Boston analysis

this just in.

Young puts Berlin as 81 seconds faster than Boston, then gives this year's race a 93-sec advantage compared to previous Bostons. A little math puts Mutai's performance at 2:03:20-ish if run in Berlin. Sounds too fast, even compared to Renato's assessment, but this is provocative of what this performance means in the bigger picture.

That is about the most nonsensical explanation yet, and there have been many in this thread.

Written by: David Monti

(c) 2011 Race Results Weekly, all rights reserved. Used with permission.

Geoffrey Mutai’s time at last Monday’s Boston Marathon was indeed fast, but according to an analysis released today by the independent Association of Road Racing Statisticians (ARRS) his mark should not be viewed as excessively aided.

Writing in this week’s “Analytical Distance Runner,” the official ARRS publication, chief analyst Ken Young wrote: “It should also be noted that Geoffrey Mutai (KEN) who ran 2:03:02 (in Boston), was ranked #2 on the ARRS competitive rankings and he had been ranked #1 for four weeks earlier in the year. When Haile Gebrselassie (ETH) set the current world record of 2:03:58.2, he was ranked #10 in the world. Hence, it is not surprising that Mutai could be capable of significantly bettering the world record. Too bad that he ran this on a course that is not eligible for records.”

According to the Young’s Race Time Bias (RTB) calculation, a measure of how fast athletes run based on their previous performances over all distances, the times at Boston were 1:37 fast for the men and 1:42 seconds fast for the women. That worked out to 2.29 seconds per kilometer for the men and 2.41 seconds per kilometer for the women. Both figures are well within the 5 seconds per kilometer value the ARRS uses to disqualify performances as excessively aided.

“The net drop is worth about 3 sec/km,” Young continued. ”The observed times were roughly 2.4 sec/km faster than expected for the 43 elite runners making up the time comparisons. Hence, one could conclude that the wind aid was sufficient to balance out the effect of the hills (such as they are).”

Young also pointed out that there were years at Boston where athletes ran much faster per kilometer than this year, based on what would be expected. The fastest year for men was 1981 when times were fast by 3.8 seconds per kilometer (2:40 total), and 1988 and 1991 for the women, where times were 4.2 seconds per kilometer (2:57 total) fast.

Interestingly, Young points out that the RTB for Boston over all years for which the ARRS has kept data averages out to +0.3 seconds per kilometer, essentially the same as a flat course. That would seem to indicate that over a long span of time the effects of the weather have canceled out (favorable and unfavorable), and the up and down nature of the course generally produces honest times, even if the course falls by 3.23 meters per kilometer, more than three times the amount allowed for setting world records.

The male and female athletes who ran the fastest based on what was expected were Ecuador’s Franklin Tenorio, whose 2:17:56 was 8:51 fast, and Colombia’s Yolanda Caballero, whose 2:26:17 was 11:01 fast.

“These RTB values for Boston are well within the range for statistically valid performances and marks from this race are included in the ARRS rankings,” Young concluded.

It's the most rigorous analysis presented in the past week though one I found surprising.

Dan K. wrote:

It's the most rigorous analysis presented in the past week though one I found surprising.

Where is the analysis? I am interested in the actual algorithm. Without that it is meaningless.

You say it is "the most rigorous analysis presented...". What is "it"? Just the text? That doesn't mean much of anything - just a conclusion that sounds like any other opinion. Show me what it is based on. What is the algorithm used? What data is used/discarded? That's what I'd like to know. Otherwise it is just, well not much really.

Wheres the Beef wrote:

Dan K. wrote:

It's the most rigorous analysis presented in the past week though one I found surprising.

Where is the analysis? I am interested in the actual algorithm. Without that it is meaningless.

You say it is "the most rigorous analysis presented...". What is "it"? Just the text? That doesn't mean much of anything - just a conclusion that sounds like any other opinion. Show me what it is based on. What is the algorithm used? What data is used/discarded? That's what I'd like to know. Otherwise it is just, well not much really.

I assure you that the method for acquiring the Race Time Bias uses sound statistical and mathematical methods. And I will personally post a good description of them if you really need them. But you are missing the point of the analysis.

Young (and the ARRS) is stating that the performance we all witnessed on Monday was STATISTICALLY VALID, and that we should not be surprised to see similar performances in the near future on record ELIGIBLE courses.

Or, to state it differently, there is VERY GOOD likelihood that Haile's record could fall in a fall marathon, but it probably won't be quite as fast as Mutai ran Boston (but they have 6 months to train).

Canova wrote: "I want to say again that coaches have the possibility to know everything can happen to their athletes, and, if they are honest with themselves, can have a real picture about the current situation of the athletes themselves. Other people speak without knowing the reality before the race, or supposing to know what happened AFTER the race." end quote

If that's true than why you had Florence Kiplagat run? She ended up in hospital? I hope that wasn't according to plan?

How is she doing is maybe a more important question though.

Again, you're not the coach of Geoffrey Mutai so all you say is speculation or otherwise on a very thin layer of actual facts.

Why do you want to claim glory and yet other marathoners under your guidance never even did close to the same? Didn't even break 2:06. And how many broke 2:07 (2?)

I am not taking away your success with Mosop but wait till its confirmed. Untill that I advice you not to be that convinced there was no wind factor.

Here are some observations about Ken Young's analysis:

- "Statistically valid" is some (arbitrarily?) cutoff defined by ARRS, and unique to them.

- Boston's overall average bias is usually 13.5 seconds slow, compared to London's 38.8 seconds fast, and and Berlin's 68.5 seconds fast. So it would be incorrect to say the "data averages out to +0.3 seconds per kilometer, essentially the same as a flat course", since, on average London is usually 52 seconds faster, and Berlin is usually 82 seconds faster.

- So if 2011 was 1:37 seconds fast (for men), compared to an expected bias 13.5 seconds slow, then in reality, 2011 was 1:50.5 seconds faster than Boston's average bias.

So 2:03:02 would have been 2:04:52, in an average year.

Likewise 2:04:58 would have been 2:06:48.

2:04:52 is still an amazing time for Boston, as is 2:06:48, considering previous course records.

The recent drops are amazing.My point is that if the wind only accounts for 40 seconds, we should begin to see (more) major drops in the official WR as well as AR.

TorontoObserver wrote:

No need to wait. It has been happening for the past few years.

World record has dropped to 2.03.59 from 2.05.

Many more runners are now running 2:04, 2:05, 206 when 2.07, 2.08 and 2.09 were the winning times.

Even for American, Ryan Hall's 2.06.17 was a significant drop from 2.08 and 2.09 than the best American born marathons had run.

Men marathoning are progressing by minutes, not seconds, on average.

There is no reason why the new top marathon runners can't improve from 2.04, 2.05 to 2.03 and 2.04 in batches.

Look on the other thread or on arrs.net, it's a regression algorithm and uses lots of data to rank athletes over time. Is it 100%right? Who knows. But it takes into account far more than anybody else on these boards have done so far in a systematic way.

I mostly agree with this. However, I don't agree that Hall was helped all that much more than thwe other elite athletes. More in the ballpark of 15 to 30 seconds in my opinion.

TorontoObserver wrote:

No need to wait. It has been happening for the past few years.

World record has dropped to 2.03.59 from 2.05.

Many more runners are now running 2:04, 2:05, 206 when 2.07, 2.08 and 2.09 were the winning times.

Even for American, Ryan Hall's 2.06.17 was a significant drop from 2.08 and 2.09 than the best American born marathons had run.

Men marathoning are progressing by minutes, not seconds, on average.

There is no reason why the new top marathon runners can't improve from 2.04, 2.05 to 2.03 and 2.04 in batches.

Someone needs to step away from the bar.

2:03:58.2 Haile Gebreselasie-2 28 Sep 2008 Berlin GER

2:04:26 Haile Gebreselasie (ETH) 30 Sep 2007 Berlin GER

2:04:55 Paul Tergat (KEN) 28 Sep 2003 Berlin GER

2:05:37.8 Khalid Khannouchi-2 (USA) 14 Apr 2002 London ENG

2:05:42 Khalid Khannouchi (MAR) 24 Oct 1999 Chicago IL USA

2:06:05 Ronaldo daCosta (BRA) 20 Sep 1998 Berlin GER

danny12345 wrote:

I assure you that the method for acquiring the Race Time Bias uses sound statistical and mathematical methods. And I will personally post a good description of them if you really need them. But you are missing the point of the analysis.

Young (and the ARRS) is stating that the performance we all witnessed on Monday was STATISTICALLY VALID, and that we should not be surprised to see similar performances in the near future on record ELIGIBLE courses.

Or, to state it differently, there is VERY GOOD likelihood that Haile's record could fall in a fall marathon, but it probably won't be quite as fast as Mutai ran Boston (but they have 6 months to train).

Sorry, but I'm not one to be impressed by things like, "I assure you..." That's nice but now can we get back to what I asked in the first place.

Regarding, "I will personally post a good description of them if you really need them." Yes. Please do so. That is what I asked already. So I guess I am asking for the same thing once more.

Finally, regarding, "But you are missing the point of the analysis." Um, no I am not. I understand quite well what Young (and the ARRS) is stating. But again that was not my question. My question was as to the validity of the analysis. And sorry, but appeals to authority don't do a whole lot for me. Let's see exactly how this analysis was done.

Now, since I've asked twice, and you've kindly offered to "post a good description of them", would you kindly proceed to do just that.

Thanks in advance.

Dan K. wrote:

Look on the other thread or on arrs.net, it's a regression algorithm and uses lots of data to rank athletes over time. Is it 100%right? Who knows. But it takes into account far more than anybody else on these boards have done so far in a systematic way.

OK, I went to arrs.net. Thanks for pointing to that for me. However, I do not see an algorithm posted on their home page. Am I missing something? Could you be more specific? Where exactly do I find the actual algorithm used?

Thanks.

- Still Wondering Where's the Beef

check out the progression of marathon world bests on the arrs.net web site. With all of the statistical smoke and mirrors they still lists marathon performances pre 1983 that we have absolutely no knowledge of the accuracy of those courses. Course measuring standards and rules weren't agreed until after the early 80's. Many of those routes were measured by cars.

Perfect storm, I think you don't understand the job of a coach that is a methodologist of IAAF, too.

I cant analyse what my athletes do, only. How I said in some other thread, before being a teacher you must be a student, and, when you are teacher, you must continue to be student.

If I want to understand how training methodology changes, I need not only to look at MY methodology, but at all what happens in training from other athletes and from other coaches, that I can know.

Do you remember, many years ago, a magazine with the title "How they train" ? The author of the magazine was not the coach of anybody, but was able to put together the infos about the individual training of many top athletes, and analysing what they did it was possible to change training in many occasions.

The fact is that coaches and athletes able to do something new always existed, and the most part of time came BEFORE any methodological and scientific explanation.

Never there was a scientist able to overtake some barrier in training. Always was some coach (many times without big scientific knowledge) having the mentality of an explorer, giving programmes for his athletes out of the normal trend, able to find athletes so strong and so motivated that overtook the limits, starting from a different mental vision of the value of the performances.

This happened, for example, with Kip Keino (WR of 3000m from 7'45"2 to 7'39"6), with Ron Clarke (from 28'15" to 27'39"), with Gebrselassie, with Paula Radcliffe.

Only AFTER their performances who was able to analyse their training discovered new frontiers in training.

So, when I speak about what athletes that I don't train do in their training, is not for claiming glory, but in simple way is for finding reasons in what happens.

And, if you find a common denominator in the training phylosophy of Geoffrey Mutai and Moses Mosop, this is a point that can become interesting for everybody, and can change something in the traditional methodology of training.

About Florence Kiplagat, what happened is completely casual and totally strange. The doctors of the Boston Hospital, after a lot of different investigations, supposed she was victim of a virus that blocked her nervous system. Of course it's not possible to think before the race something like this, so your question is a nonsense (what happened is like when Bekele dropped out in Mombasa : do you think it was part of his plan ?).

A point about the question of course difficulty;

Take a look at the posted running routes on USATF.org and click on the elevation data for courses you know to be fast or slow.

You will find that the faster course have no more than about 400 to 500 feet of elevation GAIN whether or not it is net uphill or downhill.

Some slower courses may actually be net downhill but the 800 plus feet of incline produces slower times.

Boston has over 900 feet of elevation gain, similar to the Flying Pig and the 1200 feet of downhill has not historically made the course as fast as the other major courses in the world. Most veteran Boston runners tend, in fact, to characterize the downhills as detrimental when combined with the significant elevation gain.

You know what? The week this board has been flooded with anonymous newbies who have a reason for everything, except the obvious. Board regulars, seasoned runners and/or fans who we all know all agree with what happened on Monday. (with the exception of Renato, of course, who has a horse in this race). We all can see the elephant in the living room. On the other side of the coin, it’s nothing but random handles like adsfgdhsfdh, Danny12345, Crimea River, etc, who have their eyes wide shut and caterwaul at every rational explanation, and return to the debate with nothing but red herrings. The rational side of the debate is told that they’re bitter and jealous, or they hate Ryan Hall, or that they don’t understand the new runners , or this or that, or something else – but the depth of the well of excuses and red herrings is endless.

It’s the wind Stupid.

Now I don’ have a lot of confidence in the assurance of a purely random poster named “Danny12345” saying “I assure you that the method for acquiring the Race Time Bias uses sound statistical and mathematical methods.” And “Young (and the ARRS) is stating that the performance we all witnessed on Monday was STATISTICALLY VALID.” That about as meaningless as you can get without reading it from a teleprompter.

I can assure you that any set of results you put up will be “statistically valid” . I saw the Wizard of Oz too. I was scared shitless by those f***ing monkeys but the one thing that I was sure of was that Great and Powerful Oz was trying to hide from Dorothy. The problem here is that like the Wizard of Oz you are hiding your methodology behind the curtain, and Toto is tugging at it.

On Monday we had at least four black swan events, and your saying that it was all a statistically valid happening – as if whacking 3 minutes of the course record and a minute off of the world record, which is held by the greatest runner who has ever lived, is STATISTICALLY VALID? It wasn’t even four black swans, it was more like two plaid swans, a herringbone swan, and a houndstooth check swan.