Okay, Darin. History review...
In the fall of 2001, members of this board were kicking around conversations about power-to-weight ratios of various cars and Bill M. suggested something that really clicked for me - the idea of using different multipliers to baseline weights for cars in each IT class. I started an Excel doc (an evolution of which I still have) that allowed me to start with estimates of OE- and IT-spec horsepower and weights.
I used some existing estimates to ballpark multipliers - the "Miller Ratio" number - that could then be factored onto power figures to (a) determine if cars were comparably classified, and (
ultimately treat weight as a dependent (or outcome) variable. This thinking was applied to a growing dataset of cars and, back-of-the-envelope, proved to be a pretty good predictor of real-world competitiveness among existing cars in the classes where they were actually running: That is to say, a car that fell out "light" of the average MR number for a class (better power/wt ratio) tended to be more successful in the results. Conversely, cars that fell out "heavy" tended to be dogs and the farther out they were, the doggier...
Problem was that there was a big gap emerging between A and S. Understand that my initial effort at promoting change was to apply a formulaic classification scheme to C, B, A, and S. History will tell you that this idea was roundly shouted down, supporting my belief - and that of a lot of other researchers - that people just don't understand statistics, particularly if the numbers are contrary to what they want to hear.
Given the lack of enthusiasm - ahem - demontrated for a major overhaul, I started toying with the idea of creating a new class to address some of the commonly-complained-about problems at the top of A, and both ends of S: IT2 came to life but that's not the topic of today's lesson. A later evolution of the formula applied a second multiplier to engine type, to accommodate the fact that not all engines gain the same kind of power going from stock to IT trim.
A formula could indeed be applied to the existing four classes but it would result in some really nasty compromises - hundreds of pounds of ballast in some cars, others left with no chance of meeting the minimum required to meet their class index, and lots of movement from class to class. The current MR numbers for the five-class structure are...
ITC = .048
ITB = .058
ITA = .0632
IT2 = .069
ITS = .082
The race weight formula is...
(Stock HP x engine multiplier / MR number) + 200
You will notice first that the MRs are NOT linear. They were devised to impose the least amount of weight change on existing cars as was practical. Even at that, some cars just can't stay where they are if these "best fit" numbers are applied, since they require too much ballast to be safe or can't make the required minimum. The breaks between class MR numbers are also not consistent because auto makers tend to build cars in "classes." The numbers had to accommodate cars that were already classified, remember. I couldn't just pick cars that were alike and leave others out in the cold.
The short answer Darin is that it started there - with 4 classes - but practicalities made it unworkable, even if one DID agree with the first principles of the thing. And nobody here did.
That said, your point (Darin) about making progressively tighter estimations - analogous to the foundation of calculus - is an excellent one. Each successive estimate (all of which can be seen as using the same kind of energy, time, or money) gets you closer but LESS closer each time. This kind of thinking is at the center of my long-running arguments FOR formulas. The key is coming up with one that takes the most variance out of the "competitiveness" equation as is practical, making the first estimate that gets you the most bang for your buck. We actually get TWO cuts using the current formula - engine and class multipliers.
The good news is that taking ALL of the variance due to make/model selection out of the equation is NEVER going to account for more than a small bit of any number of OTHER factors - driver skill, engineering talent, and a lot of other aspects of making a car go fast. It doesn't matter that suspension design accounts for an additional TINY little fraction of the total - it is insignificant compared to power/weight ratio, which is in turn insignificant compared to driver skill. I would love to do a study to test the hyphothesis that the single best predictor of an individual car's competitiveness is the amount of money spent on development and tires...
The magic of a weight leveling system based on physical attributes of cars would be in its CULTURAL and SOCIAL power, not in its ability to actually change what podiums might look like. Talented drivers with lots of money would still beat the rest of us and there would still be cars that were marginally better choices than others in the same class.
However, there wouldn't be any room for accusations of impropriety, everyone would know what a car's race weight would be before it was classified, and there wouldn't be any incentive to politic. All good things.
K
In the fall of 2001, members of this board were kicking around conversations about power-to-weight ratios of various cars and Bill M. suggested something that really clicked for me - the idea of using different multipliers to baseline weights for cars in each IT class. I started an Excel doc (an evolution of which I still have) that allowed me to start with estimates of OE- and IT-spec horsepower and weights.
I used some existing estimates to ballpark multipliers - the "Miller Ratio" number - that could then be factored onto power figures to (a) determine if cars were comparably classified, and (
ultimately treat weight as a dependent (or outcome) variable. This thinking was applied to a growing dataset of cars and, back-of-the-envelope, proved to be a pretty good predictor of real-world competitiveness among existing cars in the classes where they were actually running: That is to say, a car that fell out "light" of the average MR number for a class (better power/wt ratio) tended to be more successful in the results. Conversely, cars that fell out "heavy" tended to be dogs and the farther out they were, the doggier...Problem was that there was a big gap emerging between A and S. Understand that my initial effort at promoting change was to apply a formulaic classification scheme to C, B, A, and S. History will tell you that this idea was roundly shouted down, supporting my belief - and that of a lot of other researchers - that people just don't understand statistics, particularly if the numbers are contrary to what they want to hear.
Given the lack of enthusiasm - ahem - demontrated for a major overhaul, I started toying with the idea of creating a new class to address some of the commonly-complained-about problems at the top of A, and both ends of S: IT2 came to life but that's not the topic of today's lesson. A later evolution of the formula applied a second multiplier to engine type, to accommodate the fact that not all engines gain the same kind of power going from stock to IT trim.
A formula could indeed be applied to the existing four classes but it would result in some really nasty compromises - hundreds of pounds of ballast in some cars, others left with no chance of meeting the minimum required to meet their class index, and lots of movement from class to class. The current MR numbers for the five-class structure are...
ITC = .048
ITB = .058
ITA = .0632
IT2 = .069
ITS = .082
The race weight formula is...
(Stock HP x engine multiplier / MR number) + 200
You will notice first that the MRs are NOT linear. They were devised to impose the least amount of weight change on existing cars as was practical. Even at that, some cars just can't stay where they are if these "best fit" numbers are applied, since they require too much ballast to be safe or can't make the required minimum. The breaks between class MR numbers are also not consistent because auto makers tend to build cars in "classes." The numbers had to accommodate cars that were already classified, remember. I couldn't just pick cars that were alike and leave others out in the cold.
The short answer Darin is that it started there - with 4 classes - but practicalities made it unworkable, even if one DID agree with the first principles of the thing. And nobody here did.
That said, your point (Darin) about making progressively tighter estimations - analogous to the foundation of calculus - is an excellent one. Each successive estimate (all of which can be seen as using the same kind of energy, time, or money) gets you closer but LESS closer each time. This kind of thinking is at the center of my long-running arguments FOR formulas. The key is coming up with one that takes the most variance out of the "competitiveness" equation as is practical, making the first estimate that gets you the most bang for your buck. We actually get TWO cuts using the current formula - engine and class multipliers.
The good news is that taking ALL of the variance due to make/model selection out of the equation is NEVER going to account for more than a small bit of any number of OTHER factors - driver skill, engineering talent, and a lot of other aspects of making a car go fast. It doesn't matter that suspension design accounts for an additional TINY little fraction of the total - it is insignificant compared to power/weight ratio, which is in turn insignificant compared to driver skill. I would love to do a study to test the hyphothesis that the single best predictor of an individual car's competitiveness is the amount of money spent on development and tires...
The magic of a weight leveling system based on physical attributes of cars would be in its CULTURAL and SOCIAL power, not in its ability to actually change what podiums might look like. Talented drivers with lots of money would still beat the rest of us and there would still be cars that were marginally better choices than others in the same class.
However, there wouldn't be any room for accusations of impropriety, everyone would know what a car's race weight would be before it was classified, and there wouldn't be any incentive to politic. All good things.
K
