1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Ryan Tannehill best games compilation

Discussion in 'Miami Dolphins Forum' started by Brasfin, May 15, 2016.

  1. Finster

    Finster Finsterious Finologist

    3,087
    2,038
    113
    Jul 27, 2013
    On random, i would say that just because you can't predict what he will call, doesn't make it random, I get what you're saying, and understand how it would apply to many exercises, but I don't think it's applicable here, simply because there isn't any random playcalling, and random by definition means, without cause.

    As far as stats like YPA, those are gathered to determine what has been, not as a per play predictor, as is the case in this discussion, so aren't really comparable.

    The "bits" is a big part of my point, that there are so many "bits" that it would seem to me that you would need at least a decade of stats to come to any reasonable conclusions, and even then, it would still only be right a certain % of the time, jmo.
     
  2. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    On random, like I said it's by definition true (unpredictable):
    https://en.wikipedia.org/wiki/Randomness

    And there is always neural "noise" in the brain that even makes decisions in your brain "random" from a physical/biological point of view. For example, the neural spike train depends on exactly how many neurotransmitters are released, which can't be predicted either. So just because you interpret an action as purposeful doesn't mean it doesn't have a physical component of randomness to it.

    More importantly, we're talking stats here and if something isn't deterministic given the info you have, it's by definition random from a statistical point of view.

    And this proposed randomness stat is also a "has been" stat. It's not meant to be a per-play indicator. It's mean't to be a per-sequence-of-plays indicator, which in the end is the same as stuff like YPA.

    As far as estimating the sample size needed, if anything you should need fewer samples than YPA or passer rating because it's realistic to assume greater stability in the randomness of a person's decision making across conditions than stability in performance (YPA, passer rating, etc..). I mean who knows without gathering the data, but there's no a priori reason to think sample size needs to be larger here than for common stats.
     
  3. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    If something is predictable, isn't it therefore, not random? Looking at "predictability" would be the same as looking at "random," just maybe from the other side.
     
  4. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    Yup. It's just generally called a measure of randomness because historically it wasn't easy to measure how "random" something is. I guess it was a lot easier to understand "predictability".
     
  5. Pauly

    Pauly Season Ticket Holder

    3,378
    3,376
    113
    Nov 29, 2007
    A coin toss, for an example, is NOT random.

    As soon as you apply force to the coin Newtonian physics take over and the result is inevitable.
    You can demonstrate this by learning to toss a coin so that it does a half turn and it will always give you heads. With a little bit of practice you can learn how to flip a coin so that it turns 2 or 3 times in the air and almost always getting heads. I saw a TV program years ago where the engineering challenge was to build a machine that would give you a 99%+ chance of getting a heads result with the maximum number of turns in the air, and IIRC the record was 12 turns in the air.

    However, as Cbrad said earlier in this thread Kolmogorov effects take over when people are tossing coins in live situations.

    So a coin toss is better described as a non-replicable experiment than a random event.

    However, for practical purposes a coin toss can be treated as if it were a random event.
     
    resnor likes this.
  6. roy_miami

    roy_miami Well-Known Member

    1,385
    560
    113
    Oct 11, 2013
    Thats why the non-flipper calls it in the air...
     
    Fin-O likes this.
  7. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    Well.. if we're going to be technical, then every event is random as far as we know in physics. That's what quantum mechanics tells us. And importantly, it's inherently random, not simply statistically random (so random because that's how nature is, not because humans just haven't built a machine that can get the necessary information to make an accurate prediction).

    In any case, like I said before, all this is moot because we're talking statistics here and any event that cannot be predicted with the information available is by definition (statistically) random to some degree.
     
  8. Pauly

    Pauly Season Ticket Holder

    3,378
    3,376
    113
    Nov 29, 2007
    What I was getting at was that even if something isn't technically random, it can be random for functional purposes.
     
  9. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    Yeah that's true. Chaos theory might be the best scientific example. They deal with phenomena where you don't have to worry about quantum mechanics and so you can assume the system is in principle deterministic. But it turns out arbitrarily small differences between initial conditions can lead to totally different results, making the system in practice probabilistic. Weather and climate models have to worry about this.
     
  10. Pauly

    Pauly Season Ticket Holder

    3,378
    3,376
    113
    Nov 29, 2007
    So to bring it back to football, even if what the OC calls isn't a technically random choice and what the DCs call isn't technically random there is no problem treating them as if they were random for the purpose of analysis.
     
    resnor likes this.
  11. Finster

    Finster Finsterious Finologist

    3,087
    2,038
    113
    Jul 27, 2013
    Now how I thought it worked was that the unpredictability applied to random, followed the random, e.g., the roll of the dice is unpredictable because the dice are random, and not that the dice are random because they are unpredictable, the latter being basically a non sequitur, i.e., unpredictable is not synonymous with random.

    So, the lack of knowledge of an unpredictable event, does not therefor make the event random.

    At least that's how I understood it.

    However, perhaps in a "lab" type environment it is the common vernacular.
     
  12. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    Maybe I'm to simple-minded, but you can't be predictable AND random. If something is predictable, it is therefore, not random. If something is unpredictable, then it must be random, having no pattern.
     
  13. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    Yeah, there are really two types of randomness used in science: mathematical (statistical) and physical (only really used by physicists).

    Statistical randomness refers to a mathematical, not physical property, specifically the property of unpredictability. You have to understand that math is a field that exists in its own right, independent of any study of the physical world. So the mathematical foundations of statistics (where they actually try to formalize this stuff) makes no attempt to define "randomness" or "probability" in a physical sense. In fact, that's a long-standing unsolved problem: what precisely do the mathematical foundations of probability theory refer to in the physical world. Point is, while statistics is one of the most useful forms of math ever developed and applied to study physical phenomena, the math is defined independent of any notions of the physical world, and in statistics the (informal) definition of randomness = unpredictability.

    The physical definition of randomness really doesn't exist per se. A property of randomness for a physicist is unpredictability, and you're absolutely right about the logic there that the property of unpredictability follows random, not the other way around. But in practice, when a physicist says Quantum Mechanics tells us all events in nature are probabilistic (meaning they are random, but not equally random), what they mean is that it is not even in theory possible to obtain the necessary info to predict it. Nature is just built in such a way so that you actually cannot obtain that information.

    Hope that helps clear stuff up a bit.

    Think of random = black and predictable = white. Clearly there are shades of grey. Similarly, we can talk about degrees of randomness. That is, not all events are equally random or equally uncertain, and the concept we use to quantify this is probability. An event with probability 0 or probability 1 (100%) of occurring is essentially perfectly predictable, while a probability of 1/N for N possible outcomes has the highest level of uncertainty. So there's a continuum here, not just two possibilities.
     
    Pauly and Finster like this.
  14. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    Yeah...if something can be predicted with any amount of regularity, I wouldn't call it random. That may not be scientific, though.

    Like, to me, random is or it isn't. It's either random, or is not. White is always white. If it has a shade to it, it's no longer white. White can't be gray, cause then it's gray, not white. One something can be predicted successfully, to any degree, it's, to me, not random.
     
  15. roy_miami

    roy_miami Well-Known Member

    1,385
    560
    113
    Oct 11, 2013
    If you have a million sided dice with one side being green and the rest red, can you predict what color it will land on?
     
  16. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    Are we conflating probability with random here to some extent? Like, I shuffle a deck of cards, and draw six of the top. You could figure the probability of the sequence of cards I draw, but that is far different than predicting it. The act of shuffling and drawing is random. Or, I throw 3000 marbles in the air. You could find the probability of a particular marble falling into a cup, but it would still be a random event.


    I think. Lol
     
    Fin D likes this.
  17. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    I don't think so. You could find the probability, but that is not the same as predicting it. Let's say probability said that 1 in 1,000,000,000 rolls would land on green. Predicting it isn't every time you roll, you say "It will land on green," and eventually it does. Predicting would be saying "On roll 3,978,456, the die will land on green." And it does.
     
    Fin D likes this.
  18. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    What does "any" amount of regularity mean in your mind though? Suppose there are two coins A and B. Using all the information available to you in the best possible manner, suppose you can predict what happens to A (heads or tails) 90% of the time, but for B it's only 70% of the time. Clearly, there is a difference in degree of predictability here even though both events are random events.
     
  19. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    Probability is not prediction.

    Question: is it possible for me to flip a coin, and get heads (without manipulating the flip as described earlier) 100 times in a row? Of course it's possible. What is the probability? Could you predict when someone would do it, even when knowing the probability?
     
  20. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    In relation to football, let's say you know that the stats say an opponent runs off the left tackle 68% of the time in 3rd and goal. Your best guess, as a DC, if you're in that situation, is to call a defense that stuffs any run to the left side. However, that's simply playing to the probability. You're not predicting, your gambling.
     
    Fin D likes this.
  21. Stringer Bell

    Stringer Bell Post Hard, Post Often Club Member

    43,990
    21,652
    113
    Mar 22, 2008
    It is not gambling if you know the odds are in your favor.

    Sent from my SAMSUNG-SM-G920A using Tapatalk
     
  22. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    It's still gambling. Just because probability is in your favor, this might be one of the 32% of the times that the OC runs to the right, goes with a QB sneak up the middle, or rolls or right and throws.

    Anything that isn't 100% certain is a gamble.
     
  23. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    Yeah I got that from your previous post. My question to you is whether you wouldn't agree that predicting something 90% of the time correctly vs. 70% of the time correctly shows different degrees in your ability to predict things?

    I'm just pointing out it's perfectly valid to have a measure of the degree to which you can predict something correctly. What you're referring to is what would be called either 0% probability (event does not occur) or 100% probability (event is guaranteed to occur). I mean.. there's a huge utility to having a continuum here.

    Now if you're not against having a continuum, then it's just semantics we're talking about and in that case it's best to just go with common usage, which in statistics is how I'm using the words.
     
  24. Fin D

    Fin D Sigh Club Member

    72,252
    43,682
    113
    Nov 27, 2007
    wait.....what?
     
    resnor likes this.
  25. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    I bet he's in deep with some Vegas bookies. He still can't figure out why he's losing on "sure things."
     
    Fin D likes this.
  26. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    You're still not differentiating between probability and predicting. Telling the probability of an event is not the same as predicting that event.

    You can have varying degrees of success in predicting, sure. But then those events you're predicting are not random. If you can use patterns and other data to successfully predict an outcome (predict, NOT give probability of event), then the event is not random. It can't be.
     
    Fin D likes this.
  27. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    OK, let's take your sentences one at a time here. You admit you can have varying degrees of success at predicting. Then you say if that's true, then those events are not random.

    That part is patently false. Go down to the subatomic level where once again quantum mechanics tells you nature is inherently random (so we KNOW the events are random). Well.. in QM you predict the probability of those events (correctly I might add). And depending on the event, the probability and the degree to which you successfully predict the event changes.

    So before going any further, can you at least see why you can have random events that can be predicted to different degrees?
     
    Finster and roy_miami like this.
  28. Fin D

    Fin D Sigh Club Member

    72,252
    43,682
    113
    Nov 27, 2007
    Statistics.
    Random: of or characterizing a process of selection in which each item of a set has an equal probability of being chosen.
     
    resnor likes this.
  29. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    Perhaps it's a word thing...

    Tell me what you mean with "predict." I feel like when you say "predict," it means "to tell the probability of an event."

    Like I've said, those are two very different things to me. If they aren't to you, then no need to argue about it. LOL
     
  30. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    Yeah, whoever wrote that doesn't understand statistics. Wikipedia actually has it mostly correct. Look up probability theory (which is the underpinning of statistics):
    https://en.wikipedia.org/wiki/Probability_theory

    and you'll see they talk about random variables, which are the basis of the foundations of probability:
    https://en.wikipedia.org/wiki/Random_variable

    and in that first paragraph you see that random variables can take on any probability. So wherever you got that quote from, the person writing it doesn't understand probability theory correctly.
     
  31. Fin D

    Fin D Sigh Club Member

    72,252
    43,682
    113
    Nov 27, 2007
    Dictionary.com

    Is "random" and "random variable" the same thing? I mean I don't think it is ....
     
  32. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    Our understanding of "predict" is the same. I think the first source of confusion (based on your responses) is you think one can only predict individual physical events. Thing is, you can predict all kinds of things, like distributions of events, or the frequency of events over time.

    Take that latter idea. So the goal is to predict the frequency with which a coin comes up heads over 1000 flips. Now, you've already admitted that there are different degrees with which you can predict that. So maybe you only predict that 90% of the time correctly, or maybe less. Either way, these are different degrees with which you can predict something.

    OK, so far no mention of probability or randomness. First thing: assume that each coin flip was a random event (again, if you disagree in principle, just go to QM). So you can predict the frequency of random events with different degrees of success. Second thing: the (true) probability of that event (if it's a natural event) equals over time that frequency. Now you might THINK the probability is something else, but that's because your ability to predict wasn't good enough.

    So yes there's a difference between the probability of an event and prediction ability, but if you assume all available information is used as best as theoretically possible, then that difference disappears in practice.
     
  33. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    It's crazy. We all use the same words...yet somehow, when I talk to math people, we are not ever saying the same thing.
     
  34. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    No...predict for me exactly how many times I have to flip a coin to get heads 10 times in a row.

    Don't tell me the probability of it happening. Predict for me when it will happen.
     
  35. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    In statistics, a "random variable" is a mathematical function that applies the notion of "random" so that it can be used in math. The reason for my previous response was because it said "Statistics" right above the definition, and there what they wrote (the equal probability thing) is patently wrong.

    So the two concepts are different. "random variable" is a precise concept in math, while "random" is used both in math and outside of it, often with ambiguous meanings as this entire discussion demonstrates.
     
  36. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    The bold things are all probability. Those are not predicting.

    Unless those words mean different things in math world.
     
  37. Fin D

    Fin D Sigh Club Member

    72,252
    43,682
    113
    Nov 27, 2007
    I think the bold is where these concepts breakdown for football.

    In this coin flip hypothetical, you can control virtually everything, in the real world, you cannot. LOTS of pertinent information is either not tracked or not really possible to be tracked. You aren't and can't account for the unpredictability of humans doing human things. Flipping a coin has a very small set of circumstances that can affect the outcome, completing a pass in game conditions, is exponentially more complex.
     
  38. resnor

    resnor Derp Sherpa

    14,488
    8,397
    113
    Nov 25, 2007
    New Hampshire
    There's nothing ambiguous, except that statisticians have changed the meaning of the word in math.
     
  39. Fin D

    Fin D Sigh Club Member

    72,252
    43,682
    113
    Nov 27, 2007
    WADR, I don't see where you've shown the definition to be patently wrong.
     
  40. cbrad

    cbrad . Club Member

    8,697
    10,251
    113
    Dec 21, 2014
    Yeah, it's because in math you have to worry about all kinds of consequences of defining words one way or another. I mean you make the tiniest logical error and people (rightfully) throw your idea out as trash haha!

    Yeah, like I said, it's useful to talk about the degree to which you can predict something, which in this case is quantified by the probability with which you get that answer right. Not much more I can say!
     

Share This Page