700/23 vs 700/25 tires ?

In article ,
Michael Press wrote:

In article ,
Tim McNamara wrote:

In article ,
Michael Press wrote:

In article ,
Tim McNamara wrote:

It's all stochastic, man.

No, it is not.

Ooh, nice comeback. I can't speak for anyone else, but I'm
convinced now. Now perhaps you might insert an illuminating fact
here.

It is not all stochastic. Fact.

That is, I'd guess, the best we're going to get from you in this thread.
Too bad.

In article
,
" wrote:

On Mar 5, 7:18*pm, Tim McNamara wrote:
" wrote:
On Mar 3, 5:40*pm, Tim McNamara wrote:

Greg has only three other competitors in his class? *That's the
only way he'd have a 25% chance of winning each race.

This is statistically illiterate. *You believe a version of the
RBR joke that whenever there are two outcomes, the chances are
always 50/50.

Chance *is* 50/50 when there are only two possible outcomes.

No it isn't. This is innumeracy.

I might win the lottery tomorrow. Or I might not. There are two
possible outcomes. Are the chances 50/50 ?

I could arm wrestle Chuck Norris. One of us is going to win. Would
you put even odds on me?

You are restating a common misconception - that probabilities of
outcomes must necessarily be uniform. In fact everyone knows this is
not the case; Joe Average is perfectly capable of understanding the
spread on football games.

You are conflating chance and probability.

In article ,
John Forrest Tomlinson wrote:

On Sun, 08 Mar 2009 22:18:33 -0500, Tim McNamara
wrote:


Billions of people believe in one or more invisible cosmic friends.
Does that make them right?

Oh yeah, because many people overestimate the effect of bikes,
therefore bikes don't have much effect at all. The reactionarism of
RBT is always amusing.

Yes, it is.

In article ,
John Forrest Tomlinson wrote:

On Sun, 08 Mar 2009 22:18:33 -0500, Tim McNamara
wrote:

In article ,
John Forrest Tomlinson wrote:
For something you don't have data on but the vast majority of
people do it one way that's differen than, say, 25 years ago, and
you keep doing it the old way just because of the lack of
non-experiential evidence? Really?

Oddly enough, bike racing isn't very different than it was 25 years
ago. Training methods are somewhat different (way better drugs, for
one thing) and equipment is a little different in design but
fundamentally the same in function (due to UCI rules, to a great
extent).

It's interesting how you throw up stuff like this. Has anyone here
said bikes are "fundamentally" different? You're like Frank K,
"arguing" that fast bikes are very important in recreational rides.

The ball is still in your court, no matter how much you try to deflect
it. You claim that there are hundreds of tiny changes- with no
discernible effect themselves- that add up to a significant competitive
advantage. As of yet, you've identified none. Still waiting and at
this point guessing that we will wait until this thread is long over
with and you still won't come up with a list of those changes for us to
discuss.

Also interesting that you don't answer my question, which is not
about fundamental differences, but small differences.

The "question" about small differences has already been answered. Why
answer it again?

On Mon, 09 Mar 2009 20:27:05 -0500, Tim McNamara
wrote:

The ball is still in your court, no matter how much you try to deflect
it. You claim that there are hundreds of tiny changes- with no
discernible effect themselves- that add up to a significant competitive
advantage.


I don't recall the "hundreds" but maye I said it. More likely I said
"many."

But in terms of hundreds: 10 grams. Add it up a hundred times.

There's one.

As of yet, you've identified none.

Pure balonely. In this group we've heard of weight, gears, aerobars,
tire quality, wheels. (I know you've accepted aerobars.).

Still waiting and at
this point guessing that we will wait until this thread is long over
with and you still won't come up with a list of those changes for us to
discuss.

I'm still waiting for your answer on what year's bike you'd race,
assuming you were allowed your aero bars. What's the problem with a
bike from 1907 again? Oh yeah, you said the question didn't make sense
or some such dodge.

In article
,
" wrote:

On Mar 6, 9:23*am, Tim McNamara wrote:
In article , *John
Forrest Tomlinson wrote:

On Thu, 05 Mar 2009 20:20:04 -0600, Tim McNamara
wrote:

Of those 2000+ races, how many more did you win because of that
1% improvement that you wouldn't have won otherwise? *And how do
you know it was *that* 1% improvement?

it can't be known with certainty because races are too complex. *

What Ben has shown us is that it can't be known if the effect size
of the improvement is smaller than the error of measurement. *This
is a well-known problem in statistics and is a good working
definition of negligibility.

All I showed is that Frank's proposed experiment had too small a
sample size to detect a reasonable effect.

That is not a good working definition of "negligible" because it
depends on sample size.

Since by your description it is virtually impossible to have a big
enough sample size to find a reasonable effect size at a reasonable
level of confidence, you've given us a good working definition of
negligibility.

Bret's 2000 races would be a somewhat better sample size, but there
is no way to control an experiment retroactively. If there was some
small change - for example, maybe he trained smarter than everyone
else all those years - that allowed him to increase his winning
chance by 1%, though, that would be 20 races, which most would not
consider negligible.

The problem, again, is that you've shown it is impossible to isolate
that small factor because of the large influence of random events. The
outcome of races is more reliably attributable to differences in
physiological capacities (Vo2 max, power at threshold, endurance, etc.)
and to competitive decisions (when to attack, when to draft, etc.) than
to a different derailleur or a new pair of tires. In order to achieve a
1% improvement in race outcomes over a large number of races, my guess
is that the magnitude of the mechanical or training improvement or race
strategy has to be much larger than 1%- perhaps an order of magnitude
greater.

Another example: the house in casino gambling typically has an
advantage of just a few percent in any given bet (for many casino
games, not slots). If you didn't know this, and placed 10 or 20 bets,
you'd win close to half, and lose close to half. You wouldn't be able
to tell that the house had an advantage. Does this mean that the
house's advantage is negligible? In the long run, over many bets and
many bettors, the house does not think so.

Of course, if all you ever do in the casino is make 10 small bets and
then leave, maybe you don't care that the house is taking a little
from you on average. Cheap entertainment and all that. But if you
care about winning, or losing as slowly as possible to prolong the
entertainment, then you might care about small differences in the
odds. Casinos have better odds than state lotteries; blackjack has
better odds than roulette; both have better odds than slots. There
are entire websites devoted to this, so somebody doesn't think it's
negligible.

The house's advantage is overwhelming in many of these games- they win
every time the customer loses, and there are far more ways to lose than
to win. Bike racing does not function the same way. Blackjack is
perhaps the best analogue.

In article ,
Tim McNamara wrote:

In article ,
Michael Press wrote:

In article ,
Tim McNamara wrote:

In article ,
Michael Press wrote:

In article ,
Tim McNamara wrote:

It's all stochastic, man.

No, it is not.

Ooh, nice comeback. I can't speak for anyone else, but I'm
convinced now. Now perhaps you might insert an illuminating fact
here.

It is not all stochastic. Fact.

That is, I'd guess, the best we're going to get from you in this thread.
Too bad.

`It's all stochastic' is your best?

--
Michael Press

In article ,
Tim McNamara wrote:

In article
,
" wrote:

On Mar 5, 7:18*pm, Tim McNamara wrote:
" wrote:
On Mar 3, 5:40*pm, Tim McNamara wrote:

Greg has only three other competitors in his class? *That's the
only way he'd have a 25% chance of winning each race.

This is statistically illiterate. *You believe a version of the
RBR joke that whenever there are two outcomes, the chances are
always 50/50.

Chance *is* 50/50 when there are only two possible outcomes.

No it isn't. This is innumeracy.

I might win the lottery tomorrow. Or I might not. There are two
possible outcomes. Are the chances 50/50 ?

I could arm wrestle Chuck Norris. One of us is going to win. Would
you put even odds on me?

You are restating a common misconception - that probabilities of
outcomes must necessarily be uniform. In fact everyone knows this is
not the case; Joe Average is perfectly capable of understanding the
spread on football games.

You are conflating chance and probability.

Start here.
http://en.wikipedia.org/wiki/Cox's_theorem

Then look into a good book on probability such as
http://www.amazon.com/Theory-Probability-Classic-Physical-Sciences/dp/0198503687

--
Michael Press

On Mar 9, 6:19*pm, Tim McNamara wrote:
" wrote:
On Mar 5, 7:18*pm, Tim McNamara wrote:
" wrote:
On Mar 3, 5:40*pm, Tim McNamara wrote:

Greg has only three other competitors in his class? *That's the
only way he'd have a 25% chance of winning each race.

This is statistically illiterate. *You believe a version of the
RBR joke that whenever there are two outcomes, the chances are
always 50/50.

Chance *is* 50/50 when there are only two possible outcomes.

No it isn't. *This is innumeracy.

I might win the lottery tomorrow. *Or I might not. There are two
possible outcomes. *Are the chances 50/50 ?

I could arm wrestle Chuck Norris. *One of us is going to win. *Would
you put even odds on me?

You are restating a common misconception - that probabilities of
outcomes must necessarily be uniform. *In fact everyone knows this is
not the case; Joe Average is perfectly capable of understanding the
spread on football games.

You are conflating chance and probability.

That's all you have?
Captain Kirk, I'm receiving an unusually strong
Dunning-Kruger signal from this quadrant.

What does it mean to say the "chance is 50/50 when
there are two possible outcomes"? Most people use
"50/50" to mean 50 percent. I would assume that
50 percent is a statement of a probability. Do you
have some other meaning that I don't know about?

Seriously, Tim, take a refresher course in
statistics, you owe it to your clients.

Ben

On Mar 9, 6:45*pm, Tim McNamara wrote:

Another example: the house in casino gambling typically has an
advantage of just a few percent in any given bet (for many casino
games, not slots). If you didn't know this, and placed 10 or 20 bets,
you'd win close to half, and lose close to half. You wouldn't be able
to tell that the house had an advantage. *Does this mean that the
house's advantage is negligible? *In the long run, over many bets and
many bettors, the house does not think so.

Of course, if all you ever do in the casino is make 10 small bets and
then leave, maybe you don't care that the house is taking a little
from you on average. *Cheap entertainment and all that. *But if you
care about winning, or losing as slowly as possible to prolong the
entertainment, then you might care about small differences in the
odds. *Casinos have better odds than state lotteries; blackjack has
better odds than roulette; both have better odds than slots. *There
are entire websites devoted to this, so somebody doesn't think it's
negligible.

The house's advantage is overwhelming in many of these games- they win
every time the customer loses, and there are far more ways to lose than
to win. *Bike racing does not function the same way. *Blackjack is
perhaps the best analogue.

Your description of why the house wins at casino
gambling is incorrect. For many bets, there are more
ways to lose than win, but the house pays off a win at
more than 1:1. For some bets, the odds are only a little
worse than 1:1 and the house pays off at 1:1. The ratio
the house pays off at is carefully chosen to give the house
a small advantage. If the advantage were overwhelming,
the customers would lose money too quickly and not
come back.

Consider roulette. There's 38 numbers on a Vegas
roulette wheel. If you bet on a single number the house
pays off at 35:1. If you bet on red or black the house
pays off at 1:1 and its small winning margin comes
from the green numbers. More info and an odds table he
http://en.wikipedia.org/wiki/Roulette

If a roulette wheel was a bit uneven, it would take
hundreds of tries to reveal at 3-sigma significance
that some numbers were a few percent better.
However, someone who had been lucky enough
to be betting on the good numbers would be,
_on average_, ahead by that few percent. Not negligible.

Most of the time in bike racing you don't win,
even if you are Bret. If you can do something to
increase the odds of winning, it helps, even though
it will be difficult to show by a post facto evaluation
of statistics that you improved. Model-building helps
to evaluate claims of improvement, because only in
very unusual cases would you be able to show a
direct causal link between the change you made
and a win.

This really is not restricted to equipment despite your
insisting that it is a gearhead question.
If I trained harder and improved my power at
lactate threshold by 5-10 watts, I think it would improve
my chance of winning races (if I got off my ass and
raced). However, it would be difficult to analyze any
given race post facto and say that my power
improvement was the deciding factor. Except for
time trials, which are easier to analyze, but I have
heard in this thread that they don't count.

Ben