correlation again: fast runners go out hard

..|_|....................~(@:> *
_______
Blog, or dog? Who knows. But if you see my lost pup, please ping me!
<A
HREF="http://journals.aol.com/virginiaz/DreamingofLeonardo">http://journal
s.aol.com/virginiaz/DreamingofLeonardo</A>

"Donovan Rebbechi" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

I re-ran the test, this time with sub 1:20 half-marathoners who ran the
Staten Island half. I also checked them for a faster half marathon time
at another local race earlier (Colon Cancer half in March).

The bottom line:
34 data points
Correlation comes out at 0.34 or greater.
p<0.05

I ran a few variations -- one was discarding data points that appeared to
have
used the half as a training run (anyone whose split was less than a minute
off
their half time). I also tried throwing away runners with slow marathon
times
(3:03 or slower). Neither of these resulted in a negative correlation --
either
way it was positive at the 0.05 level.

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

I told our coach about your theory that your teammates did well because they
went out "hard".
He had a good laugh, he said your formula is lacking one important
ingredient, common sense.
But since you're hung up on stats, here are some results of my teammates who
ran the local Marathon in October.

HM Result
1:29 2:55
1:30 2:59
1:48 3:28
1:48 3:29

http://www.teamutopia-usa.com/results/oct04.html#MHM

That's all I have right now. I can get the rest today, 10 of 15 ran PRs.
Of the 5 that didn't two were definitely undertrained, one jogged (3:10) as
a long run in preparation for the Philly marathon and the other three are
returning from injury.
None of these people looked like they were candidates for medical assitance
after.

I guess they all should have run harder in the first half.

Parker

Quoted message said:

I told our coach about your theory that your teammates did well because they
went out "hard".
He had a good laugh, he said your formula is lacking one important
ingredient, common sense.

I agree. Foolish.

"Parker Race" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

I told our coach about your theory that your teammates did well
because they went out "hard".
He had a good laugh, he said your formula is lacking one important
ingredient, common sense.

But you forgot to tell him what a wonderful scientist, sans common
sense, that his is. Doesn't that count? 😉

-DF

Parker Race said:

I told our coach about your theory that your teammates did well because they
went out "hard".

As I've already pointed out, that's not my "theory". You're either not paying
attention or you're just plain being dishonest.

As I've already pointed out, there is a significant *ASSOCIATION* between
running a hard first half and running a fast time. I don't have a theory
about why there's such an association, but there is. However, the point is
that the association should go in the other direction if people like Galloway
(with the "rule" that says that 10 seconds lost for each second you go out too
hard)

As I've already pointed out, people who trot out the Galloway "rule" are the
ones lacking in common sense, because when one actually looks at how the "rule"
applies to real data, it's clear that it's wrong. The example I gave was
Lance's marathon last year -- according to Galloway, he could have run even
splits off a 1:32:40 first half (coming off a 1:30 half marathon)

Quoted message said:

He had a good laugh, he said your formula is lacking one important
ingredient, common sense.

Yes, but his opinion is based on your misrepresentation of what I've actually
said.

Quoted message said:

But since you're hung up on stats, here are some results of my teammates who
ran the local Marathon in October.

I can't do the analysis on those because I need the halfway split and a recent
half marathon time. I'll also probably need more than 4 data points to get a
statistically significant effect. Also, this would have the same sampling
problem that the preliminary analysis on my teammates had. But if you can
give me the remaining data, I'd be happy to redo it on that set anyway.

Quoted message said:

HM Result
1:29 2:55
1:30 2:59
1:48 3:28
1:48 3:29

[snip]

Quoted message said:

I guess they all should have run harder in the first half.

The data overwhelmingly suggest that they could have run faster in the first
half without it hurting their performance.

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

"Donovan Rebbechi" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

Parker Race said:

I told our coach about your theory that your teammates did well because
they
went out "hard".

As I've already pointed out, that's not my "theory". You're either not
paying
attention or you're just plain being dishonest.

As I've already pointed out, there is a significant *ASSOCIATION* between
running a hard first half and running a fast time. I don't have a theory
about why there's such an association, but there is. However, the point is
that the association should go in the other direction if people like
Galloway
(with the "rule" that says that 10 seconds lost for each second you go out
too
hard)

As I've already pointed out, people who trot out the Galloway "rule" are
the
ones lacking in common sense, because when one actually looks at how the
"rule"
applies to real data, it's clear that it's wrong. The example I gave was
Lance's marathon last year -- according to Galloway, he could have run
even
splits off a 1:32:40 first half (coming off a 1:30 half marathon)

Quoted message said:

He had a good laugh, he said your formula is lacking one important
ingredient, common sense.

Yes, but his opinion is based on your misrepresentation of what I've
actually
said.

Quoted message said:

But since you're hung up on stats, here are some results of my teammates
who
ran the local Marathon in October.

I can't do the analysis on those because I need the halfway split and a
recent
half marathon time. I'll also probably need more than 4 data points to get
a
statistically significant effect. Also, this would have the same sampling
problem that the preliminary analysis on my teammates had. But if you can
give me the remaining data, I'd be happy to redo it on that set anyway.

Quoted message said:

HM Result
1:29 2:55
1:30 2:59
1:48 3:28
1:48 3:29

[snip]

Quoted message said:

I guess they all should have run harder in the first half.

The data overwhelmingly suggest that they could have run faster in the
first
half without it hurting their performance.

LMAO!
OK, I'll tell them that.
The 2:55 runner ran a faster half at theis race last year and ran 3:01 and
change, her previous best half was 1:29:55 (on a tough course) but she
hasn't run a half in a while and I'm sure she would blow that away by
several minutes now.
I'll get the others half splits.
Got news for you, the 3:28 and 3:29 runners were reeling off miles at almost
the same pace that you were running your last 10k after mile 20. My coach
said those two probably went a little slower than necessary through the
half.

Quoted message said:

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

Parker Race said:

Quoted message said:

The data overwhelmingly suggest that they could have run faster in the first
half without it hurting their performance.

LMAO!
OK, I'll tell them that.

Laugh all you like, but that's what the data say.

Quoted message said:

The 2:55 runner ran a faster half at theis race last year and ran 3:01 and
change,

Which proves absolutely nothing -- the obvious explanation is that she
improved. If we ignore the obvious problem that you have an "order effect",
you also only have one data point.

Quoted message said:

her previous best half was 1:29:55 (on a tough course) but she
hasn't run a half in a while and I'm sure she would blow that away by
several minutes now.

If you think their half-marathon times don't reflect their current ability,
give me a time over some other distance instead, and I'll use it to estimate a
half marathon time.

Quoted message said:

I'll get the others half splits.
Got news for you, the 3:28 and 3:29 runners were reeling off miles at almost
the same pace that you were running your last 10k after mile 20. My coach

But I still beat them, right ? (-;

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

Doug Freese said:

"Parker Race" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

I told our coach about your theory that your teammates did well
because they went out "hard".
He had a good laugh, he said your formula is lacking one important
ingredient, common sense.

But you forgot to tell him what a wonderful scientist, sans common
sense, that his is. Doesn't that count? 😉

As opposed to being numerically illiterate as well as having no common sense ?

I've already shown that your "common sense" Galloway formula is just [censored]. Deal
with it.

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

In article <[email hidden]>,

Parker Race said:

"Donovan Rebbechi" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

I re-ran the test, this time with sub 1:20 half-marathoners who ran the
Staten Island half. I also checked them for a faster half marathon time
at another local race earlier (Colon Cancer half in March).

The bottom line:
34 data points
Correlation comes out at 0.34 or greater.
p<0.05

I ran a few variations -- one was discarding data points that appeared to
have
used the half as a training run (anyone whose split was less than a minute
off
their half time). I also tried throwing away runners with slow marathon
times
(3:03 or slower). Neither of these resulted in a negative correlation --
either
way it was positive at the 0.05 level.

Your abuse of statistics continues to astound me. It just goes to show
that anyone can run a statistics computer program and get meaningless
results.

--Harold Buck

"I used to rock and roll all night,
and party every day.
Then it was every other day. . . ."
-Homer J. Simpson

Parker Race said:

Got news for you, the 3:28 and 3:29 runners were reeling off miles at almost
the same pace that you were running your last 10k after mile 20. My coach
said those two probably went a little slower than necessary through the
half.

And seriously, I'm not sure why you still feel the need to trash my
performance. It's not like I'm the only one who had trouble after mile 20 (or
for that matter, before mile 20).

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

Harold Buck said:

Your abuse of statistics continues to astound me. It just goes to show
that anyone can run a statistics computer program and get meaningless
results.

OK, so how am I abusing statistics ?

[deafening silence anticipated]

I suppose you're going to run and hide behind your little killfile, now that
you've got your cheap shot in.

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

Donovan Rebbechi said:

As I've already pointed out, there is a significant *ASSOCIATION* between
running a hard first half

.... as compared to a recent HM performance in serious but non-elite club
runners...

Quoted message said:

and running a fast time.

.... in the New York marathon. As far as it goes, that seems perfectly
justified.

I think you're being jumped on about this for the wrong reasons. If I
remember the sequence it went something like this :

- shortly after your finish time had appeared on the website, I posted
something about the dearth of negative/even splits in the top 200 and
concluded that conditions must have been hard. This in itself was a
mistake.

- you came on later and appeared to confirm that conditions were not
optimal. Your tone in those first posts was markedly humble BTW.

- there was a lot of discussion about poor pacing (re positive splits)
and a relatively unexplored suggestion about hydration. Suggestions
about "warm" conditions made some headway, despite the fact that temps
were spot on average for New York at that time of year.

- you proposed that going out "hard" wasn't the main factor in those
positives and that maybe even the opposite was true. You did some stats
to prove it.

- the group promptly forgot that we were talking about a particular
race, in particular conditions, because your data showed HM splits which
seemed "too" close to HM PB performance and proved that positive splits
are not necessarily evil

- we tried to hit you in all directions with different takes on the
currently-held and experimentally-verified position that even pacing or
slightly negative splits, are the way to optimal performance IN GENERAL.
This involved shooting at your experimental technique amongst other
things.

- you helped maintain the confusion by arguing with the points made to
disprove things you never said, and by getting involved in
meta-arguments about technique (no offence, but that is a spaciality of
yours)

As in most long-winded discussions, the positions of the warring
factions were not necessarily contradictory, they just _appeared to be_

Am I wrong?

steve common said:

Donovan Rebbechi said:

As I've already pointed out, there is a significant *ASSOCIATION* between
running a hard first half

... as compared to a recent HM performance in serious but non-elite club
runners...

Quoted message said:

and running a fast time.

... in the New York marathon. As far as it goes, that seems perfectly
justified.

Yes, I agree. It's entirely possible that there was something weird going on
in this race.

I agree with the rest of your summary (snipped for brevity).

I also agree with those who say that even pacing (with an all-out finishing
kick that may be much faster than average pace, making for a net negative) is
the optimal racing strategy. Data from world record times overwhelmingly
supports this, and most data from races support this.

The question I'm interested in is how costly a deviation from that strategy is.
As far as I can tell, it seems that one can deviate by a fairly wide margin
without it having much of an effect.

I think my comments have been interpreted along the lines of suggesting that
everyone should go out and aim to run +10 minute splits. Again, I'd like to
emphasise that I don't recommend this. Anyone who has a crystal ball and knows
apriori how fast they can run the race should try to even-split it.

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

Donovan Rebbechi said:

Neither of these resulted in a negative correlation -- either
way it was positive at the 0.05 level.

Better training for distance runners - Martin & Coe 2nd ed 1997.
Preparing to race : Strategies for excellence
7 Strategies for racing Marathons
Break race into familiar pieces (a long run of 20mi + a 10k race)
Practice taking in fluids
Minimizing fatigue through pace management (pp357- )

This section tells of experimental data on real athletes in near-lab
conditions in Japanese marathons. It showed that :

""""a deviation by more than 2% more or less than average overall pace
is metabolically more costly than remaining within the "window" (which
you already mentioned). Athletes with early-race fast splits all too
often pay the price later in the form of a dramatic slowdown, with
proportionally slower late-race splits""""

A little later they suggest :

""""Ideally, you should run the first portion of the race a little
slower than mean pace thereby conserving carb supplies.""""

But they go on to note, given a stastically-significant slowing after
30k in the elites, that this [negative-splitting] seldom occurs because
1) races aren't run in lab conditions and minimal wind + ideal
conditions (cool and cloudy) are not available => relatively harder,
later in the race
2) athletes aren't patient and quickly forget how painful the end of a
marathon is (I believe you are forgetting some of what you learned
already) and promptly make the same MISTAKE
3)Lactate/ventilatory threshold falls during race due to dehydration
reducing blood volume. This causes reduced skeletal muscle perfusion =>
lower VO2max and also increases haemoconcentration => higher blood
acidity.

They go on to show the pacing of an elite athlete in three marathons
with different pacing each time. In the race where he went out very hard
(on schedule for a 2:03:00!!) he crashed and burned, getting slower and
slower til the end, but still finished only 3 seconds off his PB, done
on even splits 3 years earlier.

Their conclusion: how well could he have done if he hadn't paced like an
idiot?

steve common said:

""""a deviation by more than 2% more or less than average overall pace
is metabolically more costly than remaining within the "window" (which
you already mentioned).

On a track, this is just under 2 seconds a lap for someone running at about
6:00 pace. This seems reasonable -- anyone who's run on a track knows that
a sustained increase of 2 seconds a lap or more requires a fair amount of
effort.

Quoted message said:

""""Ideally, you should run the first portion of the race a little
slower than mean pace thereby conserving carb supplies.""""

Well, yes, but the problem is that you don't know what your mean pace is
beforehand.

I agree that it's probably not a good idea to go out at a pace that you know
you can't sustain, but most people are bad at estimating their finishing times.

Quoted message said:

But they go on to note, given a stastically-significant slowing after 30k in
the elites, that this [negative-splitting] seldom occurs because

Another reason I'd offer is that even the elites are often fairly inexperienced
too. The best priors they have to go on are half-marathon or 20k races, and
those aren't bullet-proof predictors. Even though they have a lot of self
knowledge on these shorter distances, it probably just doesn't help much in
the marathon.

Yet another reason is that I think the elites aren't trying to optimise the
"average case", they're trying to maximise their chance of winning or placing.
So if strategy (a) offers a 10% chance of a win and a 90% chance of a crash
and burn, they'll probably still choose it over strategy (b) which offers a 1%
chance of a win and an 80% chance of a reasonably fast finishing time. Such a
choice would be silly for us, but makes some sense for them. I think this
explains why Matt Downin was in the lead pack last year, even though some might
argue that he had no business being there.

Quoted message said:

Their conclusion: how well could he have done if he hadn't paced like an
idiot?

Hahahaha ... I bet they concluded that. In all fairness to them, I think he
would have set a new PR. But I don't agree with the Galloway rule which would
predict something like a 2:03:02 by going out 1 second slower (-;

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

Donovan Rebbechi said:

The question I'm interested in is how costly a deviation from that strategy is.
As far as I can tell, it seems that one can deviate by a fairly wide margin
without it having much of an effect.

See my next post. Martin & Coe have measured the margin as +/- 2% (we'll
have to see the original paper to find out why and what the effects were
when going "out of bounds"😉.

Given the profile of New York, terrain-related variations are well
within that range, except maybe for mile 1 up the V bridge. So the
explanation is elsewhere.

Example, for a 3:10 target, 7'14" mean pace => 7'06 to 7'22" window.
That gives a red-line-all-the-way, at-worst-moment-in-the-race first
10km split of 42'24".

steve common wrote in message ...

Quoted message said:

Donovan Rebbechi said:

Neither of these resulted in a negative correlation -- either
way it was positive at the 0.05 level.

Better training for distance runners - Martin & Coe 2nd ed 1997.
Preparing to race : Strategies for excellence
7 Strategies for racing Marathons
Break race into familiar pieces (a long run of 20mi + a 10k race)
Practice taking in fluids
Minimizing fatigue through pace management (pp357- )

This section tells of experimental data on real athletes in near-lab
conditions in Japanese marathons. It showed that :

""""a deviation by more than 2% more or less than average overall pace
is metabolically more costly than remaining within the "window" (which
you already mentioned). Athletes with early-race fast splits all too
often pay the price later in the form of a dramatic slowdown, with
proportionally slower late-race splits""""

A little later they suggest :

""""Ideally, you should run the first portion of the race a little
slower than mean pace thereby conserving carb supplies.""""

But they go on to note, given a stastically-significant slowing after
30k in the elites, that this [negative-splitting] seldom occurs because
1) races aren't run in lab conditions and minimal wind + ideal
conditions (cool and cloudy) are not available => relatively harder,
later in the race
2) athletes aren't patient and quickly forget how painful the end of a
marathon is (I believe you are forgetting some of what you learned
already) and promptly make the same MISTAKE
3)Lactate/ventilatory threshold falls during race due to dehydration
reducing blood volume. This causes reduced skeletal muscle perfusion =>
lower VO2max and also increases haemoconcentration => higher blood
acidity

I think this last point is very significant. Elites will be done with their
marathon in under 2:30 for the most part, and probably get dehydrated less
than the ~3 hour+ marathoner. Dehydration and a natural shift in energy
supply over time could well be likely reasons why somewhat positive splits
would work at marathon distance. Conditions obviously play a big role.

I think you can only conserve glycogen and ease the enegry transition to
eaten carbs if you go *much* slower, as in ultramarthon pace. You're still
going to run out of glycogen if you're trying to race a marathon, and a
slower start might delay that somewhat, but won't subtantially change it.

While I agree that anaerobic threshold is a debated topic these days, I
still think that it, or something like it, is a more limiting factor than
glycogen depletion. If you run too close to that threshold something
happens that will destroy you for later. Run fast but enough under it, and
you can stil run fast later to a point (perhaps up to marathon distance) --
tough you may be slowed toward the end by inevitable dehydration and/or a
natural shift in energy supply.

-Tony

Quoted message said:

They go on to show the pacing of an elite athlete in three marathons
with different pacing each time. In the race where he went out very hard
(on schedule for a 2:03:00!!) he crashed and burned, getting slower and
slower til the end, but still finished only 3 seconds off his PB, done
on even splits 3 years earlier.

Their conclusion: how well could he have done if he hadn't paced like an
idiot?

"Donovan Rebbechi" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

Parker Race said:

Got news for you, the 3:28 and 3:29 runners were reeling off miles at

almost

Quoted message said:

Quoted message said:

the same pace that you were running your last 10k after mile 20. My

coach

Quoted message said:

Quoted message said:

said those two probably went a little slower than necessary through the
half.

And seriously, I'm not sure why you still feel the need to trash my
performance. It's not like I'm the only one who had trouble after mile 20

(or

Quoted message said:

for that matter, before mile 20).

I'm not trashing it, but I thought that 6:40 mile pace was too fast a
beginning pace when you were stating that as a goal before. I didn't bother
to comment because I figured you were too stubborn to hear it.
You made an error in judgement, there's nothing wrong with that it happens
all the time.

Quoted message said:

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

"Donovan Rebbechi" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

On 2004-11-14, Parker Race <[email hidden]> wrote:

But I still beat them, right ? (-;

Yeah , but my friend Kimberly will still kick your butt and I'm sure she's
better looking.
Incidentally I talked to her this morning, her half split was more like 1:28
so she ran slightly negative splits.
I ran positive splits today, I don't recommend it 🙁

Parker