Our Blog


Inductive Reasoning, Non Blondes, and Applying DFS Trends

I’m going to attempt to be quick in the writing of this article. I have a lot of other pressing stuff to do. I need to bathe the dog. I need to mow the yard. I need to shower. And I definitely need to shave. Most importantly, I need to clean the house. My wife returns today, and if everything doesn’t look better than it did before she left, then I might not get to experience the pleasure of doing that thing I like to do. You know, I probably didn’t need to say that. Let’s move on.

This is the 33rd installment of The Labyrinthian, a series dedicated to exploring random fields of knowledge in order to give you unordinary theoretical, philosophical, strategic, and/or often rambling guidance on daily fantasy sports. Consult the introductory piece to the series for further explanation.

Inductive Reasoning

Based on the context in which it was situated, the phrase “that thing I like to do” might have struck you as a likely allusion to sex. The logical process that led you to that conclusion was inductive reasoning. Using your knowledge of the world (and me), you made a probabilistic guess. That’s exactly what inductive reasoning is.

A few philosophical definitions of inductive reasoning exist. Many people think of it strictly as the simple logical progression from the particular to the general, as in “I have seen only white swans my entire life, so swans can be only white.” Really, the heart of inductive reasoning lies in its reliance on the probabilistic gesture, its conjectural quality. We observe a series of related facts, and on the basis of our observations we after a time make what we hope is a reasonable assumption, such as this:

I have seen perhaps thousands of swans in my life, all of them white. They are probably representative of all the other swans out there that I haven’t seen. Therefore, it’s probable that all swans are white.

Inductive reasoning deals with uncertainty. Given the facts we have observed, the conclusions we reach might be true. This type of reasoning stands in juxtaposition to deductive reasoning, in which, given premises that are true, the conclusions we reach must be true. For instance:

Every article I write takes longer than I want it to take.
I am writing this article right now.
Therefore, this article must be taking longer than I want it to take.

This deductive argument is constructed in such a way (moving from the general to the particular) that, if the premises are true — and I assure you that they are — then the conclusion must be true as well.

Unlike deduction, induction offers no guarantees. We can never be sure that all swans are white. We can only say something to the effect of it being probable that the next swan we see will not be a Black Swan.

The Myth of Deductive Reasoning

Deductive reasoning is nice and clean. For small matters, it gets the job done. But in the real world, in which uncertainty reigns, deduction is almost useless. We are always in the realm of induction. Any important decision that you will ever make will be the result of inductive reasoning. In part, what elevates a decision to the status of “important” is that it deals with matters that possess a degree of unknowability. Deductive reasoning itself is not a myth — it exists — but it might as well not exist when one is always immersed in probabilities.

In fact, too many people rely on the frame and form of deductive reasoning. Even if they don’t know exactly what deduction is, people employ it too often, assuming that more things in the world are more certain than they are. They act as if deduction is sufficient for the creation of absolute answers when in reality deduction is not only insufficient but also misleading. For our world — be it the larger planet or the smaller sphere of DFS — only induction is sufficient to help us create reasonable answers to the questions we have. And induction isn’t just sufficient. It is necessary.

In the last installment of The Labyrinthian, I divulged that for years I used women’s (lack of) interest of Annie Hall as a screen for compatibility. In doing this, I was relying firmly on inductive reasoning. I reasoned that anyone who didn’t like the movie would probably not be a good fit for me over the long run because if someone didn’t like Annie Hall then that person probably would not have the same perspectives on life that I have. There are a lot of spoken (and unspoken) probabilistic logical moves in that last sentence. Deductive reasoning would be of only limited use in this matter.

Again, people employ the mechanisms of deductive reasoning far too often, assuming that what is uncertain is entirely certain. For instance, I know for a fact that one of my friends had this exact series of deductive thoughts in college:

Freedman likes blonde-headed women, because I’ve never seen him date a brunette or redhead.
That woman right there is a blonde.
Ergo, Freedman will like her.

Soooo much is wrong with that thought process. Let’s just set aside the potential objectification of women and deal with the logical slipperiness. The first premise assumes that I like all blondes. That’s not true now, and it definitely wasn’t true in college. Although it’s true — by coincidence — that the only girls I dated early in college were blondes (on a relatively small sample size), starting my final year of college I dated only non-blondes — again, by coincidence. All of which leads me to my next point.

Some Non-Blondes

When induction is framed as if it were deduction, people make many logical mistakes. For instance, my friend assumed (on the basis of my dating history) that I liked all blonde women. He also assumed that I liked only blond women. And, finally, he assumed that I would always like blonde women.

None of that was true. His inductive reasoning, dressed in the trappings of deduction, was highly flawed. I was in college. I liked women. That was about it. But because I was idiosyncratic I definitely didn’t romantically like all blonde women (as women who like Annie Hall, The Lord of the Rings, and The Godfather are relatively rare, regardless of hair color).

And although I was attracted to some blondes — because some women I found attractive happened to be blondes — that didn’t mean that I was attracted only to blondes. (I can still remember the redhead who got away. [Remember to delete that last sentence later.]) And just because the girls I had dated in college up to that point were blondes — I went to Texas Christian University, so blondes were kind of inescapable — that didn’t mean that I would be destined to be with blondes for the rest of my life.

I had no problems with non-blondes early in college, but my friend’s logic seemed to be based on this series of assumptions stemming from one observation:

I’ve never seen Freedman date a brunette or redhead.
He must not like brunettes or redheads.
Freedman must like blondes.
He must like all blondes.
Thus, Freedman’s preference for blondes will extend into the future.

To bring this history to a close, all I can say is that 1) I’m glad in graduate school that I asked one particular non-blonde what her favorite movie was, and 2) we often use deductive reasoning when we shouldn’t.

David Hume and the Problem of Induction

The thing is that even when we aren’t confusing induction for deduction — even when we know that we are dealing with uncertainty — we make faulty assumptions or overestimate the odds that our predictions are correct. This human inability to reason correctly David Hume called “The Problem of Induction.”

Hume was an 18th-Century Scottish thinker. Some people consider him a philosopher, historian, or economist. Really, he was just a guy who thought very deeply about a lot of things. Like many of the best thinkers in history, he was a skeptic. Eventually, he became interested in not only thinking and knowing about lots of things. He became interested in thinking and knowing about how people think and know.

That Nassim Nicholas Taleb even bothers to speak of Hume without condescension should give you a sense of how great he is. In The Black Swan, Taleb pays Hume this compliment.

Hume wrote with such clarity that he puts to shame almost all current thinkers, and certainly the entire German graduate curriculum. Unlike Kant, Fichte, Schopenhauer, and Hegel, Hume is the kind of thinker who is sometimes read by the person mentioning his work.

As an aside, I suspect that if Taleb were half the *sshole he is then he would be one-fourth as enjoyable to read. Anyway . . .

For Hume, the Problem of Induction is actually pretty simple:

  1. We apply from the particular to the general too liberally. We assign too high of a probability to the uncertain. Even though every swan we have ever seen is white, we can never be 100 percent sure that all swans are white.
  2. We project from the present to the future too liberally. We assign too high of a probability to our predictions. Even though the sun has risen each day of my life, I cannot be 100 percent sure that the sun will rise tomorrow morning.

In short, the Problem of Induction is that we can never be certain about what is and what will be.

Applying DFS Trends

Uncertainty is not just a problem with inductive reasoning. It’s a problem with life. And it’s a problem with DFS.

When you use our Trends tool to test ideas you have about DFS valuation and performance, you are always combatting the Problem of Induction. Are the trends that you create descriptive? Or predictive? Do they accurately signal actionable facts that you can leverage to your benefit? Or are they merely noise — the sound and the fury of randomness, signifying nothing?

When you create a trend, you must always attempt to gauge its reliability. What is the probability that this trend actually describes what you are wanting it to describe? What are the odds that this trend will be applicable to additional data sets in the future? Sadly, the trends cannot, with full certainty, tell you precisely what is and will be.

This is not to say that our Trends tool is useless or problematic. To the contrary, it is immensely powerful. It doesn’t need to be omniscient to possess the data to enable you to attain greater knowledge. It merely needs to be used correctly.

Walking Around in Circles

My sense is that, to get the most out of our Trends tool, one must appreciate its limitations. One must understand that not all trends you are able to create are predictive. You can spend hours in the Trends tool just going around in circles, carefully building 20-screen monstrosities that mean almost nothing. But your time would probably be better spent doing something else.

You can and should experiment by constructing a number of trends that you will never use, just so that you can learn how the machine works. Just don’t forget not to confuse those early, throwaway trends for ones that actually mean something. For some people, those early, screen-laden trends can be enticing.

For instance, if you discover that a subsection of batters in a particular division does well according to the Plus/Minus metric on a particular day of the week in a certain month, you should be skeptical. I’m not saying that this trend has no merit, because it actually might (depending on the filters that you have used and the sample size).

I am saying, though, that you should examine what this trend might actually signify. Why would this trend hold true in the future? What is it about this particular type of batter that is significant? Why would this particular type of batter benefit from being in this division? What is it about this day of the week or this month of the season that improves the performance of these batters?

What I’m saying is that Hume’s Problem of Induction applies to DFS and especially to our application of trends. We can’t take on faith — we can’t be 100 percent sure — that our trends are meaningful and predictive. We need to examine them skeptically. We must explore them so as to understand when and how they should inform our DFS decisions. And most importantly, we should endeavor to create trends that have an enhanced quality of coherence.

To use a trend that has significance, that trend should be intuitive. One should be able to understand why the filters used were applied collectively. Keeping in mind the limitations of our knowledge, one should employ trends that enhance our understanding but do not purport to be definitive representations.

To a degree, we are all bound to walk around in circles because we can never know precisely where the path leads. Ironically, that’s a problem only for the people who are absolutely certain that they know where they’re going.

———

The Labyrinthian: 2016, 33

Previous installments of The Labyrinthian can be accessed via my author page. If you have suggestions on material I should know about or even write about in a future Labyrinthian, please contact me via email, [email protected], or Twitter @MattFtheOracle.

I’m going to attempt to be quick in the writing of this article. I have a lot of other pressing stuff to do. I need to bathe the dog. I need to mow the yard. I need to shower. And I definitely need to shave. Most importantly, I need to clean the house. My wife returns today, and if everything doesn’t look better than it did before she left, then I might not get to experience the pleasure of doing that thing I like to do. You know, I probably didn’t need to say that. Let’s move on.

This is the 33rd installment of The Labyrinthian, a series dedicated to exploring random fields of knowledge in order to give you unordinary theoretical, philosophical, strategic, and/or often rambling guidance on daily fantasy sports. Consult the introductory piece to the series for further explanation.

Inductive Reasoning

Based on the context in which it was situated, the phrase “that thing I like to do” might have struck you as a likely allusion to sex. The logical process that led you to that conclusion was inductive reasoning. Using your knowledge of the world (and me), you made a probabilistic guess. That’s exactly what inductive reasoning is.

A few philosophical definitions of inductive reasoning exist. Many people think of it strictly as the simple logical progression from the particular to the general, as in “I have seen only white swans my entire life, so swans can be only white.” Really, the heart of inductive reasoning lies in its reliance on the probabilistic gesture, its conjectural quality. We observe a series of related facts, and on the basis of our observations we after a time make what we hope is a reasonable assumption, such as this:

I have seen perhaps thousands of swans in my life, all of them white. They are probably representative of all the other swans out there that I haven’t seen. Therefore, it’s probable that all swans are white.

Inductive reasoning deals with uncertainty. Given the facts we have observed, the conclusions we reach might be true. This type of reasoning stands in juxtaposition to deductive reasoning, in which, given premises that are true, the conclusions we reach must be true. For instance:

Every article I write takes longer than I want it to take.
I am writing this article right now.
Therefore, this article must be taking longer than I want it to take.

This deductive argument is constructed in such a way (moving from the general to the particular) that, if the premises are true — and I assure you that they are — then the conclusion must be true as well.

Unlike deduction, induction offers no guarantees. We can never be sure that all swans are white. We can only say something to the effect of it being probable that the next swan we see will not be a Black Swan.

The Myth of Deductive Reasoning

Deductive reasoning is nice and clean. For small matters, it gets the job done. But in the real world, in which uncertainty reigns, deduction is almost useless. We are always in the realm of induction. Any important decision that you will ever make will be the result of inductive reasoning. In part, what elevates a decision to the status of “important” is that it deals with matters that possess a degree of unknowability. Deductive reasoning itself is not a myth — it exists — but it might as well not exist when one is always immersed in probabilities.

In fact, too many people rely on the frame and form of deductive reasoning. Even if they don’t know exactly what deduction is, people employ it too often, assuming that more things in the world are more certain than they are. They act as if deduction is sufficient for the creation of absolute answers when in reality deduction is not only insufficient but also misleading. For our world — be it the larger planet or the smaller sphere of DFS — only induction is sufficient to help us create reasonable answers to the questions we have. And induction isn’t just sufficient. It is necessary.

In the last installment of The Labyrinthian, I divulged that for years I used women’s (lack of) interest of Annie Hall as a screen for compatibility. In doing this, I was relying firmly on inductive reasoning. I reasoned that anyone who didn’t like the movie would probably not be a good fit for me over the long run because if someone didn’t like Annie Hall then that person probably would not have the same perspectives on life that I have. There are a lot of spoken (and unspoken) probabilistic logical moves in that last sentence. Deductive reasoning would be of only limited use in this matter.

Again, people employ the mechanisms of deductive reasoning far too often, assuming that what is uncertain is entirely certain. For instance, I know for a fact that one of my friends had this exact series of deductive thoughts in college:

Freedman likes blonde-headed women, because I’ve never seen him date a brunette or redhead.
That woman right there is a blonde.
Ergo, Freedman will like her.

Soooo much is wrong with that thought process. Let’s just set aside the potential objectification of women and deal with the logical slipperiness. The first premise assumes that I like all blondes. That’s not true now, and it definitely wasn’t true in college. Although it’s true — by coincidence — that the only girls I dated early in college were blondes (on a relatively small sample size), starting my final year of college I dated only non-blondes — again, by coincidence. All of which leads me to my next point.

Some Non-Blondes

When induction is framed as if it were deduction, people make many logical mistakes. For instance, my friend assumed (on the basis of my dating history) that I liked all blonde women. He also assumed that I liked only blond women. And, finally, he assumed that I would always like blonde women.

None of that was true. His inductive reasoning, dressed in the trappings of deduction, was highly flawed. I was in college. I liked women. That was about it. But because I was idiosyncratic I definitely didn’t romantically like all blonde women (as women who like Annie Hall, The Lord of the Rings, and The Godfather are relatively rare, regardless of hair color).

And although I was attracted to some blondes — because some women I found attractive happened to be blondes — that didn’t mean that I was attracted only to blondes. (I can still remember the redhead who got away. [Remember to delete that last sentence later.]) And just because the girls I had dated in college up to that point were blondes — I went to Texas Christian University, so blondes were kind of inescapable — that didn’t mean that I would be destined to be with blondes for the rest of my life.

I had no problems with non-blondes early in college, but my friend’s logic seemed to be based on this series of assumptions stemming from one observation:

I’ve never seen Freedman date a brunette or redhead.
He must not like brunettes or redheads.
Freedman must like blondes.
He must like all blondes.
Thus, Freedman’s preference for blondes will extend into the future.

To bring this history to a close, all I can say is that 1) I’m glad in graduate school that I asked one particular non-blonde what her favorite movie was, and 2) we often use deductive reasoning when we shouldn’t.

David Hume and the Problem of Induction

The thing is that even when we aren’t confusing induction for deduction — even when we know that we are dealing with uncertainty — we make faulty assumptions or overestimate the odds that our predictions are correct. This human inability to reason correctly David Hume called “The Problem of Induction.”

Hume was an 18th-Century Scottish thinker. Some people consider him a philosopher, historian, or economist. Really, he was just a guy who thought very deeply about a lot of things. Like many of the best thinkers in history, he was a skeptic. Eventually, he became interested in not only thinking and knowing about lots of things. He became interested in thinking and knowing about how people think and know.

That Nassim Nicholas Taleb even bothers to speak of Hume without condescension should give you a sense of how great he is. In The Black Swan, Taleb pays Hume this compliment.

Hume wrote with such clarity that he puts to shame almost all current thinkers, and certainly the entire German graduate curriculum. Unlike Kant, Fichte, Schopenhauer, and Hegel, Hume is the kind of thinker who is sometimes read by the person mentioning his work.

As an aside, I suspect that if Taleb were half the *sshole he is then he would be one-fourth as enjoyable to read. Anyway . . .

For Hume, the Problem of Induction is actually pretty simple:

  1. We apply from the particular to the general too liberally. We assign too high of a probability to the uncertain. Even though every swan we have ever seen is white, we can never be 100 percent sure that all swans are white.
  2. We project from the present to the future too liberally. We assign too high of a probability to our predictions. Even though the sun has risen each day of my life, I cannot be 100 percent sure that the sun will rise tomorrow morning.

In short, the Problem of Induction is that we can never be certain about what is and what will be.

Applying DFS Trends

Uncertainty is not just a problem with inductive reasoning. It’s a problem with life. And it’s a problem with DFS.

When you use our Trends tool to test ideas you have about DFS valuation and performance, you are always combatting the Problem of Induction. Are the trends that you create descriptive? Or predictive? Do they accurately signal actionable facts that you can leverage to your benefit? Or are they merely noise — the sound and the fury of randomness, signifying nothing?

When you create a trend, you must always attempt to gauge its reliability. What is the probability that this trend actually describes what you are wanting it to describe? What are the odds that this trend will be applicable to additional data sets in the future? Sadly, the trends cannot, with full certainty, tell you precisely what is and will be.

This is not to say that our Trends tool is useless or problematic. To the contrary, it is immensely powerful. It doesn’t need to be omniscient to possess the data to enable you to attain greater knowledge. It merely needs to be used correctly.

Walking Around in Circles

My sense is that, to get the most out of our Trends tool, one must appreciate its limitations. One must understand that not all trends you are able to create are predictive. You can spend hours in the Trends tool just going around in circles, carefully building 20-screen monstrosities that mean almost nothing. But your time would probably be better spent doing something else.

You can and should experiment by constructing a number of trends that you will never use, just so that you can learn how the machine works. Just don’t forget not to confuse those early, throwaway trends for ones that actually mean something. For some people, those early, screen-laden trends can be enticing.

For instance, if you discover that a subsection of batters in a particular division does well according to the Plus/Minus metric on a particular day of the week in a certain month, you should be skeptical. I’m not saying that this trend has no merit, because it actually might (depending on the filters that you have used and the sample size).

I am saying, though, that you should examine what this trend might actually signify. Why would this trend hold true in the future? What is it about this particular type of batter that is significant? Why would this particular type of batter benefit from being in this division? What is it about this day of the week or this month of the season that improves the performance of these batters?

What I’m saying is that Hume’s Problem of Induction applies to DFS and especially to our application of trends. We can’t take on faith — we can’t be 100 percent sure — that our trends are meaningful and predictive. We need to examine them skeptically. We must explore them so as to understand when and how they should inform our DFS decisions. And most importantly, we should endeavor to create trends that have an enhanced quality of coherence.

To use a trend that has significance, that trend should be intuitive. One should be able to understand why the filters used were applied collectively. Keeping in mind the limitations of our knowledge, one should employ trends that enhance our understanding but do not purport to be definitive representations.

To a degree, we are all bound to walk around in circles because we can never know precisely where the path leads. Ironically, that’s a problem only for the people who are absolutely certain that they know where they’re going.

———

The Labyrinthian: 2016, 33

Previous installments of The Labyrinthian can be accessed via my author page. If you have suggestions on material I should know about or even write about in a future Labyrinthian, please contact me via email, [email protected], or Twitter @MattFtheOracle.

About the Author

Matthew Freedman is the Editor-in-Chief of FantasyLabs. The only edge he has in anything is his knowledge of '90s music.