ALMS Rnd 5: Utah 17-19 May 2007

[Fogelhund]

Originally Posted by sportscanyltics

Per your statement about the race data, the main distinction is you don't think race data is transferable to other races.

Meaningful data, used properly, can be transferable with the proper tools, and computations. I just don't see that happening in this case.

[TWK]

Without a calculator of any kind, only my eyes, a cursory look at timing and scoring, and following the flow and strategies of the race in real time, like any fan, here is my conclusion following Miller:

In 2006 on this track, Porsche needed a late yellow to get into position to challenge Audi for the win. This season (regardless of Allan’s protestations) we think the same late caution could well have meant a win for Audi. Certainly there is no doubt that the larger restrictor allowed Porsche the win at this track, but it wasn’t exactly a parade, was it? So the rule has put Miller Motorsports Park in the Porsche column. Would any of the previous races been any different with the ACO mandated 5% reduction for the LMP 2s? Not likely. The R10 was nearly helpless at Long Beach. The Audis had a tough time at Houston. The other two – Sebring and St. Pete – they won. Of seven remaining venues, three are almost certainly Audi wins, regardless of the restrictor rule. Similarly, two are likely for Porsche. That leaves two in the balance. Adopt the ACO restrictor for the Porsches tomorrow, and the outlook is just the same.

No mean, golden or otherwise, required. No deviation (deviant?) standard or otherwise, required.

[jhansen]

I find the exercise of analyzing race data to be interesting. However, race engineers and the like go much farther beyond the "mean" laptime and other basic statistical presentations of the data. Race engineers have access to a wealth of information that we will never see. Not to mention, they are much better placed to estimate fuel loads, types of tyres, etc. No disrespect, but I just don't see how you can draw any meaningful conclusions with the data that we have available. Not only that, but then to be able to forecast for the upcoming races? Seems a bit far fetched.

[chewymonster]

Originally Posted by JAG

The point is a factory petrol P1's should have over 700bhp, otherwise somethings wrong.

AER and Judd claim 650bhp, so likely have 20-40bhp more, and you would think a factory team, if battling like Acura and Porsche are, would be pushing 700bhp+.

The anti disel agenda some people are waging expects us all to believe P2's are pushing 550bhp+, diesel P1's 700bhp+, but find it inconceivable a factory, petrol, P1 could be making more than 650bhp, so clamour for the diesels to be hobbled.

A petrol P1 could theoretically be the best of package for most circuits, big power, but a nimble package to negotiate slow corners were P2's excel, but diesels struggle.

Kneejerk responses to Peugeot and Audi, based on comparisons with privateers like Dyson, convince me evermore the ACO are right to sit back and see how things play out for a little longer.

Very much true. Surely Porsche/Acura can make a P1 car as quick as their current P2??? But then they will lose the fuel economy advantage of running a lighter vehicle that is just as quick (but shouldn't for all common sense)

[sportscanyltics]

Originally Posted by Fogelhund

Meaningful data, used properly, can be transferable with the proper tools, and computations. I just don't see that happening in this case.

Call me curious, but what meaningful data, proper uses, proper tools, and computations are you refering to?

[sportscanyltics]

Originally Posted by TWK

Without a calculator of any kind, only my eyes, a cursory look at timing and scoring, and following the flow and strategies of the race in real time, like any fan, here is my conclusion following Miller:

In 2006 on this track, Porsche needed a late yellow to get into position to challenge Audi for the win. This season (regardless of Allan’s protestations) we think the same late caution could well have meant a win for Audi. Certainly there is no doubt that the larger restrictor allowed Porsche the win at this track, but it wasn’t exactly a parade, was it? So the rule has put Miller Motorsports Park in the Porsche column. Would any of the previous races been any different with the ACO mandated 5% reduction for the LMP 2s? Not likely. The R10 was nearly helpless at Long Beach. The Audis had a tough time at Houston. The other two – Sebring and St. Pete – they won. Of seven remaining venues, three are almost certainly Audi wins, regardless of the restrictor rule. Similarly, two are likely for Porsche. That leaves two in the balance. Adopt the ACO restrictor for the Porsches tomorrow, and the outlook is just the same.

No mean, golden or otherwise, required. No deviation (deviant?) standard or otherwise, required.

If one uses data (of any sort) in conjunction with the methods you use to come to a similar (or disimilar) conclusion, does it make the use of data any less viable?

Another if: If I've ever misrepresented data in any of the items on Sportscar Analytics please let me know and I will do my best to rectify it in the future.

[Spyderman]

Originally Posted by sportscanyltics

Not true, much more time in the pits (under green flag conditions as well) for the Audis was their undoing.

O.K. Call it semantics ..but then it would be more proper to say that the Audi's were quicker whilst on track during the entire race as opposed to just quicker over the course of the entire race.

Anyway, glad to hear it! Certainly takes the wind out of the sails of the pro-diesel brigade argument.

[TWK]

Originally Posted by sportscanyltics

If one uses data (of any sort) in conjunction with the methods you use to come to a similar (or disimilar) conclusion, does it make the use of data any less viable?

Another if: If I've ever misrepresented data in any of the items on Sportscar Analytics please let me know and I will do my best to rectify it in the future.

Since I have no idea what data has been used or how it has been manipulated, nor what conclusions are reached, having never read any of the material, the answer to both your questions is also "I have no idea."

It was a simple point. The paragraph I quoted was written without resort to statistical analysis.

This all reminds me of a wasted semester in a class called Operations Research. Ten thousand calculations to arrive at the obvious conclusion that the best return is to build thirty-five widgets of type "A" and seventeen of type "B" as long as one ignores the market demand for either - or neither. (Which in the real world is critical, isn't it?) If stats float your boat, knock yourself out.

[Fogelhund]

Originally Posted by sportscanyltics

Call me curious, but what meaningful data, proper uses, proper tools, and computations are you refering to?

For example, how have you adjusted your data for elevation, track temperature and ambient temperature? How do you obtain what engine mapping each team is using at this event, and how that compares to what they will use elsewhere?

[sportscanyltics]

Originally Posted by Spyderman

O.K. Call it semantics ..but then it would be more proper to say that the Audi's were quicker whilst on track during the entire race as opposed to just quicker over the course of the entire race.

Anyway, glad to hear it! Certainly takes the wind out of the sails of the pro-diesel brigade argument.

But when one is talking about lap times in relation to an entry being "quicker", isn't that usually what we're talking about? Time in the pits has nothing to do with the performance of a car, but it can have everything to do with a result of an entry. Sorry I didn't make that clear!

[sportscanyltics]

Originally Posted by TWK

Since I have no idea what data has been used or how it has been manipulated, nor what conclusions are reached, having never read any of the material, the answer to both your questions is also "I have no idea."

Fair enough, I can't really discuss it with you if you haven't read it...

Originally Posted by TWK

This all reminds me of a wasted semester in a class called Operations Research. Ten thousand calculations to arrive at the obvious conclusion that the best return is to build thirty-five widgets of type "A" and seventeen of type "B" as long as one ignores the market demand for either - or neither. (Which in the real world is critical, isn't it?) If stats float your boat, knock yourself out.

I appreciate your last statement, and I'm sorry your course work lead you to an indifferent approach to statistics. I believe that when done correctly (i.e. only utilizing stastics to approach very particular questions) and used in conjunction with the methods you use to arrive at conclusions, stastics can be an added layer of understanding.

[sportscanyltics]

Originally Posted by Fogelhund

For example, how have you adjusted your data for elevation, track temperature and ambient temperature? How do you obtain what engine mapping each team is using at this event, and how that compares to what they will use elsewhere?

But wouldn't that denote that I'm actually using engine data for whatever conclusion I'm drawing, or that whatever conclusion I've arrived at is pertinent to engine data? For whatever way I read into the results of one race, can't I also keep in mind that there are certain variables (just a few of which you mentioned above) that I must remain aware of when drawing a conclusion for any upcoming race?

[Fogelhund]

Originally Posted by sportscanyltics

But wouldn't that denote that I'm actually using engine data for whatever conclusion I'm drawing, or that whatever conclusion I've arrived at is pertinent to engine data? For whatever way I read into the results of one race, can't I also keep in mind that there are certain variables (just a few of which you mentioned above) that I must remain aware of when drawing a conclusion for any upcoming race?

That is fair enough, but now we've come back to my original statement that there are no real implications to be drawn at this time, that are relevant to future races. Incomplete information is just that, and I'm not one to draw mathematical conclusions based upon a fraction of the variables in the equation.

If we want to talk about about the race, and analyze it.. fine... but let's not pretend, or present that accurate mathematical analysis has occurred, and that we can draw conclusions based upon this.

[jhansen]

Originally Posted by sportscanyltics

If one uses data (of any sort) in conjunction with the methods you use to come to a similar (or disimilar) conclusion, does it make the use of data any less viable?

Another if: If I've ever misrepresented data in any of the items on Sportscar Analytics please let me know and I will do my best to rectify it in the future.

In my opinion, you don't have enough data. Side note, Ross Brawn (former Ferrari F1 tech. director) recently said he didn't think the tv coverage of F1 was good enough. Simply being that he did not have the level of information that he was used to when he sat on the pit wall. As fans, we just don't have enough information to conduct the kind of analysis you are suggesting. At least in my opinion.

[sportscanyltics]

Originally Posted by Fogelhund

That is fair enough, but now we've come back to my original statement that there are no real implications to be drawn at this time, that are relevant to future races. Incomplete information is just that, and I'm not one to draw mathematical conclusions based upon a fraction of the variables in the equation.

If we want to talk about about the race, and analyze it.. fine... but let's not pretend, or present that accurate mathematical analysis has occurred, and that we can draw conclusions based upon this.

I ask you again, what analysis have I done that was misrepresented by the data I used for the conclusion I drew?

[sportscanyltics]

Originally Posted by jhansen

In my opinion, you don't have enough data. Side note, Ross Brawn (former Ferrari F1 tech. director) recently said he didn't think the tv coverage of F1 was good enough. Simply being that he did not have the level of information that he was used to when he sat on the pit wall. As fans, we just don't have enough information to conduct the kind of analysis you are suggesting. At least in my opinion.

Heck, I'd love to have all of the data from every car out there... however, you have to work with what you've got, and you also have to start somewhere. In the same question I asked Brett, I'd like to ask you too: for the purpose of the questions I was trying to answer, and the considerations I made in answering those questions, where the data I used was not appropriate?

[jhansen]

Originally Posted by sportscanyltics

Heck, I'd love to have all of the data from every car out there... however, you have to work with what you've got, and you also have to start somewhere. In the same question I asked Brett, I'd like to ask you too: for the purpose of the questions I was trying to answer, and the considerations I made in answering those questions, where the data I used was not appropriate?

I think you're missing my point. What I am saying is that basic statistical analysis has no application here beyond the ability to inject numbers and make graphs. The relevant data is not known, thus the statistics have little meaning.

However, I think anyone can come to reasonable and rudimentary conclusions just by watching the race unfold (ie. lap times, when cars pit, apparent strategy changes, etc.).

[sportscanyltics]

Originally Posted by jhansen

I think you're missing my point. What I am saying is that basic statistical analysis has no application here beyond the ability to inject numbers and make graphs. The relevant data is not known, thus the statistics have little meaning.

However, I think anyone can come to reasonable and rudimentary conclusions just by watching the race unfold (ie. lap times, when cars pit, apparent strategy changes, etc.).

I'm definitely not missing your point... for the purpose of the questions I've attempted to answer, how was the statistical analysis used not appropriate or used with "little meaning"? And taking your sentiments even further, if we could know everything from just "watching the race unfold", why would results including number of laps run, fastest laps, qualifying times, gaps to the leader etc. etc. ever be published or payed attention to by even the most casual of fan?

[jhansen]

Originally Posted by sportscanyltics

I'm definitely not missing your point... for the purpose of the questions I've attempted to answer, how was the statistical analysis used not appropriate?

One more time, because it doesn't account for all variables.

[sportscanyltics]

Originally Posted by jhansen

One more time, because it doesn't account for all variables.

One more time, in which of my uses of statistical methods could I have accounted for more variables?

[jhansen]

Originally Posted by sportscanyltics

One more time, in which of my uses of statistical methods could I have accounted for more variables?

You can't, that's why I don't think it's worth while.

[sportscanyltics]

Originally Posted by jhansen

You can't, that's why I don't think it's worth while.

So once again, no answer to my question...

If you're allergic to numbers so be it, but to say that looking at lap times over the course of a race isn't "worth while" is ridiculous. A mean lap time for an entry, driver, or even a standard distribution makes absolute sense when trying to understand which car, package, driver, etc. is the fastest; ESPECIALLY in endurance sportscar racing.

Sure, you can try to make inferences about a race from just using your eyes and/or your brain, but to try and strike down someone who actually uses sound numbers (lap times, laps per stint, etc. etc.) to show his or her methods of arriving at a very specific conclusion about a particular race is absolutely ridiculous.

Furthermore, its not like I've been trying to make sweeping generalizations based on data from a single race. The only analysis any of you could be referring to is the Houston edition of Inside the Race. (Just in case you haven't read it, and maybe you haven't, it's available here: http://sportscaranalytics.wordpress....-rd-4-houston/) If a mean lap time or a variation of a mean is so ridiculous, why is it that the ACO uses the exact same method as a primarily tool for determining performance equivalency? I just think that maybe you're mixing apples and oranges here in what I've written and what you think I'm trying to accomplish.

[Adam43]

Perhaps a bit of time out from this is a good idea. Thread closed.

It will also give me time to read today's postings!

[Adam43]

This is not the place for a general debate on the usefulness of statistics. At the very least it is tedious and as the above shows sportscar fans don't seem to be very good at discussing that particular topic!

Specific statistics relevant to Sportscar racing are allowed here and can have merit. However they are not gospel and the specific statistic can be challenged.

Key challenges may concern accuracy, confidence, and the assumptions made.

Thread re-opened.

[Adam43]

I have a few questions.

How long did the top runners actually spend in the pits?
Which laps are (and aren't) considered for the mean?

What race are we talking about? Houston or Salt Lake? Mainly Salt Lake?