Tuesday, May 15, 2012

A Philosopher Examines Crovelli's View on Mises and Probability

A philosopher emails:

Bob,

In view of your battle with Crovelli, you might find my take of interest:

His arguments don't make any sense. He says that R.von Mises defines probability as relative frequency. In doing so, he rules out any other sense of probability by definition. But of course von Mises doesn't do this.He is trying to give a correct theory of probability. He isn't saying that the classical theory is ruled out by definition. Rather, the assumptions that this theory makes are wrong.

It's also silly to think that you need to have the correct theoretical account of probability in order to perform probability calculations. This invalidates his point that insurance can't depend on the relative frequency theory, because insurance was around long before that theory was formulated. Of course it was; but so what?

He is also wrong that Mises ignores the possibility of subjective probability estimates. That's precisely the point of Mises's distinction between class and case probability.

46 comments:

  1. Professor Hoppe sent me the same comments, Wenzel. A few observations about what he has written:

    1) His (and R. von Mises's) claim that the assumptions underlying the classical method are mistaken is beside the point. Of course the assumptions are unrealistic to a certain extent. But this is also true of virtually all abstractions in applied mathematics, INCLUDING Richard von Mises's idea that probabilities converge on the limit. Also, Richard von Mises's claim that the classicicists' assumptions were wrong was based on his view that probability must be a positivist science like any other. Why Hoppe can agree with this type of epistemological arguement, I do not know.

    2) To think that people were "preforming correct probability calculations" before the development of the relative frequency method is just bizarre. There is no evidence whatsoever that I am aware of that anyone prior to John Venn was using this sort of method. How could anyone before Venn have performed correct probability calculations" without the very method that is supposedly correct? Is Hoppe admitting here that it is possible to assign numerical probabilities (contra the Mises brothers) by methods besides the relative frequency method? I think he is.

    3) I'm not sure if Hoppe is speaking here of R. von Mises or L. von Mises. If he is speaking about Ludwig, I tend to agree with this statement, as I argued in my 2010 paper. However, this statement conceals the fact that both brothers denied that NUMERICAL PROBABILITY could be applied to singular events. This is why the application of NUMERICAL probability to sports games, weather forecasting, the search for resources like oil and gold in singular locations, et cetera ad nauseaum matters so much. Hoppe would have us condemn all of these applications outright, solely on the belief that Richard von Mises's favorite method is the "only" correct one.

    ReplyDelete
    Replies
    1. 1. If you agree that the assumptions underlying the classical method are wrong, then you must accept that you were wrong to claim that casinos set odds not according to any prediction of the outcome, but according to balancing the money flows on both sides of the bet.

      2. As L. Von Mises showed in economics, this is very common in other fields of inquiry as well. It is not bizarre that people engage in practically useful mental tools of action, before such mental tools are formalized and expounded by academic professionals. This happens in finance all the time as well.

      3. Numerical probabilities to singular events would be thrown out not because it contradicts R. Von Mises' preferred method, but because the underlying assumptions are contrary to reality.

      Delete
    2. Typo:

      first paragraph should read:

      1. If you agree that the assumptions underlying the classical method are wrong, then you must accept that you were wrong to claim that casinos set odds according to any prediction of the outcome, because they actually set odds according to balancing the money flows on both sides of the bet.

      Delete
  2. I should note as well that Hoppe is plain just wrong to say that Richard von Mises does not define the classical method (and every other method) out of existence.

    To define probability ITSELF as “The probability of an attribute…within a collective IS the limiting value of the relative frequency with which this attribute recurs in the indefinitely prolonged sequence of observations.” is to define away every other method that does not involve frequencies.

    If the classicists had said something similar (e.g., "the probability of an attribute IS the number of times that the attribute can emerge divided by the number of possible outcomes"), they would have defined every other method out of existence. Keynes did something very similar. He simply defined probability as a "logical relation" and excluded everything else simply by definition.

    These types of method-centric definitions are all question begging.

    ReplyDelete
    Replies
    1. "To define probability ITSELF as “The probability of an attribute…within a collective IS the limiting value of the relative frequency with which this attribute recurs in the indefinitely prolonged sequence of observations.” is to define away every other method that does not involve frequencies."

      This isn't a definition. This is (a portion of) his theory of probability.

      http://www.stat.berkeley.edu/~aldous/Real-World/von_mises.html

      You do know the difference between a theory and a definition, right?

      Your criticism that R. Von Mises "defines away" all other possible methods, is silly. That criticism can be made against ANY theory, by interpreting a theory as mere verbal play, and then positing some other verbal play, and then transposing it back to theory before anyone notices. Go on, try it. Take any theory you want, such as gravity, and then say "Einstein merely defined away any other possible force that could explain our observations." No, Einstein was just trying to give a correct theory of gravity. He wasn't defining away all other definitions, he was rejecting all other theoretical explanations as inherently wrong, to make room for his theory.

      You're a very sloppy thinker.

      Delete
  3. Let me bring you up to speed here, Pete, since you clearly do not understand the issues that are being debated. You seem to think that this debate revolves around constructing a "theory" of probability, but it is really nothing more than a definitional debate.

    The papers that I have written offer an argument that probability must be defined subjectively. Why must probability be defined subjectively? Because every event that happens in the world has a cause or causes that necessitate a certain outcome. There is no uncertainty "in" the world if everything that happens has a cause of some sort. The uncertainty is "in" us. Hence, probability is a measure of OUR uncertainty about what is going to happen in the world, not a measure of a "real" property in the world.

    If you are an indeterminist like Richard von Mises, you do not have to accept this argument. If you are a determinist, however, only in the weak sense that everything that happens has a cause of some sort, then you must define probability subjectively. The great probabilist I.J. Good made this observation.

    The upshot of this is that, while it was not contradictory for indeterminists and positivists like Richard von Mises to define probability as a "real" physical property in the world (e.g., a relative frequency), it is contradictory for causal determinists like me or Hoppe or Ludwig von Mises to define probability this way.

    The entire debate revolves around what probability "is." Hoppe sides with the indeterminist/positivist Richard von Mises, while I would side with men like Einstein and Ludwig von Mises in saying that everything that happens has a cause. "God does not play dice." This means that I must also side with men like Nassim Taleb and Bruno de Finetti in saying that "probability does not exist."

    The fact that this debate involves the definition of probability explains why it is question begging to drag particular methods into the picture. You cannot say that any particular method is legitimate or useful until you know what the hell probability is in the first place. To drag particular methods into the picture before you have established what probability is is to beg the question.

    The fact that Richard von Mises and Hoppe do not like the assumptions underlying the classical method is beside the point in a definitional debate. We are trying to figure out what probability "is," not trying to figure out the best method to measure it. It is also blatantly hypocritical for them to ignore the obvious problems with the assumptions underlying Richard von Mises's method. Quite obviously, we can never, ever obtain the limiting values of relative frequencies, and it is dubious at best to believe that these frequencies converge on the limit (The Black Swan shows us the fallacy in this idea).

    So, if you really feel like joining this debate, Pete, (unlike Wenzel), you need to approach this from a definitional perspective.

    The central question is "What is probability?" Is it a "real" and "objective" thing, as Hoppe would lead us to believe, or is it subjective, as I argue?

    ReplyDelete
    Replies
    1. Let me bring you up to speed here, Pete, since you clearly do not understand the issues that are being debated.

      You're in no position to be condescending towards anyone in this debate. You don't even have the guts to admit that you were wrong in your 2009 paper where you laughably suggested that casinos set the odds according to predicting what they perceive as the likely outcomes, instead of the truth, which is setting a line according to cash flows to balance the sides. You took the notion of subjectivity in probability way too far and you got burned for it. And you've just been digging a deeper hole ever since.

      You seem to think that this debate revolves around constructing a "theory" of probability, but it is really nothing more than a definitional debate.

      No, that's false. The point Wenzel is making, the point Hoppe is making, and the point that is going over your head, is that YOU are the one trying to spoil these points as mere definitional quibbling, when in reality they are arguments concerning theory.

      A single event that is attributed as having a probability of 80%...what does that mean exactly? R. Von Mises argued that it is incoherent without reference to some grounding of what the 80% refers to. He argued it can only make sense if it is grounded in repetition of a class of events. E.g. 80 successes out of 100 repeated events.

      This is what he was doing when he tried to come up with a theory of probability. What you're doing is corrupting this down to mere verbal games, as if R. Von Mises was merely defining away any other method. That isn't what he was doing, and your attempts at trying to re-interpret what he was doing fails. It contradicts what he wrote!

      There is no uncertainty "in" the world if everything that happens has a cause of some sort. The uncertainty is "in" us. Hence, probability is a measure of OUR uncertainty about what is going to happen in the world, not a measure of a "real" property in the world.

      As quantum physicists have shown, there is in fact uncertainty "in" the real world. You have it backwards. The uncertainty is in the real world, and the causality inference is "in" us. Human actors infer causality in the real world because we ourselves act to change the world to achieve subjectively desired ends. Causality cannot be observed or sensed. Causality is presupposed in acting.

      Delete
    2. 2/2

      The central question is "What is probability?" Is it a "real" and "objective" thing, as Hoppe would lead us to believe, or is it subjective, as I argue?

      No, the central point is that you were wrong to claim in 2009 that casinos make money according to making predictions of outcomes, and you continue to refuse to admit it. The only time you came within a hair's breath of actually addressing your error was when you said this:

      "So what? Am I right or not? Do casinos predict the outcome of games in order to set the initial odds or not?"

      http://www.economicpolicyjournal.com/2012/05/crushing-crovelli.html?showComment=1336997463522#c8574161116651813719

      You never said INITIAL odds in your 2009 paper.

      You have no class. You are refusing to admit you goofed.

      The point you are missing is that sports books DON'T NEED to make any predictions concerning the outcome. That is not how they make money. They could set the initial line at 50/50 and almost immediately start moving the line as the bets come in. Yes, they COULD make a particular prediction to get things moving, but this is not necessary. It's akin to an auctioneer setting an original price and then letting the buyers determine a final price as the offers come in. The auctioneer could set the initial price at $0.01 if he wanted and it wouldn't matter. He just sets an higher initial price to save time and get things moving faster. Neither the auctioneer nor the buyers are obligated in any way to the initial price, nor is it necessary for anyone to make a prediction for what the final price will be.

      Delete
    3. Pete, you clearly did not read what I wrote about determinism. What I said is that we are required to define probability subjectively IF we are determinists. I am a causal determinist, and so was Ludwig von Mises, so we must define probability subjectively. A deterministic world is a world where there is no uncertainty "in" it, so probability cannot be a measure of something "out there" in the world.

      If you believe that the world is indeterministic at the atomic level, like Richard von Mises, then you are free to define probability any way you want. (Although you must assume the world is causally deterministic in order to use the relative frequency method, but that's a whole other topic).

      Delete
  4. I'm not really a part of this debate, but I'd like to ask - why is this question binary?

    If I'm shooting particles through a slit, there's a precise probability that they will hit the screen at a certain location. This probability is a physical quantity and it's not subjective at all; it is there, and it is fixed, regardless of the observer.

    If I am trying to predict the next election however, I can't possibly have objective data to quantify my prediction. Moreover, my own prediction affects the outcome itself! Meaning that, in this case, the subjective component is more obvious.

    So it seems probability has both an objective component (due to physical qualities) and a subjective component (due to imperfect knowledge and reflexive behavior). Everything has a cause, but some causes have more than one possible effect.

    Definitional wars are rarely productive...

    ReplyDelete
    Replies
    1. You ask a very good question, Alex. I think this is the first comment that truly addresses a substantive issue.

      I think it is preferable to use a less complicated example, because this example brings into the picture a whole host of extremely complicated epistemological questions. (This type of experiment, for example, brings us back to the problem proposed by Gene Callahan a few weeks ago. Every point on this continuous screen would have a probability of 0).

      Let's look at a simpler example: tossing a coin. We want to know the probability of tossing a head with the coin. How would we go about figuring out what the probability is? Well, the answer we come up with depends upon the definition we adopt for probability.

      There are of course a number of different methods we could use to generate a number between 0 and 1 (which is all that is needed in order to apply Kolmogorov's axioms for probability). We could toss the coin a thousand times and generate a relative frequency. We could simply assume that the coin is fair and thus that each side is equally likely. Or, we could use indirect evidence and develop a model that will predict the outcome as best as is possible.

      In the real world we are usually restricted to using just one of these methods. In football, for example, the teams are not in a position to generate a relative frequency prior to the game to ensure that the coin for the coin toss is fair. Both teams just assume that it is a fair coin and that each side is equally likely to be tossed. In other areas, like life insurance, we are in a position to generate relative frequencies, so of course we use that method. In other areas, like the generation of sports odds, neither one of these methods will work, and so we have to model the outcome as best as we can with the indirect information that we have.

      Whether any of these methods are legitimate for generating numerical probabilities depends upon the definition for probability that we adopt. We need to know what probability is in order to know whether any particular method is capable of legitimately generating numerical probabilities.

      I have followed I.J. Good in arguing that the definition of probability depends upon the nature of the world. If the world is deterministic (as Ludwig von Mises and Einstein claimed), then probability cannot be a "real" property in the world. Every event in the world would be the necessary result of the causal factors that produce it. "No deviation
      from the necessary course of affairs is possible. Eternal law regulates everything," as Ludwig von Mises puts it. This means that probability must be measuring something "inside" us, not something "out there" in the world.

      Does this clarify anything?

      Delete
    2. The Great ObserverMay 16, 2012 at 10:02 AM

      Crovelli,

      More proof of your confused mind.

      You write:

      "In the real world we are usually restricted to using just one of these methods. In football, for example, the teams are not in a position to generate a relative frequency prior to the game to ensure that the coin for the coin toss is fair. Both teams just assume that it is a fair coin and that each side is equally likely to be tossed."

      No they don't have to assume the coin is fair. All they have to consider is that the other side does not know how the coin is skewed (if it is). This gives equal probability on the coin toss to both teams.

      Even if the coin is so skewed that it lands 100% of the time on heads, if neither team knows this, the odds remain 50/50 on the outcome for either team.

      Delete
    3. First of all, Observer, if you were a frequentist like Hoppe, you would have to say that there can be no numerical probability put on this event UNTIL there is a frequency that has been calculated. So, for you to say that there is a probability of .5 (50/50 odds) for a coin that has not yet been tossed, you are saying that Hoppe and the frequentists are wrong.

      Thank you for taking my side!

      Second, a "fair" coin IS a coin that does not have any inherent bias, i.e., it is not "skewed.". Both teams ASSUME that this is the case because neither side knows whether it is or not. Duh.

      Delete
    4. The Great ObserverMay 16, 2012 at 11:32 AM

      Crovelli

      It is simply not the case that both teams have to assume that the coin is "fair". If neither team knows how it is skewed, they both have a 50/50 chance, before one team calls "heads or tails".

      Let me make it simple for you. If the coin is so skewed that it will always turn up heads, but neither team knows this, the the team that calls the toss will still have a 50/50 chance of winning, depending what they call. No assumption of fairness is necessary.

      Quit being so stubborn about admitting all your errors. You are creating a frequentists dream of developing a track record of causing a new error 100% of the time, every time you try and defend an old error.

      Delete
    5. Ha! I notice that you did not touch on the most important point of all. You continue to say that it is possible to assign a numerical probability to a singular event WITHOUT actually generating a frequency. This means that you agree with me and not with Hoppe! Thank you again, Observer! Finally an ally on this site!

      As far as the coin example is concerned, you are completely missing the point. I said that both teams ASSUME the coin is fair because it is not practical to "test" it before the toss. This is just another way of saying that they DO NOT KNOW whether it is indeed fair. You are saying exactly the same thing. Moreover, your focus on the teams' beliefs about the coin's fairness instead of the frequency is exactly how a good subjectivist should think. Bravo!

      You and I are saying EXACTLY the same thing! Glad to have you on board!

      Delete
    6. The Great ObserverMay 16, 2012 at 11:58 AM

      This is absurd. Don't you have a day job? Your continual assertions of victory by sliding off point are serious signs of delusion.

      You assert, "You continue to say that it is possible to assign a numerical probability to a singular event WITHOUT actually generating a frequency." I said no such thing. And the point is not what two teams assume, a team could assume they have a 75%, but the outcome of winning the toss is 50/50 regardless of their assumptions.

      I am not on board with you, you argue in a dishonest fashion, as Banacek has pointed out.

      Delete
    7. No, it is not 50/50 if you are frequentist like Hoppe or Richard von Mises. They say that putting a numerical probability on such an event is ABSURD AND MEANINGLESS until you actually have a frequency. You are saying the probability is 50/50 a priori, which is EXACTLY what Richard von Mises says is impossible.

      Hence, you are indeed on board with me against Hoppe. Thanks again.

      Delete
    8. The Great ObserverMay 16, 2012 at 12:13 PM

      You are creating a false simple model. The model is much more complex than you are creating. We are not talking about the outcome of the flip of the coin itself. We are talking about the tautology that with two possible outcomes and with two parties being impacted by the outcome of the call of the coin flip they both have a 50/50 chance of the outcome coming out in their favor before they call the toss and if they don't have insider knowledge about the coin which could impact the outcome. This is a given. It is not about the coin itself, which would require frequency tests to determine if it was fair.

      Delete
    9. No, this is completely false according to frequentists. They EXPLICITLY say that there is no probability of 50/50 UNTIL the coin is actually tossed over and over. You are using the classical approach that they tried to get rid of.

      For you to say that there is a 50/50 chance BEFORE tossing the coin, you are directly contradicting the frequentist position.

      Delete
    10. Mark, thanks for answering. I see two problems:

      1. I think you're brushing away the physics example because it's inconvenient, but it doesn't make it less true - in modern physics, the world is NOT deterministic. So at the very least, you are forced to concede that SOME probabilities (e.g. the ones in Shroedinger's equation) are physical qualities and not subjective evaluations, and SOME causes have more than one effect.

      2. The two approaches as you define them are both too pure to be useful in the real world. If one refused to assume the cointoss is roughly 50:50 until he actually measured it on a large sample, he'd be a frequentist fool; and if one assumed it was 50:50 even after seeing it land on heads 1000 times in a row he'd be a determinist fool...

      In reality, to make good predictions you need to have both - an initial prediction based on your (subjective) understanding of physical properties, and an ongoing adjustment based on (objective) measurements.

      That would also answer the bookmaking question - you start at your best prediction based on the data and then you adjust as the money flows. Starting at 1:1 is foolish, but failing to adjust is also foolish.

      ---
      If a definition is really necessary, I'd suggest something like: the actual probability is the frequency in an infinite sample; the measured probability is the frequency in an actual sample; the expected/perceived probability is what we can predict based on (our understanding of) physical properties and measured probability.

      P.S. The difference between actual probability and expected probability is imperfect knowledge, and the fact that it exists already refutes most classic economists... ;-)

      Delete
    11. I am not ignoring the physics, Alex. I say explicitly that IF you are an indeterminist like Richard von Mises, then you should feel free to define probability in any way you wish. If you think that things actually occur at the subatomic level randomly and without cause, then there is no problem for you to define probability "objectively."

      For those of us who are causal determinists, however, like Ludwig von Mises, Einstein, and myself, we are not entitled to define probability as anything other than a subjective measure of human uncertainty. For us, there is no such thing as randomness or uncertainty in the world, even at the subatomic level. We would concede that there are problems of measurement at the subatomic level, but this by no means implies that what is going on is truly "random" or spontaneous. "Eternal law governs everything," as L. von Mises says.

      There may be no reconciling our views about determinism here, but allow me to make one observation in defense of determinism. It is self-contradictory and absurd (as Hoppe and Popper have shown) to use scientific instruments that could only "work" in a deterministic world in order to try to prove that the world is indeterministic.

      Delete
    12. "like Ludwig von Mises, Einstein, and myself"

      Crovelli your ego is immense, only superseded by you confusion.

      Delete
    13. Please explain to me, big guy, how pointing out that we are all determinists is egotistical. The three of us ARE determinists!

      Delete
    14. Mark, it's true that Einstein has never accepted quantum theory, but we have almost 90 years of scientific advances to make the decision much easier for us. Brilliant people can be wrong too.

      Specifically, Einstein's last attempt at an alternative explanation (local realism) was refuted by the aspect experiment 30 years ago. I think we have passed the point where the other side of this argument was valid a long time ago - uncertainty is not just "problems of measurement" as you called it; it is a property of reality itself.

      Could it be that belief in determinism (over empirical measurements) also creates a tendency to hold on to predetermined beliefs, even when they are contradicted by scientific data?

      Delete
    15. There are a couple of issues here, Alex. First, , it is not possible to prove that particles at the micro level have no causal reason for behaving the way that they do. Heisenberg's own uncertainty principle rules out the possibility of investigating these particles beyond a certain point, so it is misleading for you to assume that physicists have "proven" that particles do not have causal reasons for behaving as they do. It is not even theoretically possible to prove the negative assertion that particles have no causal reason for behaving as they do.

      It is not as though physicists have really given this much thought since Einstein. Even Niels Bohr and Heisenberg stopped looking at this question in the 30's. I remember seeing a quote from a physicist today who said something to the effect that modern physicists don't even bother thinking about this problem because they assume it was settled, and that if they could only spare 20 minutes one afternoon they could figure out why it was settled.

      Finally, you have to remember that we are dealing with a scientific theory that relies upon empirical evidence. Thus, even if the evidence suggested that there is no deeper cause for the behavior of subatomic particles, this by no means "proves" the theory to be true. New evidence can emerge that contradicts the theory. "Paradigm shifts" are still possible (to use the phrase of Thomas Kuhn, who was a physicist in addition to being a philosopher of science).

      Delete
  5. Oh, Pete, you want to rehash the odds argument that even Wenzel himself has abandoned. That's just fine with me. I thought you were moving on to something more substantive, but let's just flay that dead horse some more.

    Sports books do indeed predict the outcome of games. There is no other way that a casino could predict how millions of dollars in bets are going to come in unless they predict the outcome of the game as part of their model. This is simply a fact.

    The fact that casinos do not open their lines at 50/50 on every game should alert you to this fact. The fact that lines typically don't change very much in the lead up to the event should also alert you to the fact that casinos are extremely adept at predicting the line that is needed to balance the books right off the bat.

    opening odds are NOT simply arbitrary numbers. Any casino that arbitrarily set its initial odds would get absolutely killed by professional betters. Professional betters would just wait for a line to open and pile in on the side that was undervalued or overvalued by the casino's arbitrary odds. Classic arbitrage. The professional betters wouldn't even bother betting after the line changed and became more reasonable, because they could make a killing just betting on the days these ridiculous odds opened. The casino would get obliterated every time.

    You, like Wenzel, are interpreting the statement "casinos COULD grope along without models predicting the game's outcome" to mean that casinos DO IN FACT operate this way. This is a silly non sequitur.

    The fact that casinos can consistently generate odds that make them money and cannot be exploited by professional gamblers means that the numbers they generate are pretty damn accurate. That is PRECISELY what I wrote in my paper, and it is 100% correct.

    ReplyDelete
    Replies
    1. Crovelli,

      You are an amazingly dishonest. Wenzel didn't abandon any of his positions. After I poinedt out that the quote you made was made up by you and can't be attributed to Wenzel, you say you were just mocking Wenzel and then you put the same line in quotes again. Let me spell it out DISHONEST.

      You should work for the government.

      Delete
    2. Oh, Pete, you want to rehash the odds argument that even Wenzel himself has abandoned. That's just fine with me. I thought you were moving on to something more substantive, but let's just flay that dead horse some more.

      At this point it is clear you lost.

      Wenzel didn't "abandon" any of his arguments. He is just addressing multiple errors you are making, and you are incorrectly interpreting that as him abandoning the original argument he made regarding the errors of your 2009 paper.

      Sports books do indeed predict the outcome of games. There is no other way that a casino could predict how millions of dollars in bets are going to come in unless they predict the outcome of the game as part of their model. This is simply a fact.

      No, that is false. You can repeat this fallacy as many times as you want, but it won't make it true. Sports books make the line based on cash flows. That is how they make money. Their job is not to predict outcomes. They don't have to predict any outcome for them to act as middlemen.

      The fact that casinos do not open their lines at 50/50 on every game should alert you to this fact.

      You're missing the point. Casinos don't have to open their line at ANY particular value. They don't make money by making predictions. They make money by balancing the books. You even admitted this after Wenzel corrected you on this point.

      The fact that lines typically don't change very much in the lead up to the event should also alert you to the fact that casinos are extremely adept at predicting the line that is needed to balance the books right off the bat.

      Hahahaha, so now you're saying casinos predict a line that would balance the books. That is a far cry from predicting the outcome of games.

      opening odds are NOT simply arbitrary numbers. Any casino that arbitrarily set its initial odds would get absolutely killed by professional betters. Professional betters would just wait for a line to open and pile in on the side that was undervalued or overvalued by the casino's arbitrary odds. Classic arbitrage. The professional betters wouldn't even bother betting after the line changed and became more reasonable, because they could make a killing just betting on the days these ridiculous odds opened. The casino would get obliterated every time.

      Nobody claimed the initial line MUST be "arbitrary." You're twisting things so badly now that it is clear you're not intellectually honest.

      You, like Wenzel, are interpreting the statement "casinos COULD grope along without models predicting the game's outcome" to mean that casinos DO IN FACT operate this way. This is a silly non sequitur.

      No, that's false. We are just going by your incorrect assertion regarding how casinos set the line. They don't care what the outcome is. The only thing that matters to them is balancing the books, to split the betters in two.

      You admit this in your latest response to me, but you don't even have the cajones to admit you were wrong.

      The fact that casinos can consistently generate odds that make them money and cannot be exploited by professional gamblers means that the numbers they generate are pretty damn accurate. That is PRECISELY what I wrote in my paper, and it is 100% correct.

      The reason why they rarely lose money is NOT because they correctly predict the outcome. No, they rarely lose money because they successfully predict what line would balance the books. Jeepers, are you dense or what?

      You're being completely inconsistent in your claims.

      Delete
    3. Pete, you are as dense as Wenzel. In order for a casino to make money it has to predict how the money will be bet. HOW DOES THE CASINO PREDICT HOW THE MONE WILL COME IN? I truly cannot believe that you and Wenzel do not see this problem. You are so focused on the fact that the line can move that you cannot see that the line has to be generated in the first place in order to move it!

      The way that the casino predicts how the money will come in is to predict the outcome of the game and adjust that number according to various variables in the proprietary model. That is just a fact about how casinos operate.

      If I am wrong here you are either saying that the opening odds casinos generate are either 1) completely arbitrary, or 2) they just fall out of the sky in the "correct" position to balance the books. Either way, this is completely asinine.

      How else could a casino generate opening odds that will balance the books. How could they predict how the bets will come in without knowing anything about the thousands of people who are planning to place bets.

      You tell me. Wenzel certainly hasn't tried to say how this could be done.

      Delete
    4. Pete, you are as dense as Wenzel.

      Is that what you call people who make you look like a fool? OK, we're all "dense."

      HOW DOES THE CASINO PREDICT HOW THE MONE WILL COME IN?

      IT DOESN'T HAVE TO PREDICT HOW MONEY WILL COME IN.

      You are so focused on the fact that the line can move that you cannot see that the line has to be generated in the first place in order to move it!

      The line doesn't have to be based on predictions of outcomes. That's the point that is going over your head. You're so focused on trying to cover your error that you're missing the forest for the trees.

      The way that the casino predicts how the money will come in is to predict the outcome of the game and adjust that number according to various variables in the proprietary model. That is just a fact about how casinos operate.

      False. They do not predict how cash will come in based on predicting outcomes you nerfherder. Those are two different things. You're tacitly presuming that the cash that comes in automatically follows the same prediction as the casino. That's wrong.

      If I am wrong here you are either saying that the opening odds casinos generate are either 1) completely arbitrary, or 2) they just fall out of the sky in the "correct" position to balance the books. Either way, this is completely asinine.

      Hey why not a false dichotomy to add to the mix. No, the choice isn't between predicting outcomes or setting a completely arbitrary opening line. If they do pick an opening line that is contingent, it will be contingent on the expected cash flows, NOT on the predicted outcome.

      If I am a sports game setter, and there is a horse race, and I predict horse A to come in top three, then I won't set the opening line based on that prediction. If I do set an opening line contingent on something, it will be contingent on what I expect the bets will be.

      Are you really so incredibly obtuse that you don't even understand the difference between outcomes and cash flows? You're hopeless.

      How else could a casino generate opening odds that will balance the books.

      Haha, now you're saying they set the line that will balance the books again!

      How could they predict how the bets will come in without knowing anything about the thousands of people who are planning to place bets.

      They do research on what they expect the cash flows will be. They don't set a line based on who they think is going to win.

      You tell me. Wenzel certainly hasn't tried to say how this could be done.

      He doesn't have to. All he did was show you how you were wrong in 2009 when you claimed casinos make money year after year by predicting outcomes. Now you're saying both outcomes and balancing the books.

      You're so confused it hurts.

      It's over. You lost. Be a class act and admit you were wrong so we can go home. You're being a child.

      Delete
    5. "The fact that casinos can consistently generate odds that make them money and cannot be exploited by professional gamblers means that the numbers they generate are pretty damn accurate."

      That isn't accurate at all about professional gamblers not being able to exploit the casinos in Vegas. It demonstrates that you simply do not understand what you are talking about with vegas oddsmakers. They are able to make money because the lines are based upon how people bet, not what they think will happen. You would be shocked at how there is one main person who sets the initial lines -- and is currently in Costa Rica bc it is a non-extradition nation on gambling -- though there are other people who will be the face of those initial lines for boxing and MMA based upon spreadsheets he has about judges history, the fighter's history, etc while also taking into account how they think most people will bet. For instance, a fight with a British fighter like Hatton always involved some ridiculous betting on the fighter that was completely delusional but expected since the fans are british, and that was always taken into account. From this point on, the odds change because of how people bet, not based on who they think will win.

      This fact is exactly why certain people can and do take advantage of the gambling odds set by the casinos all of the time -- which would be much harder to do if they were based around who would win instead of how people bet. It is much easier to predict how people will bet.

      http://www.cbsnews.com/2100-18560_162-7243443.html

      http://www.cbsnews.com/video/watch/?id=7253011n

      Billy Walter has for years and years taken advantage of the thing you claim cannot be done, and that is the betting system set up by the vegas casinos. Watch or read the above links if you don't believe me.

      You really should know what you are talking about before attempting to act so condescending toward others. And yes, I actually know exactly how the gambling lines are set, and my comments above about how responsible one specific individual is for setting most of the initial lines in Vegas who is no longer in America is a pretty good indication for anyone who actually knows the industry.

      Delete
    6. "In order for a casino to make money it has to predict how the money will be bet. HOW DOES THE CASINO PREDICT HOW THE MONE WILL COME IN?"

      Which again shows how clueless you are on gambling. It is very easy to keep spreadsheets and even just remember how whenever a British fighter comes to vegas how much money is bet on him, regardless of his chances of actually winning. Most people don't understand the fight game very well at all, and it is easy to break down how they will gamble based upon nationality, race, and just who pop culture likes when it comes to how the masses will bet when it comes to fighting.

      But it, just like football or basketball as documented by 60 minutes, is very, very easy to take advantage of with the right methods and info. The casinos still make so much money off of the clueless masses that it doesn't matter that a few people devote the time and energy as well as having the intelligence to beat the system.

      Delete
    7. You have got to be kidding me, Anonymous. "Spreadsheets?" If you honestly think that the numbers on this webpage http://scoresandodds.com/ are generated just by keeping "spreadsheets," you are beyond clueless. Needless to say, while people can put numbers into a spreadsheet, a spreadsheet cannot generate the numbers.

      Delete
    8. For the benefit of everyone on here, especially Wenzel, here is a quote that should finally put this matter to rest.

      "In addition to trying to balance bettors on either side of a bet, sportsbooks seek to price the odds of each bet so that each sporting event is close to a "centered game," or a bet whose pricing reflects the actual expected probability of that event to occur. If the bets are priced with the true exact probabilities, bettors will only be able to win 50% of their point-spread bets (and appropriate moneyline winning percentage) -- and the sportsbooks will collect the 4.5% profit margin in the long run due to the cushion of the vig. In this example, the actual bets that a sportsbook takes won't matter in the long-run (because by definition, the proper pricing will prevent bettors from making outsized gains)."

      http://www.sportsinsights.com/betting-tools/sportsbook-profit-margins.aspx

      End of story. I win. Wenzel loses.

      Delete
    9. The Great ObserverMay 16, 2012 at 12:20 PM

      Are you an idiot?

      The balancing money flow number and the attempting to gauge the true exact probabilities number are two different numbers. How can you simultaneously do both? You have to be a total moron to think both these things can be done at the same time.

      Delete
    10. Who said it was done at the same time? I didn't. Modeling anything involves combining different variables into a coherent model. Wenzel says that the outcome does not factor into this model at all.

      He is wrong and I am right.

      Delete
    11. No, you WERE wrong and Wenzel WAS right. You've changed your tune.

      You are still conflating predicting outcomes with predicting cash flows.

      Sportsbooks care about the latter. That's how they make money "year after year."

      Your quote above is not NECESSARY to the sportsbook process. That's why it is said "in addition" to balancing bettors, rather than "another method is" or "this is what is ultimately required."

      Delete
  6. First of all, Banacek, I responded to you over an hour and half ago, but Wenzel's pathetically slow comments section still hasn't posted it yet. I have never seen a more inefficient website (assuming that these constant delays are not intentional).

    You know very well that I was simply mocking Wenzel with that pseudo quote. It was directed at him, not you, and he knows very well that he didn't write that verbatim. I was trying to bait him into writing something substantive instead of making ad himonem comments about my mother.

    As for the claim that Wenzel did not abandon his argument, point to one comment, just one comment, since his original post where he has defended the idea that casinos do not predict the outcomes of games in order to predict the way bets will come in. He has not, and he started to back off the idea as soon as Robert Fallon made his first comment, which means he has (wisely) abandoned the argument. He is free to jump in here and dispute this.

    I hope he does.

    ReplyDelete
    Replies
    1. As for the claim that Wenzel did not abandon his argument, point to one comment, just one comment, since his original post where he has defended the idea that casinos do not predict the outcomes of games in order to predict the way bets will come in.

      He already defended his idea multiple times, you just won't admit that you were wrong. You can't expect people to repeat themselves 54,675 every time you cry foul.

      Educate yourself:

      http://www.free-live-odds.com/pages/odds/odds.php

      "But the goal of oddsmaking for the casino oddsmakers is not to predict the outcome of a game, it is to furnish the bettors with a betting line that will split the public in two with half of the people betting one side and half on the other. This is where “juice” or vigorish comes into play. The “juice” is the 10% that bettors have to pay if they bet football or basketball at 11 to win 10."

      and

      "Oddsmakers work for the house so their goal is to make the sportsbook money by putting out good lines that will draw interest on both sides."

      Do you comprehend how betting works yet? You are so clueless that you keep insisting that oddsmakers set the line based on predicting outcomes. No, they don't do that. If they do set a contingent line, it will be contingent on attracting the most cash from both sides. They don't give a flying rat's ass what the outcome will be. They set a line based on balancing the books and maximizing cash flow from all parties.

      You keep arguing from ignorance by wailing "But but but how else would they set the opening line?" as if the fact that you don't know how the real world of betting works somehow entitles you to make any old claim that suits you.

      Instead of insisting that the world of betting and oddsmaking works the way you want it to work so that your fallacious theory of probability is rescued, you should instead observe reality, and then adjust your theory accordingly.

      You're so wrong it hurts.

      Delete
  7. I would like to refer everyone on this site who thinks that opening odds are arbitrarily set to this website: http://scoresandodds.com/ (Of course, there are thousands of websites like it). Have a look at the opening lines and the current lines.

    Examine the column "Line Movements," and see how very little the line has moved since it opened. If Wenzel and Pete are right that odds are just arbitrarily set or picked out of the sky, it is absolutely miraculous that the line moves so little! How could that be?

    Could it have something to do with the fact that casinos' models of how the money will be bet are amazingly accurate? Of course. How could the casino know how people that they do not even know will bet on a game? The answer: they predict the outcome and then adjust this prediction with various other variables that are included in the proprietary model.

    Case closed.

    ReplyDelete
    Replies
    1. You continue to miss the elephant in the middle of the room. The line moves little because Vegas purposefully sets the line in accordance with expected cash flows. They purposefully set the line to maximize attraction from both sides of the bet! That's why the line moves little after the initial odds.

      The fact that the line moves little HAS NOTHING TO DO WITH WHAT THE ODDSMAKER PREDICTS THE OUTCOME TO BE. It doesn't mean the oddsmaker "guessed the right outcome." It means they guessed the right cash flows.

      For the millionth time, outcomes are not the same as cash flows that balance the books!

      Finally, it is NOT "arbitrary". They don't just pick any old line out of thin air. They estimate the one closest to the one that will attract the most interest from both sides.

      If an oddsmaker had to choose between setting a line based on his prediction of the outcome, or a line based on his prediction of cash flows, he will choose the latter, not the former.

      I would like to refer you to this site:

      http://www.free-live-odds.com/pages/odds/odds.php

      Read the paragraph titled "How are odds made?":

      "But the goal of oddsmaking for the casino oddsmakers is not to predict the outcome of a game, it is to furnish the bettors with a betting line that will split the public in two with half of the people betting one side and half on the other. This is where “juice” or vigorish comes into play. The “juice” is the 10% that bettors have to pay if they bet football or basketball at 11 to win 10."

      Clear as mud now? Or are you going to continue to make a fool of yourself?

      Delete
    2. Pete, you have absolutely no idea what this debate is about. It is not about whether casinos try to balance bets for and against. Of course they do. Who has ever said that casinos don't try to balance the money coming in? I certainly have never said that, so can you please stop punching that old straw man.

      What the debate has been about literally from the beginning is whether casinos predict the outcome of the game in order to predict how bets will come in. Here's what I asked weeks ago when this debate started:

      "The problem is how do casinos generate the INITIAL odds to start taking bets. They can't start taking bets without an INITIAL line, so where does it come from. No one doubts that odds and lines can be adjusted AFTER they start taking bets, but where do the initial odds come from?
      Quite obviously, the initial line is built, in part, on a forecast of who is going to win. Duh."

      Wenzel says that the casino DOES NOT predict the outcome of the game, while I say they do. Here's why I am completely right and he's completely wrong:

      "In addition to trying to balance bettors on either side of a bet, sportsbooks seek to price the odds of each bet so that each sporting event is close to a "centered game," or a bet whose pricing reflects the actual expected probability of that event to occur. If the bets are priced with the true exact probabilities, bettors will only be able to win 50% of their point-spread bets (and appropriate moneyline winning percentage) -- and the sportsbooks will collect the 4.5% profit margin in the long run due to the cushion of the vig. In this example, the actual bets that a sportsbook takes won't matter in the long-run (because by definition, the proper pricing will prevent bettors from making outsized gains)."

      Delete
    3. Give it up Crovelli, you lost a long time ago, and it is clear you refuse to admit you were wrong.

      Yes, this whole debate has in fact been casinos balancing the bets for and against. That's what Wenzel corrected you on, and everything else your throwing into the mix are nothing but red herrings.

      Your question that you asked "weeks ago" contains the crucial word "INITIAL" that was NOWHERE to be found in your 2009 paper. It is only AFTER you were corrected that you started to separate the two.

      You were completely wrong, and you still occasionally make the mistake of claiming oddsmakers' job is to predict the outcome of the game when in reality they do not.

      In the source you got that quote:

      http://www.sportsinsights.com/betting-tools/sportsbook-profit-margins.aspx

      It further states:

      "However, there is one sure thing that rings true: human nature. Bettors have certain tendencies. For instance, on average, bettors like to take favorites. Sports fans also like "jumping on the bandwagon" and riding the coattails of perennial winners. Sportsbooks can use these biases to shade their lines and increase their profit margins."

      THIS is the actual "predictions" sportsbooks make. They ultimately based their predictions on cash flows. Any "INITIAL" line they make that is based on predicting outcomes is something that is NOWHERE to be found in your 2009 paper. You clearly claimed in your paper that casinos make money "year after year" by correctly predicting outcomes. No, they don't. Year after year, they make money by setting the line that balances the books.

      "the goal of oddsmaking for the casino oddsmakers is not to predict the outcome of a game, it is to furnish the bettors with a betting line that will split the public in two with half of the people betting one side and half on the other."

      http://www.free-live-odds.com/pages/odds/odds.php

      Delete
  8. Human actions are not governed by causality but by teleology, by goals and the scarce means to reach those goals. That is why in praxeology one must speak the language of methodological dualism and why it is wrong-headed and meaningless, no matter how much statistical and econometric modeling one concocts, to assign probabilities to singular future human events, especially those that really matter such as the start of WWI or the 1998 Russian financial crisis. Just ask the geniuses at Long Term Capital Management.

    ReplyDelete
    Replies
    1. That's the other extreme and it also goes too far. LTCM didn't blow up because they assigned probabilities; it blew up because they bet the farm on it.

      Numerical prediction is perfectly fine as long as you recognize your fallibility. People assign probabilities to future events all the time: on intrade, in options trading, in poker.. there is nothing wrong-headed or meaningless about it.

      Where I disagree with LTCM is - I don't bet everything I have on a single poker hand, no matter how good I think I am. They did...

      Delete
    2. I could not agree more with you, Alex. There is nothing wrong with assigning numerical probabilities to risk or anything else. What IS wrong is thinking that probabilities are "objectively true." This was one of the most important points made by Nassim Taleb.

      Delete