Re: Bill Reid, Kelly Criterion


"Bill Reid" <hormelfree@xxxxxxxxxxxxxxxx> wrote in message
news:frCyj.694320$kj1.312893@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

<GoldenGemNetwork@xxxxxxxxx> wrote in message
news:591d6fc7-789b-4ab0-8873-aa840bf381d4@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

OK I WILL NOT NEED TO MEET YOU IN PERSON, BUT YOU CAN PERHAPS
UNDERSTAND THAT I GET CONFUSED BY THE >>>>>>>>> FOR EXAMPLE
I DON'T KNOW IF YOU OR I HAVE SAID SOMETHING WITHOUT COUNTING
WHETHER A STRING OF >>>>> HAS AN EVEN OR ODD NUMBER OF >.

Yeah, what'd I say...this is just like being in a bar! Bars are noisy,
you
can't hear what the other person is saying, SO YOU WIND UP SHOUTING
AND NOBODY UNDERSTANDS ANYTHING THAT IS SAID!!!

WHAT??!?!!??

I SAID "NOBODY UNDERSTANDS ANYTHING THAT IS SAID!!!"

WHAT??!!!?!!???

I SAID "NOBODY UNDERSTANDS ANYTHING THAT IS SAID!!!"

OH, YEAH, I THINK THIS BAND SUCKS TOO!!!

WHAT???!!!??!!

HENCEFORTH, THEREFORE, I HAVE DELETED THE MOST RECENTLY
INSERTED LAYER OF > AND HAVE DECIDED TO PUT ASTERISKS * BEFORE
MY OWN POSTS.

WHAT??!!?!!!??

Hi Bill Reid,

Can we meet each other in person????

uh, no...why??? Are you named "Michael"?

This newsgroup stuff confuses me.

It's exactly like meeting somebody "in person", except "in
writing"...

*******************************************************
* OK now that I have decided to use asterisks.*
******************************************************

Well, at least you're not "top-posting"...

I think you understand a lot about investing....... what state are you
in anyway?

According to "Lowbrow", I live in "Oakland CA"...of course he's rarely
right about anything and is a pathological liar, so your guess is as
good as his (or probably better!)...

I have just had a sort of terrible experience, I built a house on some
land, and someone bought the road to the house, and won't let me even
walk on it. What do you think of that?

Well, now I know what "state" YOU live in...sounds like New Jersey
to me...

It is the valid concept of
landlocking someone's land here in England. Damn, it is legal for them
to do it.

Remind me never to live in England, then. They do the same thing
in the US, though, they just call it "property taxes"...

****************************************************************************
**********************************************
*Yes, correct, so at least that land is not losing money you correctly
guessed we do not have them here.*

****************************************************************************
**********************************************

Hmmmm, interesting...property taxes are the primary way that
municipalities raise money for fire, police, schools, etc., in the
US...and frankly, though an imperfect taxation system (like all
taxation systems), they seem like a rational way of "fairly" charging
for the "infrastructure" that makes so-called "private property"
ownership possible...

Anyway, not making much sense am I.

Well, it's a crazy world, and if you describe it accurately, it makes
YOU sound like a nut!

It would be nice to find you wherever you are, and have a few drinks,

Again, I don't get it...are you a better mathematician when you're
drunk? Do you think more clearly? I really don't drink that much,
mostly just a single beer (ONE!) on our weekly (most weeks)
"pizza night", and anything more is just on "special occasions"...

*************************************************************************
*Hi, your writing below has convinced me to stop drinking now,*
*I just had been looking for an escape from this financial mess*
*with the land.... I'll follow your example from now on.
*
*************************************************************************

Well, I'm sorry about your problems...where I ACTUALLY live
(NOT "Oakland CA") property sales have slowed down quite a
bit but prices really haven't declined much unlike many other
areas hit by the "sub-prime crises", because of the strong local
economy and the fact that even with the average home price
being upwards of $1million that is just "pocket change" for
a LOT of the people who live here...

It also reminds of an incident working for this failing flailing
dot-com a few years back during the whole dot-com "boom"...I've
always had an ability to walk into ANY company and be able
to tell you within mere hours or at most days just how long
the company will be in existence (many years ago I walked
into what was considered to be the most successful, fastest
growing company in the world at the time, a "household
name" that was lauded in best-selling books as the one
of the best companies in history, and calmly announced
they would be bankrupt in two years...and I was essentially
right almost to the minute).

You see, it's all about the people in the company, and most
importantly the MANAGEMENT of the company (sorry, fans
of "democracy", all people are NOT born "equal", BY A LONG
SHOT), and how MANAGEMENT either leads or doesn't
lead their employees to profitable, succesful, productive
behavior. So I'm in a meeting with a "manager" and a group
of "subordinates" in a company that I've already explicitly
stated would "padlock the doors in about a year" (which was
again 100% accurate, as about a year later they actually did
physically lock out all their employees as they came to
work one day). And this was a very "lively" meeting (if
you know what I mean, which you probably don't), in which
all kinds of "decisions" were being made on forward "direction",
but they were WAY off-track on some known and demonstrable
technical issues, so I interjected though the "hub-bub" that
they were making some egregious technical mistakes to
support their "decisions".

And the "manager", who literally was flush with excitement (he
was laughing and red-faced with the ruddiness actually extending
all the up to the top of his completely bald head!), interrupted me
with these words of profound "pity" for me:

"I'm sorry, we've got this one covered, but if you want to talk about
it
meet me at a bar after work and we can discuss it over drinks."


****************************************************************************
**
* OK OK I'm
convinced!
*

****************************************************************************
**

Actually, drinking is not THAT big of a problem, being an excitable
unprofessional idiot is the real point of that story...

I think they actually fired him for "cause" a few months later, but
it hardly mattered, because as the Italians like to say, "The fish
stinks from the head", and although the guy in the corner office
WAS the smartest "manager" in the whole company, he was
a TOTAL wash-out and blew through HUNDREDS OF $MILLIONS
in financing to acheive abject failure in a RAPIDLY GROWING
MARKET, STILL JUST ABOUT THE MOST LUCRATIVE BUSINESS
OPPORTUNITY I KNOW FOR PEOPLE WHO CAN "MANAGE"
IT PROPERLY (which virtually nobody on this planet is apparently
capable of doing).

and talk about the real meaning of investing,

Is this "Akash"? The REAL "meaning of investing" is to make
"money", NUF SED!!!

and finally show you
some of my programs that you refuse to look at, just so you can
appreciate the graphics, and tell me how terrible they are.

Why do you think they are terrible? As I told you, I have to exercise
a minimum amount of caution in this world in general, as it is
demonstrably
full of "Lowbrow"s who are "up to no good"...so either "discuss" it
here, or forget it...

Here, tell me how "terrible" this totally non-graphic portion of my
trading program is:

unsigned reallocate_new_category(unsigned new_category_num) {
unsigned user_option;
download_category *category=download_init.categories;
void *move_destination;
size_t move_size;

REALLOCATE_CATEGORIES :

if((category=realloc
(category,sizeof(download_category)*(download_init.num_categories
+1)))
==NULL) {

user_option=alert_user(ALERT_ERROR,memory_failure_menu,
"Could not reallocate memory for new download category");

if(user_option==RETRY) goto REALLOCATE_CATEGORIES;

else return FALSE;
}
else download_init.categories=category;

if(new_category_num<download_init.num_categories) {

category+=new_category_num;
move_destination=category+1;

move_size=
(download_init.num_categories-new_category_num)*
sizeof(download_category);

memmove(move_destination,category,move_size);
}

category=download_init.categories+new_category_num;

category->menu_descrip[0]=NULCHAR;
category->target_path[0]=NULCHAR;
category->schedule=0;
category->category_num=new_category_num;
category->automatic=0;
category->num_sources=0;
category->sources=NULL;

download_init.num_categories++;

return TRUE;
}

***********************************************************
* What language is that? *
*
***********************************************************

Good old-fashioned "C"...I tend to write in "C", "C++", with
an unwelcome amount of "Object Pascal" forced upon me
against my wishes due to the goofy development package
I use at home to write the "GUI" versions of my applications.

Here, I'll do a little of my own critique...first thing I notice
is this:

REALLOCATE_CATEGORIES :

if((category=realloc
(category,sizeof(download_category)*(download_init.num_categories
+1)))
==NULL) {

user_option=alert_user(ALERT_ERROR,memory_failure_menu,
"Could not reallocate memory for new download category");

if(user_option==RETRY) goto REALLOCATE_CATEGORIES;

else return FALSE;
}
else download_init.categories=category;

This is of course a garden-variety "safe reallocation" routine, which
every "C" programmer and his brother (and me!) has written a generic
library function to handle...but for some reason I'm still doing the
whole smear manually, when I could have just called my library
function:

if(!reallocate_memory
(category,sizeof(download_category)*(download_init.num_categories+1))
return FALSE;

Which does the whole thing above using a temporary memory pointer
and calling my user_alert() library function with a timed "quit
applications
and retry or abort" menu on failure...also, it is generally considered a
"bad idea" in "C" to use ALL-CAPS for "goto" destination labels since
ALL-CAPS "should" be reserved for MACROS only to avoid confusion...

Likewise, all of this:

if(new_category_num<download_init.num_categories) {

category+=new_category_num;
move_destination=category+1;

move_size=
(download_init.num_categories-new_category_num)*
sizeof(download_category);

memmove(move_destination,category,move_size);
}

category=download_init.categories+new_category_num;

category->menu_descrip[0]=NULCHAR;
category->target_path[0]=NULCHAR;
category->schedule=0;
category->category_num=new_category_num;
category->automatic=0;
category->num_sources=0;
category->sources=NULL;

...is also covered by a generic library routine I wrote for inserting new
"list" items (with a "callback function" for performing the
initialization),
which itself actually calls the "safe reallocation" library function, so
the whole thing could have been reduced to about three lines of
code (and "C++" programmers playing along at home will note that
ALL of those library routines that I wasted time writing and using
are part of the "STL" list "templates", so would correctly note that
I am just ridiculously reinventing the wheel here).

So in a nutshell, IT SUCKSSSSSS!!!

I only bring this up because yesterday I was looking at my whole
"strategy" for configuring automatic data downloading from the
Internet,
thinking if I could make it a little more "user-friendly" (I'm the
only
user,
and I get confused!), and tie it in more directly and automatically
to my target database...but mostly I just sat there and couldn't
come up with any brilliant new ideas but I did know if I "touched"
anything I would most certainly break it and I'd be tied up fixing
it for the next two days...

As far as "graphics" in general goes, I hardly use them, as I've said
I make my decisions mostly based on discrete numbers...I DO have
VERY interesting charting routines in my "Utilities" module...they
are "interesting" because I can "chart" ANY kind of DATA in my
database (overlay long-term stock price with fundamental growth
data, for example), and sometimes the results are startling and VERY
illuminating, but rather than rely on a "pretty picture", I use some
form of correlation coefficient to come up with a "hard" value that
tells me just how truly related two or more datasets are...

Other than that, I also have a utility that can plot out the "Lorenz
attractor", which looks really neat, but again, I'm less interested in
a graphical display than in a single number that encapsulates the
"chaos" (as opposed to "randomness" or "linearity") of the
dataset(s)...

Life is strange, and at times cruel

MANY (I dare not say most, even if true) people are born cruel,
and only "understand" the "law" of greater cruelty to "keep them
in line"...life WILL appear to be eternally "strange" to anybody who
is born with REASON who can't grasp this simple fact because of
the natural tendency to project your outlook on life on other
people...


****************************************************************************
*****
* That is so very interesting. I have heard that before,
from *
* my carer/partner Dimitra, and never believed it. she said
my *
problems getting the land stolen were because of my
passivity *
and assuming people are good. I told people everything so of
course they could anticipate my plans. Being foreigners we're
not liked, and when we moved into a house she told me
she sensed the underlying meaning of people saying 'Do
you plan to live here long? She relates to me telling them,
'Yes, I we plan to live here forever' just to relish watching
their hearts sink. She explains to me, you can't take
things personally as you do. You have to realize when
* people are your enemy, and go on the attack against them.
*

****************************************************************************
*****

"It's a dog eat dog world, and I'm wearing Milk-Bone(TM) shorts."
- NORM!!!

OK, I'm kinda getting an idea here you may not be a native
English speaker and probably don't understand American idioms
and popular TV show references...so that's why I'm using them.

In any event, there may be legal remedies available to you, but
since the whole thing seems a little crazy to me and I don't know
any specific laws on the subject I don't know what you can do...I
would say that it just sounds like you "bought" property in the
middle of somebody else's property (kind of like me selling a
patch of my backyard to somebody), and although the "trespass"
laws here would prevent you from crossing their property on
foot, likewise they are forbidden to step onto your property
(you could have them arrested if they entered your house)...maybe
you could fly a helicopter in and out (I actually know people who
own large amounts of land not serviced by any public roads, and
they have a landing strip and fly their plane in and out)?

In any event, I did want to clear up one of MY stupid mistakes,
just for the record. Previously, I responded to your incorrect
assertion
about the "Kelly criterion" for a particular probability distribution:

This is the neatness of your Kelly idea. If you have ten dollars,
and a trade gives a 50% probability of leaving you with just 5
dollars, and a 50% probability of leaving you with 20 dollars, Kelly
says it is an even trade.
No, that is the "expected value" of the "trade", and you
calculated it wrong; it is:
(0.5*-0.5)+(0.5*2.0) = -0.25+1 = 0.75
That's a 75% expected value!!! In the words of "RoboCop", I'd buy
THAT for a dollar!!! (Every dollar you "invest" in that would be
"worth"
$1.75.)

Of course, the correct "expected value" of the "bet" is:

(0.5*-0.5)+(0.5*1.0) = -0.25+0.5 = 0.25, or you make on average
25% each bet, or each dollar you bet is "worth" $1.25. STILL a
GREAT DEAL, but not as great as I "calculated"...


****************************************************************************
****
* Bill, take a deep breath. Now relax.

I'm almost always relaxed and generally remember to breathe...

I did say something wrong, but it is not wrong in
the way you thought. I shall explain. Suppose my
total wealth is 10 dollars (actually it is less so this
is being optimistic here).

* A bet of 10 dollars, on a coin
toss *
* where heads leaves you with only 5
dollars, *
* and tails wins you 20, while it has a
positive *
* expected return, has zero log return. It
is *
* an even bet according to the Kelly
criterion. *

Not the "Kelly criterion" I'm familiar with, and you just mis-expressed
the problem again by saying "tails wins you 20", when you really mean
"tails wins you 10" (which is why I made my first mistake calculating
the "expectation")...just HOW are you "calculating" the "Kelly criterion"?
The correct way is to have a "log optimal" value at a certain fraction
of your total wealth "bet", and you're not showing that fraction...

* If you change it so tails wins 21 dollars, the Kelly
criterion *
* says take the bet. If you change it so tails wins only
19, *
* the Kelly criterion says do not take the
bet. *

Show your "work", because none of this makes a lick of sense
to me...

* The thinking is, if you follow such a strategy, of taking
* a sequence of INDIVIDUAL bets where you bet the entirety
* of your wealth,

OK, there's your problem, you're not "searching" for a "log optimal"
wealth fraction, but just betting it ALL...which leads to the question,
"why would you do such a thing?" (Other than the fact that due to
the high expectation of the bet, and the impossibility of "losing it all",
a "maximum boldness" strategy may even itself be "log optimal"...but
of course it really isn't, because as you point out below I think your
wealth-doubling and wealth-halving would be equi-probable so it
is just a giant waste of time. However, this would only be slightly
affected by changing the win payout from $10 to $9, so I still don't
get what your point is.)

with a 50% chance of halving and a 50%
chance
* of doubling your money then half the time you will have
* a lot less money than you started out with, and half the
* time a lot more.

For a single "bet", that is true. And for multiple "bets" of ALL your
wealth, that is true. But over time (DIVERSIFICATION over "time",
or the "long run") by dividing your wealth into the appropriate fractional
units,
you will acheive your "expectation" and will make TONS of money.

Once again, the point is: just like "Kelly" can't "magically" make
money where none can be made (zero or negative "expectation"),
"Kelly" doesn't allow you (actually PREVENTS you) from making
stupidly-sized "bets" and losing money.

According to Bernoulli, human enjoyment
* of their wealth too is logarithmic, so your expected
happiness
* is unaffected by the bet.

Over the long run, as you expressed the bet size and probability
distribution, that would be correct...still not sure what the point is...

* NOW, I SHOULD HAVE SAID (important) that this assumes
* that the 10 dollars you started with is the totality of your
* portfolio (SORRY I FORGOT TO MENTION THAT -- OOPS).

YES, EXACTLY, YOU REALLLLLLY SHOULD HAVE SAID THAT!!!

* If you have 20 dollars, and can invest in two such bets,
* then the expected returns with probability 25%, 50%, 25%
* are 5+5, 5+10, 10+10 which are gains of .5, 1.5, 2 so
expected
* log return is .25 ln(.5) + .5 ln(1.5) + .25 ln(2) . The end
terms
* are negatives of each other and the middle term is positive
* since 1.5 > 1.
*
* If you are allowed to divide your money into pennies, and
* take the same bet (with 50% probability of ending up with
* 1 penny for each bet, and 50% probability of ending up with
2)
* then the Kelly criterion begins to agree with the decision
* according to the ordinary rule of expectation.

Well, yeah, that's what I've been trying to tell you!!!

* Hence a principle which I think would not be too hard to
* prove, that the Kelly criterion degenerates into ordinary
* calculation of expectation for a sufficiently diversified
* portfolio.

!!!

EXACTLY!!! I'VE ONLY SAID THIS LIKE A BILLION TIMES!!!

However, we have to be careful here, and just to be sure we're
clear, when you say "sufficiently diversified" you mean diversified
appropriately according to the "Kelly criterion". This is where
you can HURT your "log optimal" growth by OVER-diversifying
(some market gurus have called this "de-worsification").

Your RETURN (money "won"/money "bet") for any fractional "bet"
size at or below the "Kelly/log optimal" will OVER TIME increasingly
EQUAL your "expectation" (consistently "betting" OVER the "Kelly
criterion "guarantees" you will LOSE your wealth, even WITH a
positive expectation). BUT (and this is the really big important
BUTTTT) our "LOG GROWTH" (total increase of your total wealth)
OVER TIME will be greatest right at the "Kelly criterion" wealth
"bet" fraction.

Get it? There's a few things to keep in mind there, but they're
all equally important to grasp...

* The only time the Kelly criterion gives a surprisingly
* divergent answer from ordinary expectation is when
* someone is trying to put too many eggs into too few
* baskets.

Uh, I've tried to express it as succintly as possible above,
and I'm not sure what value you ascribe to "surprisingly", so
maybe we're on the right track here anyway...

* I hope you will read this and think about this as
* earlier I mentioned diversification, and you answered
* in a superficial way. Here I am making a straightforward
* point about the relationship between the Kelly criterion
* and expectation. I am stating, as a theorem, that
* the Kelly criterion decision agrees with the decision
* based on expectation in the limit as a portfolio
* becomes increasingly diversified.

UP TO A POINT. Again, again I say, "DE-WORSIFICATION" is
actually a REAL thing, acknowledged by market "gurus" AND
"Kelly"...


****************************************************************************
****

I would guess you were attempting to illustrate "coin toss odds"
(heads you win your bet, tails you lose your bet):

(0.5*-1.0)+(0.5*1.0) = -0.5+0.5 = 0

But aside from both of our laughable failures to correctly calculate
the "odds", the important thing for the thread subject of "Kelly
Criterion"
is that the above "coin toss odds" can NEVER deliver a POSITIVE
growth factor, despite your claim that the "Kelly says it is an even
trade" (and that somehow we can "manage" our portfolio using "Kelly"
to transform money-losing trades into overall money-makers).

Again, the "math" to prove this (for me) consists of a "brute" or
"binary" search for the highest (or least negative) Kelly growth
factor
(more computer code if you want it) for a certain "bet" size, but
again the math is less important than being able to "grok" the
math. Understanding why a "coin toss" is a MONEY-LOSER
NO MATTER WHAT is as simple as understanding the following:

WHY WOULD YOU RISK ANY MONEY AT ALL TO MAKE
NO MONEY AT ALL? IT'S PURE STUPIDITY!!! YOU WILL
ALMOST CERTAINLY BE AT A LOSS AT SOME TIME DURING
YOUR INFINITY TRIALS, WITH NO EXPECTATION OF
MAKING MONEY AT INFINITY TRIALS, AND THAT IS
NOT "GROWTH". IF YOU WERE TO BET YOUR ENTIRE
NET WORTH ON A SINGLE TOSS OF A COIN, YOU COULD
LOSE IT ALL, AND THAT CERTAINLY IS NOT "GROWTH"!!!
EVEN IF YOU WON, IF YOU CONTINUED TO BET THIS
WAY, YOU WOULD EVENTUALLY LOSE EVERYTHING,
AND THAT CERTAINLY IS NOT "GROWTH"!!! IF YOU
"CLEVERLY" ONLY BET HALF YOUR NET WORTH ON
EACH TOSS, YOU WOULD ALSO EVENTUALLY LOSE
ALL IT ALL, AND THAT'S NOT "GROWTH" EITHER!!!
NO MATTER WHAT YOU DO, YOU CAN'T "GROW" YOUR
NET WORTH FLIPPING A FLIPPIN' COIN!!!

********************************************************************
* No need to calm down, Bill, I appreciate the *
* sincerity which you are investing in our debate. *

I'm cuckoo for Cocoa-Puffs(TM)!!!

* I have answered the larger point earlier. Here, I will *
* answer your immediate point. If I were to state that
* based on a coin toss, you should receive from me,
* the totality of Warren Buffet's assets if the coin should
* land heads, and Bill Gates' assets if it should land tails,
* would you still say that no matter what you do you cannot
* grow your assets by flipping a coin? *
**********************************************************************

Of course not. Now I just have to wait the infinity of the universe
for THAT deal to come along...

Speaking of great deals, there's a TV game show here called "Deal
Or No Deal" which is very popular primarily because they have a bunch
of naked women on the show. But it is interesting from a probabilistic
standpoint, because they essentially "give" you about $78,000 to
play the game, based on the probabilities of the game (you can
win up to a $1million by guessing where the $1million is in 26
briefcases with amounts ranging from one penny to $1million, but the
average amount in all the cases is about $78,000). So the "strategy"
for playing the game is to take the "deal" when they offer you MORE
than $78,000 to quit the game, because that is all you can reasonably
EXPECT to win in the first place...

THEREFORE,
YOU DON'T EVEN HAVE TO DO THE "MATH" (SAVE THOSE
COMPUTER CYCLES AND THE ATTENDANT "GLOBAL
WARMING"!), BECAUSE "COMMON SENSE" SHOULD
TELL YOU ALL YOU NEED TO KNOW IN THIS CASE.

The bottom line, again: the "Kelly Criterion" is NOT some
magical way to subvert the laws of the universe, it is only a
humble tool that simply reveals the "truth" about those laws
in regard to a "bet" with certain probability characteristics...

OK, I think maybe we've accomplished something here, not
much, but more than what could be done in a bar while drunk...

LOL Looks like you have an admirer Bill. Maybe a stalker even...
I would reach down and do a package check though before inviting her home
for a drink...
20-yr. ago it would seem like a rude act of groping.. now a days it is
considered playing it safe..


.