[davej]

out of interest whats the oldest ascent of Mecca?? I heard Keith Sharples was trying it how did he get on??

My money would be on Simon Reed for the oldest ascent of Mecca, possibly late 40s / early 50s?

I'm sure someone will be along to correct me soon.

Good effort

[duncan]

We need more data but...Graph 1 closely resembles plots of objective measures of athletic performance with age, suggesting assuming a linear progression in grades could be correct.  Muscle power deteriorates similarly, strength declines more slowly, supporting what jwi says about the importance of power. 

What us middle aged climbers are really interested in is people who have maintained or improved their grade at at an advanced age. That, e.g., Moon can manage 8b+ or whatever it is at the ripe old age of 48 or so is of course impressive, but not so astonishing given what he did when he was 40 and what he did in his 30s. Is Haston unusual, even unique, in managing to improve in his 40s and up until the age of 50 (in terms of sports grades at least)? I'm thinking of people who climbed to a decent standard beforehand of course - it would be easy to improve if one bumbled around in one's 20s and 30s.

Perhaps Stevie didn't focus on Sport Climbing as much as he could have in his 30s and 40s, but I don't think you did either.

Then, the crucial question, can I improve without having to shed 10kg?

It would be very hard.

[tomtom]

I'm afraid Habrich gets 2/10 for his graphs. No axis titles, no y axis units or title. Curved line to link dots - and clearly not a linear relationship as the line isn't straight.. Though it could be if the y axis were log, but you can't tell as no units. Back around...

Duncan however gets 6.5/10. Axis titles needed and a few tweaks with the label formatting at line colours would get you a first. Tick.

[petejh]

Don't worry Habrich you get better at making charts as you get older...

[jwi]

Method: Sample all scorecards from 8a.nu, for each age group compute average best grade. This would take less than an hour if 8a.nu made its database available in usable form.

If 8a.nu had an API to its database I would be willing to forgive some of Jens's jackassery

[tomtom]

Don't worry Habrich you get better at making charts as you get older...

Christ - what was he like at school then

[tomtom]

I'll hold fire on Habrichs charts and instead offer some positive re-inforcenment "well done that's better" *patrinising pat on head*

But what they do raise (possibly inadvertently) is a debate about grading. Habrichs charts could either show that there is not a linear relationship between age and ability - or that the grading scheme is not linear.. Or both!

[Johnny Brown]

Manolo's whole cv is worth a read, peaking at 9a+ at 48 amongst others.

Is there a version on the web anywhere?

He was 50 in 2008:

    9a+/5.15a:
        Bimbaluna - Saint-Loup (SUI) - 20 gennaio 2008 - Ripetizione della via di François Nicole del 2004[5]

    9a/5.14d:
        Eternit - Vette Feltrine/Baule (ITA) - 24 agosto 2009 - Prima salita, prosecuzione di O ce l'hai...o ne hai bisogno[8]
        Bain de Sang - Saint-Loup (SUI) - 2006 - Ripetizione della via di Fred Nicole del 1993[4]

    8c+/5.14c:
        Roby Present - Val Noana (ITA) - 24 marzo 2012 - Prima salita, via dedicata a Roberto Bassi[9]

    8c/5.14b:
        Eroi Fragili - Val Noana (ITA) - 5 marzo 2011 - Prima salita[10]
        Stramonio - Val Noana (ITA) - 10 ottobre 2010 - Prima salita[11]
        Thin ice - Terlago (ITA) - 25 aprile 2009 - Via di Nico Favresse del 2007[12]
        El sior Favonio - Fonzaso (ITA) - 2006 - Prima salita[13]
        Diabloluna - Fonzaso (ITA) - 7 gennaio 2006[14]
        L'Arte di Salire in Alto - Celva (ITA) - 2001 - Via di Rolando Larcher del 1992[2]
        The Dream - Val Noana (ITA) - ottobre 1991 - Prima salita

    8b+/5.14a:
        Appigli Ridicoli - Vette Feltrine/Baule (ITA) - 1990 - Ora 9a per la chiusura di due appigli[3]
        Il Maratoneta - Paklenica (CRO) - 1988
        Malvasia - Dvigrad (CRO) - 1988

    8b/5.13d:
        O ce l'hai… o ne hai bisogno - Vette Feltrine/Baule (ITA) - 1990
        Ultimo Movimento - Totoga (ITA) - 1986 - Prima salita e primo 8b italiano

A vista

    8b+/5.14a:
        Rock and Blues - Kalymnos (GRE) - 19 giugno 2009

[slackline]

Assuming linearity between age and grade...



Clearly not linear so lets fit a Loess regression line...



Data (i've added Adam Ondra, Alexander Megos, and Beat Kameerlander who climbed 8b+ (trad) at 50 with "Prinzip Hoffnung")

Code: [Select]

name,age,grade,gender
"Chris Sharma",31,9b+,"male"
"Stevie Haston",52,9a,"male"
"Manolo",51,9a,fe"male"
"Francesco Marin",60,8b+,"male"
"Lee Sheftel",59,8b+,"male"
"Herman Gollner",70,7c+,"male"
"Lee Sheftel",68,8a,"male"
"Stimson Bullitt",83,6a,"male"
"Fred Beckey",93,4,"male"
"Adam Ondra",19,9b+,"male"
"Alex Megos",20,9a,"male"
"Beat Kammerlander",50,"8b+","male"

R Script

Code: [Select]

## Filename     climber_age.R
## Created      2014-11-21
## Author       slackline
## Description  Takes data on climbers age and the grade they
##              have climbed, plots it, calculates correlation
##              and makes a poor attempt at regressing age on
##              ability (poor due to the paucity of data)

## Load libraries
## library(dplyr)
library(ggplot2)
library(mlogit)

## Read in the CSV
climbers <- read.csv("climber_age.csv",
                     header = TRUE)

## Define grade as a true factor variable
climbers$grade <- factor(climbers$grade,
                         levels = c("4", "5",
                                    "6a", "6a+", "6b", "6b+", "6c", "6c+",
                                    "7a", "7a+", "7b", "7b+", "7c", "7c+",
                                    "8a", "8a+", "8b", "8b+", "8c", "8c+",
                                    "9a", "9a+", "9b", "9b+", "9c", "9c+"))
climbers$gender <- factor(climbers$gender,
                          levels = c("female", "male"))


## Regress age and gender onto grade
##
## This means that if/when female data is added the difference
## between the sexes will be accounted for in the variation
## observed in grade
##
## NB - At present gender _isn't_ included because all of the
##      data is on males only.  If/when data on females is
##      available/included then you should change each and
##      every 'grade ~ age' to 'grade ~ age + gender' and
##      this will then account for differences in gender
##      that affect the grade climbed
##
## First we fit a linear relationship between age and grade
## accounting for the differences between gender.  This is
## quite crude, but serves as a starting point
regress.linear <- lm(grade ~ age,
                     data = climbers)
## Look at the results, the number underneath the age variable
## tells you how much of a decrease in grade occurs with an
## increase in age by 1 year.
regress.linear

## But the relationship might not be linear so now we fit
## a regression model using the Loess method which performs
## local regression (see http://en.wikipedia.org/wiki/Local_regression)
regress.loess <- loess(as.numeric(grade) ~ age,
                       data = climbers)

## But thats still not the best approach since grades whilst
## offering some indication of relative difficulty might not
## show a linear relationship, so now try fitting a multinomial
## logistic regression model
##
## NB - Doesn't actually run yet
## regress.mlogit <- mlogit(grade ~ age,
##                          data = climbers,
##                          shape = "wide")

## Plot the Relationship
##
## This is done twice, the first one fits a line that represents
## a linear relationship as fitted by the linear grade regress
png(filename = "age_grade_linear.png", width=1024, height=768)
plot.linear <- ggplot(climbers,
                      aes(age, as.numeric(grade),
                          shape = gender,
                          fill  = gender)) +
                 geom_point() + geom_smooth(method = lm,
                                            fullrange = FALSE) +
                 xlab("Age (Years)") + ylab("Grade (French Sport)") +
                 scale_shape_manual(values = c(1, 2)) +
                 scale_y_continuous(breaks = seq(1:length(levels(climbers$grade))), labels = levels(climbers$grade)) +
                 ggtitle("Climbers Age and Maximum Grade Climbed (Linear)")
plot.linear
dev.off()
## The second fits a curve by a method called Loess which performs
## local regression
png(filename = "age_grade_loess.png", width=1024, height=768)
plot.loess <- ggplot(climbers,
                     aes(age, as.numeric(grade),
                         shape = gender
                         fill  = gender)) +
                geom_point() + geom_smooth(method = "loess",
                                           fullrange = TRUE) +
                xlab("Age (Years)") + ylab("Grade (French Sport)") +
                scale_shape_manual(values = c(1, 2)) +
                scale_y_continuous(breaks = seq(1:length(levels(climbers$grade))), labels = levels(climbers$grade)) +
                ggtitle("Climbers Age and Maximum Grade Climbed (LOESS)")
plot.loess
dev.off()

If you can be bothered to collect more data download the CSV and R script and add the data to the "climber_age.csv" in the correct format (if including females then enter their gender as "female" and NOT "Female").  Then download RStudio and you can re-run the script (you might have to install.packages("ggplot2") in order to plot the data).  There are instructions within the script on how to account for gender in the regression models (the results of which I've not included in this post as I've not got time to explain them and besides there is too little data to say anything meaningful).  Shouldn't need to do anything to code for the plots to get them to display each gender (although the Loess plots confindence intervals might not differentiate correctly).


Whats really needed is more data though, multiple peoples maximum grades at different ages, since currently this is cherry picking the most exceptional ascents at each age, which is not really a "benchmark" which is commonly held to be a common standard.  There are perhaps two relatively simple approaches to addressing this, firstly scrape the data from 8a.poo or I guess I could setup a survey for denizens here to complete (and include historical aspects of people's maximum grades in the past and the age they achieved them), but I don't really have the time.

Since this is vaguely related to previous threads on benchmarking and a few people asked recently for a copy of those results the PDF is available here (as the Wiki where I originally posted it no longer hosts PDFs).  I might if I can find the time and motivation re-run the UKBenchmarking as there has been more data collected in the form I left open, but thats unlikely to happen anytime soon.

[SA Chris]

Just count the rings...

I would be hoping they each had just one. In the usual place?

[Fiend]

Fuck the charts, get back to the real issues of how did these guys manage it, what was their previous experience, how did they avoid injury, how many of them DIDN'T have a background in other regular intensive exercise activities, etc etc. Need something to read in between doing weighted pull-ups in the park...

[tomtom]

Just count the rings...

I would be hoping they each had just one. In the usual place?

Twas a dendrochonological reference... I wood have hoped you'd twigged.

[SA Chris]

I know exactly what it was. I was just trying to piggyback a shit joke on the rear end of yours.

[tomtom]

I know exactly what it was. I was just trying to piggyback a shit joke up the rear end of yours.

Does this mean we pun about sphincters or trees now - I'm confused

[SA Chris]

Neither. For fear of the wrath of Dense.

[Palomides]

For the women there's Muriel Sarkany with PuntX (9a) at age 39 - not sure of any others?

Audrey Sniezek did Lost Horizons last summer - 8c at 42 according to 8a.nu

[petejh]

Clearly not linear so lets fit a Loess regression line...



Data

This graph should be perma-pasted to 'Significant repeats' and 'News' as a handy visual aide to calibrate spray against:
''Climbed 8c!''. Are you 50? Back around.

[cheque]

 

Also needs merging with a graph of youngest ascents so we have a definitive age range that each grade is impressive within.

[JamieG]

Very nice plots slackline. Just wondered about a couple of things.

Clearly grades are subjective at the best of times and really constitute categorical data rather than continuous data. But with those caveats can you correct the data some how for the fact that clearly going from 6a to 7a is easier than 8a to 9a. I.e. grades are really not a linear representation of difficultly. I sometimes log correct data to solve that kind of problem, but not sure if that would make it worse in this case. What do you think?

[JamieG]

Haha, sorry Habrich!

I looked at the graphs and then thought . . . . well exactly what everyone else is thinking, but didn't read the comments!   
 

[JamieG]

 

I feel like such a spanner now I've read through most of the thread . . . . back around Jamie, back around!

Anyway, very interesting topic. I have no excuses being only 30. I'll have to wait until i'm in my 80s before being cutting edge (if i'm alive!) :-)

[tomtom]

I think you'll find that I made a similar interpretation of your data...

[Stu Littlefair]

Has everyone missed the post where someone posted a graph from other sports showing the decline in absolute performance with age?

It seems to me that the correspondence in shape between this graph and habrich's first graph is reasonable evidence that the steps between grades are linear. Which is what that post  said.

[a dense loner]

I must have missed it

[a dense loner]

This is all well and good, and interesting, could we have a seperate thread for the same type of info for bouldering?