly for both women and men).

In other words, based on the results of the Division III meet, the suits might have mattered, but not nearly as much as the swimmers and their training.

Maybe the outcomes will prove to be different for older swimmers, but the outcomes in the NCAAs paint a picture that is, at best, unclear.

Speedo
March 29th, 2010, 05:06 PM
C ~ 1 * P

There's nothing more elegant than that, and it is also correct.

But I'm having a problem with how it can be correct yet not accurate, which is the other criteria.

jim thornton
March 29th, 2010, 05:31 PM
C ~ 1 * P

There's nothing more elegant than that, and it is also correct.

But I'm having a problem with how it can be correct yet not accurate, which is the other criteria.

Hahahahaha.

Well, you do earn points for elegance. Not since

.....................................2
E = MC

have I seen quite such elegance.

But as the vlog indicates, the Cup will be awarded only on the basis of accuracy, even if this requires a garage filled with mathematical scribblings.

James, I may be wrong about this, but I don't think either Lochte or Phelps swam NCAAs last year (well I know Phelps didn't, but I thought Lochte ended his career with the Gators in 2006.)

In any event, the winning formula will be the one that best predicts the change in masters swimming times from this year's SCY Nationals to next year's SCY Nationals. I understand there are all sorts of confounding variables, not the least of which is that most of us try harder the years when we first age up to the next, presumably easier age group; and moreover, we need to compare a given swimmer's change from before and after the suit (since for many age groups--definitely mine--new beasts seem to be coming up the ranks each year, smashing records.)

Perhaps I shall need to more specifically calibrate the criteria for a winning Cup Formula.

For now, just keep in mind the following chestnut: If a predictive formula were easy to produce, I would have no need to scavenge for a Bowling and/or Mathlete Trophy for our eventual winner!

And on this note, I ask that you return now to your high level mathematical cogitations.

Karl_S
March 29th, 2010, 05:41 PM
Is this hypothetical formula supposed to predict; (1) a given swimmer's 2009-2010 performance based on her/his 2008-2009 performance, OR, is it to predict (2) the time of the Nth-place finisher in 2009-2010 based on the time of the Nth-place finisher in 2008-2009? I would suspect that the latter would be a more pure measure of the tech-suit effect, (albeit with complicating factors mentioned in earlier posts).

If (1), are we looking of the season's best performance, or the performance at a particular meet? If the latter, what meet? Most of the big zone and national meets haven't been held yet.

james lucas
March 29th, 2010, 05:51 PM
I may be wrong about this, but I don't think either Lochte or Phelps swam NCAAs last year (well I know Phelps didn't, but I thought Lochte ended his career with the Gators in 2006.)
Right you are.

But, while we're looking at Lochte, look at the top times in the 100 back:

American record, unbroken since 2006: Lochte - 44.6
NCAA Div. I champ in 2010: Godsoe - 45.11

My larger point is that, if the suits made such a difference, someone in a tech suit would have taken Lochte's name off the record board - and in the post-suit year, Godsoe wouldn't have come so close to Lochte's very fast time.

knelson
March 29th, 2010, 06:36 PM
The results from last year's NCAA Division I meet were colored by the "Olympics effect" - that is, the results were skewed by swimmers like Lockte who were nearing their Olympic peaks.

But last year was 2009 and the Olympics were in 2008.


My larger point is that, if the suits made such a difference, someone in a tech suit would have taken Lochte's name off the record board - and in the post-suit year, Godsoe wouldn't have come so close to Lochte's very fast time.

Seems like you're cherry-picking data, though. No, the 100 back record was not broken last year but many records were (11 of 18 events). On the other hand this year not a single record was broken despite several swimmers returning who had set records in '09.

james lucas
March 29th, 2010, 07:55 PM
Seems like you're cherry-picking data, though ... this year not a single record was broken despite several swimmers returning who had set records in '09.
That's fair.

But the results from the Division I meet only tell part of the story, also. If the suits made such a difference, then how could 12 records fall in the 40-or-so men's and women's events that were contested in the NCAA Division III meet in 2010?

Leonard Jansen
March 29th, 2010, 08:28 PM
The competition seeks to find a formula that will prove the most accurate, in both the men's and women's divisions (conceivably, two formulae will be needed), in predicting the change in masters swimming times as we make the sadly disheartening (or, I suppose, to some, happily playing-field-leveling) transition back to the past (though not distant past when Johnny Weismuller and his Olympic peers were allowed, indeed, forced, to swim in body suits).


You need to define "accurate" since I assume that this will be applied across any # of people and races. The other thing is that the statistical measure must be commensurate with the type of data - it is wrong to use things like measures of central tendency (e,g, standard deviation) if the data doesn't fit that model.

An inelegant way, which would be spot-on, would be to construct a cubic spline type function. A cool way, but difficult to express, would be to feed old times vs. new times into a Hopfield-type neural network and see if it converges to a stable solution.

Sorry - I'm being a math wonk here.... I'll go back to my dungeon now.

-LBJ

aquaFeisty
March 29th, 2010, 09:33 PM
That's fair.

But the results from the Division I meet only tell part of the story, also. If the suits made such a difference, then how could 12 records fall in the 40-or-so men's and women's events that were contested in the NCAA Division III meet in 2010?

I would think that there is a lot more noise in Div III data. Think about it: the absolute best, non-pro, collegiate swimmers are in Div I. There might be an exception here and there, but I would guess it is rare. (Just look at how many DIII event winners would even make finals in DI.) But a very good Div I swimmer could transfer to a Div III school and just blow away Div III competition. These transfers happen and can account for wacky things happening in DIII results, which could make suit effects negligible... or at least, muddy up the data with lots of noise.

aquaFeisty
March 29th, 2010, 09:36 PM
You need to define "accurate" since I assume that this will be applied across any # of people and races.

Leonard, I'm taking a wild guess here, but Jim might accept accurate as any model which predicts a time slower than what Jim goes next year in a jammer... thus showing that Jim has improved over and beyond what the suit was contributing...

:D

jim thornton
March 29th, 2010, 11:33 PM
Leonard, I'm taking a wild guess here, but Jim might accept accurate as any model which predicts a time slower than what Jim goes next year in a jammer... thus showing that Jim has improved over and beyond what the suit was contributing...

:D


Carrie, you appear to have stumble on the Daily Double! Now get to work and make me proud of the first ever Female Thornton Cupster.

Karl_S
March 30th, 2010, 09:54 AM
[stuff cut] A cool way, but difficult to express, would be to feed old times vs. new times into a Hopfield-type neural network and see if it converges to a stable solution.

-LBJ

This is an idea that I was thinking of, (although I don't know what a "Hopfield-type" nn is), but it still isn't clear to me what exactly we are to predict based on what. (See my earlier post.)

knelson
March 30th, 2010, 12:30 PM
If the suits made such a difference, then how could 12 records fall in the 40-or-so men's and women's events that were contested in the NCAA Division III meet in 2010?

Great question. I'm not really sure. It does seem like D III swimming has been getting much faster lately, though. The decrease in D I schools with swimming may have something to do with it, but I wouldn't want to attribute it all to that. In any event, you're right, times were very fast at Division III NCAAs.

lefty
March 30th, 2010, 02:57 PM
The decrease in D I schools with swimming may have something to do with it, but I wouldn't want to attribute it all to that. In any event, you're right, times were very fast at Division III NCAAs.

That is exactly what it is. Well no, not 100% but it sure is a contributing factor. Would Zach Turk have passed on UCLA for Kenyon?

jim thornton
March 30th, 2010, 05:17 PM
Bill Sherman reported via a comment on my blog:

Late getting in on this...one of our locals has a son who just finished the NCAA D2 meet, where the average swim of the field was .43 seconds per 50 slower this year than last for the the 50, 100, and 200 freestyles. Pretty broad sample there!

Do DII and DIII schools offer swimming scholarships?

I don't think DI schools do.

Not sure if this affects things or not. But the top 2 divisions saw times slow, and only the bottom division saw records broken. It makes me think that faster swimmers are being squeezed by porgram shrinkage into the bottom division.

james lucas
March 30th, 2010, 07:46 PM
Would Zach Turk have passed on UCLA for Kenyon?
UCLA no longer has men's swimming. In any case, Google shows it was Hawaii, not UCLA, on which Turk passed:

http://www.cleveland.com/sports/college/index.ssf/2009/03/lords_of_the_rings_kenyon_swim.html (http://www.cleveland.com/sports/college/index.ssf/2009/03/lords_of_the_rings_kenyon_swim.html)

Div. III schools give no athletic scholarships, and Div. II schools can give fewer swimming scholarships than Div. I teams. Under NCAA rules, fewer than half of a fully funded Div. I team is swimming on scholarhips, so this is only partly about scholarships.

Swimming seems to be surviving in many of the traditional Div. III swimming conferences. Though UCLA is one of several Pac-10 schools to drop men's swimming, all eight of the schools in Southern California's small-college league still compete in both men's and women's swimming.

Back to the Thornton formula. If such a thing could be done, it probably would look like the following (assuming the swimmer shaves, wears a compression jammer and doesn't get psyched by not having a special suit):

2010 100 free time = [pi * (BMI points over 28) / pi] + [% change in season's workout yardage and/or effort / 50] + 2009 100-yard free time - [2009 100-yard free time * .005 (to adjust for age-related slowness)].

In other words, if you're lean and kept working out at the same pace, and didn't get any older, and you can still keep your head screwed on straight without being constrained like a black torpedo, your times might not be so different. That was the experience in the Div. III meet: in this year's post-suit meet, a senior from Johns Hopkins broke the 200 back record he set last year while wearing a Super Suit.

lefty
March 31st, 2010, 10:42 AM
UCLA no longer has men's swimming. In any case, Google shows it was Hawaii, not UCLA, on which Turk passed:


That was my point. If UCLA had a swim team, perhaps he would have gone there instead (I have no idea. I don't know the guy at all. It was just an example.).

And for the record: D3 is as fast D2.

elise526
March 31st, 2010, 01:58 PM
In the case of women's teams like Emory University (this year's DIII National Champ), such teams are faster than most DII teams and smaller DI teams.

In looking at the times from Nationals, the team's top sprinter who has been there since her freshman year dropped time. Others on the team that have been there since their freshman year dropped time as well.Many swimmers that go to DI are already made, so you aren't going to see the kind of improvement that you would in DIII. Swimmers that may not be good enough to be on the team at top DIV I swimming schools that are also tops in academics (Stanford, Cal, UVA, etc.), often opt to go to smaller, top academic schools like Emory. Also, with lots of attention from the coach and no scholarship pressure, many swimmers who leveled out or felt burned out in high school dramatically improve in DIII.


I would argue that time improvements at DIII Nationals this year can most likely be attributed to improvement and not the transfer effect. I say all of this to say that for those masters swimmers that have not hit their peak potential, they will continue to improve and the suit will be a non-issue.

I'm confident that with new and better training techniques, even those swimmers that appear to have hit their peak may still improve. Perhaps we can look at the suit as we would a new training technique. It made us faster, but now that it has been taken away from us, we must find new training techniques to make up for it.

stillwater
March 31st, 2010, 03:14 PM
The elite at DIII have always been fast. Some very fast swimmers are in this division due to academic issues, money, maturity, and many other intangeables.

There is a greater gap in times from swiftest to slowest, but it is a quality meet, with great swimmers.

gull
March 31st, 2010, 04:15 PM
Your formula needs to include a Coefficient of Mediocrity, since the suits benefit those of us who are mediocre swimmers to a greater degree.

jim thornton
March 31st, 2010, 10:21 PM
Your formula needs to include a Coefficient of Mediocrity, since the suits benefit those of us who are mediocre swimmers to a greater degree.


Excellent point! I think I may have an unusually high C of M.

For those who have been proposing formulae, I should point out that SwimSuitAddict's self-described "adorable" equation has popped into first place. You can see it in the comments section of this vlog entry: http://forums.usms.org/blog.php?b=8806

Peter Cruise
April 1st, 2010, 02:15 PM
Good one, Gull! May I propose a Coefficient of Fiction that compensates for our aging memories of just how fast we used to be?