How do you decide what you call "True"?

[ygolo]

I am curious what set answers we will get from this question.

Note, I am nat asking the the more static "What is truth?" question.

I am asking from a day-to-day, minute-to-minute decision making, how do you decide what you consider "true" (in practical, theoretical, or sptitual matters)?

How do time constraints on those decisions effect your desicions in this matter?

How would you like to improve your day-to-day desicion making process about what is true?

 

[ptgatsby]

Interesting question. True to me is predictive. When I do research, I'm attempting to build a predict model so that I can base my next actions upon it. Anything that fails to be predictive is inherently false.

Of course, that gets into the whole probability thing, which makes it difficult. In cases where probability is low (like 50/50), I either need a large sampling of trial events or I don't consider it true.

In logical models, which I'll give an example of, I will use historical data to validate it as much as possible. In those cases, I also tend to try to use as many sets of data as possible;

For instance, one of the recent problems I'm facing is how to invest money - logically, it follows that taking all of you home equity (interest only loan) and investing it in the market, and then leveraging that money as much as possible, is the correct decision. The reason is that inflation adds to your earning power while the loan remains fixed, and the assets you buy are generally inflationary as well.

However, that is data-free. As such, I would pull out as much historical data (such as bank stock prices, broad indexes, foreign markets) and run the same calculation on each market - I would write the full calculation, work out the risk of ruin, and then calculate it over multiple periods of time (ie; 5, 10, 25 yrs) and output the average return, average risk of ruin, etc. No only would this data either support or deny the logical solution, it would provide variation and other data (ie: objective risk of ruin rather than just the mathematical model).

Only after that would I consider the theory true. That goes for nearly anything I do.

Of course, when it comes to acting, I use as much "true" (ie: validated) information as possible, but I will make my best guess as to the rest and move on. No one has unlimited time!

 

[ygolo]

For instance, one of the recent problems I'm facing is how to invest money - logically, it follows that taking all of you home equity (interest only loan) and investing it in the market, and then leveraging that money as much as possible, is the correct decision. The reason is that inflation adds to your earning power while the loan remains fixed, and the assets you buy are generally inflationary as well.

Reasoning like this is why I think loss aversion is a really good human bias (No offence). You are not already doing this are you?

Your risk of ruin analysis makes more sense. I think the utility of money deteriorates as you have more of it (So it actually makes more sense to me to be loss averse).

Consider the following two gambling situations, would you choose to proceed?:

1. On a flip of a coin you will either get your own networth in money plus $1, or you will loose your networth in money.
2. On the flip of a coin, you will gain $1 and a $1/(your net-worth) or lose $1.

Expected value analysis makes sense in one case, probabablity or ruin in the other, but value at risk analysis, I think, makes sense in both, and lets you know what sort of analysis to do going forward.

Anyway, I don't mean to be judgemental, but I have friends who were burned by investing too agressively.

 

[Zergling]

I tend to just assume what I see in front of me is true.

If I hear something from other people,I tend to figure it is true unless I have some reason to think otherwise.

 

[ygolo]

I tend to just assume what I see in front of me is true.

If I hear something from other people,I tend to figure it is true unless I have some reason to think otherwise.

I do too, usually. Is this a good thing? Would you like to improve?

I have become better about remember who told me what, so that I can start knowing how well to trust my sources over time.

But in trickier situations, I do have too look beyond the surface and theorize and test. I'm sure you do too. Consider when you have to "fix" something. The first thing I do is to check if the thing is actually broken (I have actually that in most cases when someone asks me to fix something, it is simply in a weird mode, not actually broken).

 

[ptgatsby]

Reasoning like this is why I think loss aversion is a really good human bias (No offence). You are not already doing this are you?

No, I haven't completed my analysis yet.

Your risk of ruin analysis makes more sense. I think the utility of money deteriorates as you have more of it (So it actually makes more sense to me to be loss averse).

That depends on your goals - the ROR is tied into the whole model. I operate on the principle that you figure out what you need to know, then how you can objectively find out, then build the model, then test the model, then apply the model.

I'm currently at the "build the model" stage.

Consider the following two gambling situations, would you choose to proceed?:

1. On a flip of a coin you will either get your own networth in money plus $1, or you will loose your networth in money.
2. On the flip of a coin, you will gain $1 and a $1/(your net-worth) or lose $1.

Yes, I know what you are saying. FWIW, I'd take the first regardless, because I do not have time for infinite iterations of the second. The second would offer roughly me less than 0.0005% benefit if iterated daily for the next 10 years (I'm basing this upon my current projection using excel), while a single iteration that was in my advantage would reduce my currently time from by nearly two years (and, as such, only increase my respective time frame by a few years; the relative value says I wouldn't do either). To put that in perspective, it'd take thousands upon thousands of iterations early on, to have any significant impact... as such, the second doesn't even factor into it at all and as such, holds no interest to me.

Model first, then determine what will achieve your goals. The logical model needs to be proven, then feasibility needs to be shown.

Anyway, I don't mean to be judgemental, but I have friends who were burned by investing too agressively.

I was using it as an illustration on how I make decisions and how I support them - I haven't concluded anything.

FWIW, the inflation argument is also more similar to your second example the the first... although pruchasing power and asset inflation is like rolling doubly weighted dice, which is why the risk of ruin is the most important factor in determining proper leverage. It is the length of time that has the most significant impact on the viability of the strategy.

(But not to derail the thread - I'm more than willing to talk all about this if you are interested... or anyone for that matter.)

 

[reason]

bachelors are unmarried men - true

 

[ptgatsby]

bachelors are unmarried men - true

How do you know someone is a bachelor (or an unmarried man)?

 

[reason]

How do you know someone is a bachelor (or an unmarried man)?

I didn't claim to.

 

[ptgatsby]

I didn't claim to.

So you don't know if anyone is married, then? This is not a group you ever identify?

 

[reason]

So you don't know if anyone is married, then?

That's right.

This is not a group you ever identify?

Oh, I identify the group. I just have no way of knowing that anybody is married.

 

[ptgatsby]

Oh, I identify the group. I just have no way of knowing that anybody is married.

I take it this prevents you from hitting on women...?

 

[ygolo]

bachelors are unmarried men - true

Got it. Tautologies and definitions are true by definition. But do you really make this type of decision on a day-to-day basis?

If so, you just made me feel incredibly normal.

 

[ygolo]

Yes, I know what you are saying. FWIW, I'd take the first regardless, because I do not have time for infinite iterations of the second. The second would offer roughly me less than 0.0005% benefit if iterated daily for the next 10 years (I'm basing this upon my current projection using excel), while a single iteration that was in my advantage would reduce my currently time from by nearly two years (and, as such, only increase my respective time frame by a few years; the relative value says I wouldn't do either). To put that in perspective, it'd take thousands upon thousands of iterations early on, to have any significant impact... as such, the second doesn't even factor into it at all and as such, holds no interest to me.

Keep in mind, as your networth increases, in both cases your expected rate of return decreases (both options are lousy investments).

But, lets see how the "bet-everything-everytime-on-the-same-coin-strategy" works out over the long run in terms of probability of ruin....

The formula is simple: For n iterations, p=1-(.5)^n is the probability of ruin. after which you will only get $1 back at a time. Even after just 10 iterations, your probability of ruin is 99.9%. Stochastic processes that have a continual non-zero probability of extension, do become extinct given enough time, even though the expected value of the new population on each iteration is a higher value. There are many different types of probabilistic convergence. What would you consider the formula for the long run expected value of this process?

This is why financial experts advise others to "diversify". Remember the central limit theorem. When you average independent variables, you reduce the variance, and therefore the value at risk, and your probability of ruin are also reduced.

(But not to derail the thread - I'm more than willing to talk all about this if you are interested... or anyone for that matter.)

Perhaps, we can create a ptgatsby's get rich scheme thread.

 

[ptgatsby]

Keep in mind, as your networth increases, in both cases your expected rate of return decreases (both options are lousy investments).

But, lets see how the "bet-everything-everytime-on-the-same-coin-strategy" works out over the long run in terms of probability of ruin....

Which was the point - neither are relevent. Pretend, for example, that you have clawbacks to your pension if you have a certain income - flipping the coin once actually becomes weighted towards flipping it because the ROR doesn't have a "zero" return. What if, instead of resetting to 0, it resetted to 0.3* original networth, but you are unable to flip again. Or an absolute number... or in the pension case, a graduated amount dependent on your final networth.

I'm not mathematically astute enough to derive these models, I fear, so it isn't an option for me. I also feel that these models are inherently unsuitable - they work great in defining... in reducing data down to a model... but they rarely convey understanding of how things work. I'd rather see simulations than the created model because I can step through it to understand the options involved.

What would you consider the formula for the long run expected value of this process?

Meaning mine? I'm not sure, actually.

Lemme see, I'd need to know the degree of expected variation, the EV and the degree of margin... or at least, the point at which I would consider myself ruined.

If I was to use that scenario, my initial capital would be (a), taken from (b), the home residence. The relative cost of (a) decreases by 2% per year while the relative value of (b) increases by 2% per year. I take (a) and buy a non-correlated (structurally) asset, which I'll call (c), initially, c=a, but (c) will increase, on average at 2%. I then leverage (c)'s value, maintaining 50% in equity, meaning I now have a loan that is (d), which is initially d=c. Over time, however, the relative cost of (d) drops at 2%, and now c becomes 2c.

That makes sense so far. What is risk of ruin here? I would define it as a completely failure when a+d <= c+b, IOWs, the FV of the loans is greater than the FV of the investments.

So the value of (a+d) decreases at (a+d)/1.02 while (2c+b) increases at the rate of (2c+b)*(1+(0.02+market real return rate))... now I just need to know the variation for c and b... hrmmm. Except, that doesn't account for the cost of carrying (a+d). And that's correlated to the market rate... though I could define the cost of a, b would be floating... Hrmmm. And that depends on outside income, since a+b are a tax writable expense.. which ignores further income being injected... and it matters because excess income is added to the portfolio, changing the relative mix of c and d compared to a and b.

Now I remember why I can't do this I'm going to stick with "I don't know" and I can't even begin to imagine my ability to calculate this with even the most complex of models... I know there are some models that allow you to put in this, but they are simplistic ROR measurements based upon margin and the like, which isn't the whole picture.

Though I certainly don't mind anyone telling me how I could do this mathematically, I'm pretty sure I wouldn't be able to understand it, making it rather useless for working things out on my own

Perhaps, we can create a ptgatsby's get rich scheme thread.

Scheme? *shrug* Simplify the problem if the leverage bothers you.

Do you pay off your mortgage or carry an interest-only long term loan?

Do you invest the money that would be in your mortgage in the market?

That's what I'm trying to determine. If it also applies to investments... then why own investment property for those advantages if I can do it through leverage? And then, if that is true, how is ROR affected by having two scenarios running at the same time.

The only way I know to do that is to model it; I'm not sure of the correlations between housing and interest rates, housing and markets, interest rates and markets and so forth... so to model, using historical data seems the best to me.

 

[reason]

I take it this prevents you from hitting on women...?

No. That isn't what keeps me from hitting on women.

 

[reason]

Got it. Tautologies and definitions are true by definition. But do you really make this type of decision on a day-to-day basis?

Oh, my position regarding this is incredibly complicated and difficult to understand. I want to respond in full, but then realise how long it would take!

That's why I am giving these intriguing, but flippent, little responses.

 

[ygolo]

Oh, my position regarding this is incredibly complicated and difficult to understand. I want to respond in full, but then realise how long it would take!

That's why I am giving these intriguing, but flippent, little responses.

Perhaps you can attempt bigger chunks?

 

[reason]

Perhaps you can attempt bigger chunks?

I'll think about it.

 

[Zergling]

I do too, usually. Is this a good thing? Would you like to improve?

I have become better about remember who told me what, so that I can start knowing how well to trust my sources over time.

But in trickier situations, I do have too look beyond the surface and theorize and test. I'm sure you do too. Consider when you have to "fix" something. The first thing I do is to check if the thing is actually broken (I have actually that in most cases when someone asks me to fix something, it is simply in a weird mode, not actually broken).

I haven't really thought about the exceptions yet to how I view things. For the example of information learned from people, I tend to view things like science data, dates, addresses, etc. as trustable. Political information, or information that has direct benifits that might encourage people to fudge it, I tend to take with a partial grain of salt, unless I hear it from a number of different sources, where I take it with less of a grain of salt. Information that I have no need to act on, or is outside any way to directly test, I tend to not view as true or untrue, I simply build up information and have an opinion ot fall back on if I need to decide what I think about the information.