I don’t see why no one has done a Monte Carlo Simulation for the comparison of disc wheels versus other wheels such as tri-spoke using course profile data. It would be instructive.
I am sure there are some programming gurus who could do this. It wouldn’t seem that hard and would let people determine the optimal wheel selection based on course.
On what course would one be debating between a rear disc and rear tri-spoke?
monte carlo for this situation does not work well for the same reason you cannot program a chess computer for every permutation. you can take a wheel in a vacuum and say this one is more aero than another. but when you put it on a course with an individual rider and you take into account all his/her variables and the course variables, you simply cannot generate a reliable outcome as you suggest using a simplified quant simulator like mc. notably you lack the clear maximum across subject that you might find in a trading analysis (max profits).
ultimately what is optimal for one may not be optimal for another. take eagleman for example – flat and windy. any simulator is going to tell you the fastest wheelset is going to be a disc. so why isn’t everyone running a disc? well you can throw out a lot of reasons, but a notable one might be that you are 115 lb female. i don’t care what a simulator says, but a disc is not the optimal if this is you. now consider you are a 160 lb guy who can run a 1:25 half off the bike – the run is your weapon. are you going to run a disc? how you program for that? you can’t.
I disagree. Obviously there are many factors that go in to a race. However, a Monte Carlo simulation would be given input data that would be used to run the MC sim. For example, data such as frontal area, weight of bike, as well as CP60 or other wattage data could be used to determine what the best wheel selection is. Moreover, this is precisely analgous to trading equities, rather than maximizing profits one is minimizing time. In fact, the variables on a rider are largely known when compared to the effects of exogenous forces on equity prices.
Furthermore, the simulation would and could account for various race strategies (in this case one is finding a maximin soultion). Maximum best time on bike while using the least amount of watts.
In fact, I think some sort of iterative method could be constructed to first determine a rider’s best power strategy for a given course and then use that information to run the MC to determine wheel selection. Obviously it is an approximation. But that are approximations of an individual rider’s
Monte Carlo is not an appropriate analytic technique when one can gather actual data. Monte Carlo is for when you can neither model nor field-test a proposition. You have no data, so you have to randomize your inputs.
One can make these wheel decisions with great precision using a combination of data gathering and modelling.
Moreover, this is precisely analgous to trading equities, rather than maximizing profits one is minimizing time.
It doesn’t work for trading equities, either. No need to invent data when the real data is right there before us. Correlation, standard deviations and expected returns are wildly unstable. Markowitz optimization, Monte Carlo – it’s all bunk. (Markowitz himself told me this in person – but he chuckled and said he has no intention of giving back his Nobel.)
if you can’t write the program, i’m not sure what basis you have to disagree.
unfortunately, you cannot program for individual preferences, where people value things differently. this is why deep blue does not win 100% of the time. theory and practice are two different things. you believe you should be able to do it, but you can’t, not with reliable results. that’s why no one has done it and published the results.
of course you can program individual preferences…
so why can’t you program a perfect game of chess?
its an indeterministic system, that’s why. see the works of aron katsenelinboigen, the language of predisposition.
Of course you can program a perfect game chess. You are misunderstanding basic concepts and it has nothing to do with being able to program individual preferences.
Btw, before you tell me I am wrong, this was the topic of my PhD thesis, making the connections between individual preferences in different paragidms and game theory.
You can program a game chess. The problem is that by the time the program is done, humanity will be done too. It sure IS NOT non-deterministic. Chess is untractable which is very different. It just means that there is no deterministic polynomial solution to a game of chess.
I invite you to sit in my class on the topic one of these days. 
there: a deterministic approach to evaluate individual preferences in multicriteria decision making…
www.cs.utep.edu/fmodave/ipmu98.pdf
if you want something in the framework of decision under uncertainty or decision under risk, or something more efficient from a computational point of view, there are other papers on the same page. Most of them are in english.
Does this mean you know how to change your AC filters now too?
yeah yeah…come to PCB I owe you a beer
.
I just might do that. The timing might be right. I am thinking of going from there to Biloxi to find some dirt to build on. Rumor has it they might need some reconstruction work.
i actually don’t think i am mis-understanding the basic concept. what i am saying is that as a result of individual preferences, the math becomes too cumbersome to program efficiently. too many permutations. chess is rather simple and defined system. if we can mathamatize it, how can we do the same thing with much more complex system. the answer is we can’t, or at least until artificial intelligence bridges the gap.
the application to this situation is that there are too many individual variables, that are not necessarily logic driven, to program for to make a high quality predictive equation. your better off in this scenario riding a course 5 times with different set-ups and simply making your decision. if there was science too it, we would see all the top pros with the same wheel set-up for a given course, but we don’t. yet they likely all have the same equipment available to them. that to me says individual preference out games math in making a wheel selection.
i’ll take a pass on sitting in your class. i’ve sat in far too many phd economics course than i would like to remember.
I actually confirm you are misunderstanding the basic concepts…
The maths are not ‘too cumbersome’ to program…I did it…for the army for instance.
you can do it with more complex systems…you can do it for biological systems which is as complicated as it gets. The problem then is tractability.
AI has NOTHING to do with this btw…
There is a science to it. The problem is always the tradeoff between accuracy and complexity. The more accurate the answer, the more complex it becomes to give a solution in a reasonable amount of time.
the answer, gentlemen, is quantum computing. why not be able to evaluate all possible outcomes with a wide range of variables? shouldn’t be long. I think I’ll build a quantum computer this afternoon.
interesting answer…indeed…quantum computers, DNA computing, non-deterministic computers are all implementing the same idea of massively parallel computers with no bounds nearly…
actually i think i do get it, my conclusion is the same as yours – there is a tradeoff between accuracy and complexity. since the level of effort and complexity in this situation would in fact be cumbersome, you are better off using actual data. what makes it cumbersome is individual variances. if there was no preference, the math would be pretty straight forward.
I don’t think you get the point of a monte carlo simulation. Furthermore, my inability to actual implement the programming does not undermine my ability to conceptualize the theory and understand the value of a monte carlo simulation with regard to wheel choice selection in an ironman distance event.
It is precisely because of the difficulty in solving the problem purely analytically that the wheel choice selection could be better approximated using a Monte Carlo simulation. Additionally, I think that it could be used to optimize an individual’s power output for an ironman distance race. Constant power is not necessarily the most efficient power strategy for racing the bike leg.
The testing that Ashburn has documented elsewhere regarding rolling resistance on 1 mile stretches and other testing is 1. time consuming and 2. not readily reproducible and 3. expensive (need a powermeter), 4. not necessarily applicable to conditions found on ironman courses.
From Mathworld:
Any method which solves a problem by generating suitable random numbers and observing that fraction of the numbers obeying some property or properties. The method is useful for obtaining numerical solutions to problems which are too complicated to solve analytically. It was named by S. Ulam Eric Weisstein’s World of Biography, who in 1946 became the first mathematician to dignify this approach with a name, in honor of a relative having a propensity to gamble (Hoffman 1998, p. 239). Nicolas Metropolis Eric Weisstein’s World of Biography also made important contributions to the development of such methods.