Author Topic: position sizing  (Read 28445 times)

Ice77

  • Newbie
  • *
  • Posts: 42
Re: position sizing
« Reply #20 on: February 09, 2021, 03:57:12 PM »
I've dabbled with Kelly mostly as a tool to determine relative allocation between competing ideas (could also just use expected value which tends to correlate reasonably well with the kelly allocation and is easier to remember). Qualitatively have been guided simply by this dictum - size it sufficiently big that it makes a difference to your life but not so big that you lose sleep over it.


Mohammed Al Alwan

  • Newbie
  • *
  • Posts: 29
Re: position sizing
« Reply #21 on: February 10, 2021, 12:43:08 AM »
Agree with all of the points mentioned above.There was a reading in CFA level three that really changed the way I approach my personal portfolio  as well.The reading was about asset allocation for human and financial capital .The idea is that early in your life most of your net worth is in human capital which are cash flows distant in the future and your financial capital is little.

So, if you consider your financial capital allocation only ,you might be underinvested and not concentrated as you may think. It touched on the correlation between human capital and financial capital. for-example ,you work in wall street and most of you pay is variable and equity like, by concentrating say on financial stocks you are actually more concentrated than you think because at worst you may lose your job (human capital) and your financial capital (stocks go down ).i think now it's being revised under risk management for individual portfolios worth a read .

wabuffo

  • Hero Member
  • *****
  • Posts: 763
    • Twitter
Re: position sizing
« Reply #22 on: February 10, 2021, 07:29:07 AM »
I read this book written by two NY Times reporters years back called "Blind Man's Bluff" about the Cold War that involved true stories about submarine espionage.  One of the stories involves the USS submarine Scorpion that disappears in 1968 to the bottom of the Atlantic with all hands lost.  This sets off a frantic search by the US Navy to find its lost sub.  All they knew was its last reported position, the path it was on (it was heading back to base at Newport News, VA after a tour of duty in the North Atlantic) and only vague other bits of information.  The area to be searched was a large part of the North Atlantic near the US coast.  It was kind of hopeless that the sub would ever be located.

https://en.wikipedia.org/wiki/USS_Scorpion_(SSN-589)

So the Naval Officer in charge of the search operation (a kind of Hunt for Red October Jack Ryan type of guy, I guess) comes up with a novel plan.  He doesn't just reach out to experts like other submarine commanders.  He assembles a wide range of folks - some with submarine knowledge, mathematicians, salvage ops men - basically a diverse set of knowledgeable people but not all experts in subs.  He briefs them with all the information and data that the Navy has related to the USS Scorpion's last voyage.  He then asks them to go off on their own, sift through the data and independently offer their best opinion on why the submarine ran into trouble, its speed and its steepness of descent so as to locate where it might have touched bottom.  He makes it interesting by offering a reward and prizes to the winner who comes closest to the actual location if/when the sub is located.

The individuals' guesses were assembled on a map and using some fancy math (Bayes Theorem) -- a composite guess of the "crowd" was isolated based on the collective estimate of the group as to where the sub might be located.  The location was not a spot any individual member came up with or picked.  But a search was started focusing on the spot identified by this collective method and five months later the sub was located within 220 yards from where the group's estimate said it would be.

Ok - why tell this story.  Because collective wisdom is what the stock market operates on.  It is why it is generally an efficient market.  The various participants individually all have guesses about the fair value of a stock.  These guesses are made up of lots of random guesses + a tiny bit of signal information in each guess.   As these guesses are aggregated, the random parts cancel each other out and what remains is mostly pure signal.

I long ago gave up on using the Kelly Criterion.  That is because according to the Kelly Criterion the size of the bet is determined by the formula:  Edge/Odds = size of bet.  But here's the thing - according to the formula if your Edge = zero, then the size of bet is "Don't Bet!"

If most of us are truly honest with ourselves, do we really have an edge in picking an individual stock vs the market.  Do we think our guess as to where the "missing sub" is located is going to be more accurate and correct than the collective wisdom of a diverse group of actors with money on the line?   Are we really just punters when we think we are experts?

The good news is that unlike most gambling games (which are negative sum), the stock market is a positive sum game.  All we need is to go with the market via diversification in a group of 10-15 high quality businesses and we will do fine.  So my recommendation is to put away the Kelly position sizing stuff because if you think you have an edge then you are making big bets that you alone can locate a "missing sub" better than the market can.

wabuffo

p.s. the USS Scorpion story is also featured in James Surowiecki's "Wisdom of Crowds" book.
« Last Edit: February 10, 2021, 09:33:12 AM by wabuffo »

BG2008

  • Hero Member
  • *****
  • Posts: 1967
Re: position sizing
« Reply #23 on: February 10, 2021, 07:34:47 AM »
I read this book years back called "Blind Man's Bluff" about the Cold War that involved submarine espionage based on true stories written by two NY Times reporters.  One of the stories involves the USS submarine Scorpion that disappears in 1968 to the bottom of the Atlantic with all hands lost.  This sets off a frantic search by the US Navy to find its lost sub.  All they knew was its last reported position, the path it was on (it was heading back to base at Newport News, VA after a tour of duty in the North Atlantic) and only vague other bits of information.  The area to be searched was a large part of the North Atlantic near the US coast.  It was kind of hopeless that the sub would ever be located.

So the Naval Officer in charge of the search operation (a kind of Hunt for Red October Jack Ryan type of guy, I guess) comes up with a novel plan.  He doesn't just reach out to experts like other submarine commanders.  He assembles a wide range of folks - some with submarine knowledge, mathematicians, salvage ops men - basically a diverse set of knowledgeable people but not all experts in subs.  He briefs them with all the information and data that the Navy has related to the USS Scorpion's last voyage.  He then asks them to go off on their own, sift through the data and independently offer their best opinion on why the submarine ran into trouble, its speed and its steepness of descent so as to locate where it might have touched bottom.  He makes it interesting by offering a reward and prizes to the winner who comes closest to the actual location if/when the sub is located.

The individual's guesses on a map were assembled and using some fancy math (Bayes Theorem) -- a composite guess of the "crowd" was isolated based on the collective estimate of the group as to where the sub might be located.  The location was not a spot any individual member came up with or picked.  But a search was started focusing on the spot identified by this collective method and five months later the sub was located within 220 yards from where the group's estimate said it would be.

Ok - why tell this story.  Because collective wisdom is what the stock market operates on.  It is why it is generally an efficient market.  The various participants individually all have guesses about the fair value of a stock.  These guesses are made up of lots of random guesses + a tiny bit of signal information in each guess.   As these guesses are aggregated, the random parts cancel each other out and what remains is mostly pure signal.

I long ago gave up on using the Kelly Criterion.  That is because according to the Kelly Criterion the size of the bet is determined by the formula:  Edge/Odds = size of bet.  But here's the thing - according to the formula if your Edge = zero, then the size of bet is "Don't Bet!"

If most of us are truly honest with ourselves, do we really have an edge in picking an individual stock vs the market.  Do we think our guess as to where the "missing sub" is located is going to be more accurate and correct than the collective wisdom of a diverse group of actors with money on the line?   Are we really just punters when we think we are experts?

The good news is that unlike most gambling games (which are negative sum), the stock market is a positive sum game.  All we need is to go with the market via diversification in a group of 10-15 high quality businesses and we will do fine.  So my recommendation is to put away the Kelly position sizing stuff because if you think you have an edge then you are making big bets that you alone can locate a "missing sub" better than the market can.

wabuffo

p.s. the USS Scorpion story is also featured in James Surowiecki's "Wisdom of Crowds" book.

Wabuffo,

I think you're pretty good with this Garret Motion stuff.  So I think you are underselling yourself a bit. 

writser

  • Hero Member
  • *****
  • Posts: 2279
Re: position sizing
« Reply #24 on: February 10, 2021, 08:01:19 AM »
I actually think the Kelly Criterion can be useful to play around with a bit if you have never done that - just to get a feel for the theory of position sizing and to compare relative position sizes (i.e. why do I have a 5% position in Google but only a 0.5% position in Microsoft even though my thought process about both names is about the same), but I agree that in practice in the stock market its application is limited and I hardly ever use it. l I think for me there are three prime drivers of position sizing:

1. Your personal situation (age, net worth, do you have a job, etc). If you are young and have a good job you can afford to take some risks. It doesn't work out? Write it off as a life lesson, worst case. But if you are a relatively poor retiree? You are fucked.
2. (downside) Risk. What happens if your thesis does not work out as expected? Are you buying a cash box or a growth story at 200x price / sales? E.g. "never lose money". I'd be extremely hesitant to put a large amount of money in a warrant or equity stub or something with an equal risk profile.
3. Level of confidence - the most tricky one. How sure are you that this is a good idea? Do you think you can correctly estimate all tail risks? What's the chance that you are missing an important piece of the puzzle? Can you point out what mistake others are making in their valuation? Why does this opportunity exist? What would you do if the stock cratered 50% today?

The big problem with number 2 and especially number 3 is that they are very hard to quantify. A skeptical view of your own abilities is required. Often when I see people making huge bets / running concentrated portfolio's I feel like they are overestimating their own abilities. In that case Kelly isn't going to help you either because you will always overestimate your own edge and underestimate the risks.

I guess for me it often comes down to: buy as much as you feel 'comfortable with' - and probably a bit less. I know, that's a vague concept, but if you are thinking all the time about a specific position or are anxious something will go wrong it's probably too big. And (at least in my case) very often plans and ideas turn out to be shit, no matter how confident you are initially. So even if you think: I would be totally comfortable with this being a 5% position, why not start with 3%? Often the exact moment of buying is not that time-critical anyway. So you can take it easy. Nibble a bit, take a break.
« Last Edit: February 10, 2021, 08:09:41 AM by writser »
I'm sorry if I have offended you. Please contact this forum's safe space coordinator to work things out.

@thewritser

cherzeca

  • Guest
Re: position sizing
« Reply #25 on: February 10, 2021, 08:20:32 AM »
I think if you do have a large position in a stock, you should have already decided upon an exit strategy. holding forever like Buffett works fine if you own insurance companies that provide you float. the longer you hold a large position the more likely something will go wrong with the stock or the market, and the pain is magnified with a large position. a bull market that increases over time can lull one into complacency, and it is dangerous to be complacent with a large position.

ourkid8

  • Lifetime Member
  • Hero Member
  • *****
  • Posts: 1108
Re: position sizing
« Reply #26 on: February 10, 2021, 08:30:30 AM »
I totally disagree, you should only sell when the fundamentals change.  If the business continues to strengthen it's moat and it's fundamentals keep on increasing, why would you sell? I have held CN Rail for well over a decade and plan to hold it for the rest of my life. 

I think if you do have a large position in a stock, you should have already decided upon an exit strategy. holding forever like Buffett works fine if you own insurance companies that provide you float. the longer you hold a large position the more likely something will go wrong with the stock or the market, and the pain is magnified with a large position. a bull market that increases over time can lull one into complacency, and it is dangerous to be complacent with a large position.
ATCO TSU FFH CNR RY FB WFC

LC

  • Hero Member
  • *****
  • Posts: 5588
Re: position sizing
« Reply #27 on: February 10, 2021, 08:34:26 AM »
Quote
the longer you hold a large position the more likely something will go wrong with the stock or the market, and the pain is magnified with a large position

The opposite holds true as well, though. Good companies + time is a recipe for success. Of course, identifying good companies is difficult and needs to be consistently re-evaluated as ourkid8 mentioned.
"Lethargy bordering on sloth remains the cornerstone of our investment style."

coc

  • Sr. Member
  • ****
  • Posts: 331
Re: position sizing
« Reply #28 on: February 10, 2021, 11:10:46 AM »
I long ago gave up on using the Kelly Criterion.  That is because according to the Kelly Criterion the size of the bet is determined by the formula:  Edge/Odds = size of bet.  But here's the thing - according to the formula if your Edge = zero, then the size of bet is "Don't Bet!"

If most of us are truly honest with ourselves, do we really have an edge in picking an individual stock vs the market.  Do we think our guess as to where the "missing sub" is located is going to be more accurate and correct than the collective wisdom of a diverse group of actors with money on the line?   Are we really just punters when we think we are experts?

The good news is that unlike most gambling games (which are negative sum), the stock market is a positive sum game.  All we need is to go with the market via diversification in a group of 10-15 high quality businesses and we will do fine.  So my recommendation is to put away the Kelly position sizing stuff because if you think you have an edge then you are making big bets that you alone can locate a "missing sub" better than the market can.

wabuffo

p.s. the USS Scorpion story is also featured in James Surowiecki's "Wisdom of Crowds" book.

Serious question, if you feel you have no 'edge' (which is a quantitative concept i.e., better odds than average), why are you picking stocks at all? There would be no point.

Mohammed Al Alwan

  • Newbie
  • *
  • Posts: 29
Re: position sizing
« Reply #29 on: February 10, 2021, 11:30:39 AM »
I read this book written by two NY Times reporters years back called "Blind Man's Bluff" about the Cold War that involved true stories about submarine espionage.  One of the stories involves the USS submarine Scorpion that disappears in 1968 to the bottom of the Atlantic with all hands lost.  This sets off a frantic search by the US Navy to find its lost sub.  All they knew was its last reported position, the path it was on (it was heading back to base at Newport News, VA after a tour of duty in the North Atlantic) and only vague other bits of information.  The area to be searched was a large part of the North Atlantic near the US coast.  It was kind of hopeless that the sub would ever be located.

https://en.wikipedia.org/wiki/USS_Scorpion_(SSN-589)

So the Naval Officer in charge of the search operation (a kind of Hunt for Red October Jack Ryan type of guy, I guess) comes up with a novel plan.  He doesn't just reach out to experts like other submarine commanders.  He assembles a wide range of folks - some with submarine knowledge, mathematicians, salvage ops men - basically a diverse set of knowledgeable people but not all experts in subs.  He briefs them with all the information and data that the Navy has related to the USS Scorpion's last voyage.  He then asks them to go off on their own, sift through the data and independently offer their best opinion on why the submarine ran into trouble, its speed and its steepness of descent so as to locate where it might have touched bottom.  He makes it interesting by offering a reward and prizes to the winner who comes closest to the actual location if/when the sub is located.

The individuals' guesses were assembled on a map and using some fancy math (Bayes Theorem) -- a composite guess of the "crowd" was isolated based on the collective estimate of the group as to where the sub might be located.  The location was not a spot any individual member came up with or picked.  But a search was started focusing on the spot identified by this collective method and five months later the sub was located within 220 yards from where the group's estimate said it would be.

Ok - why tell this story.  Because collective wisdom is what the stock market operates on.  It is why it is generally an efficient market.  The various participants individually all have guesses about the fair value of a stock.  These guesses are made up of lots of random guesses + a tiny bit of signal information in each guess.   As these guesses are aggregated, the random parts cancel each other out and what remains is mostly pure signal.

I long ago gave up on using the Kelly Criterion.  That is because according to the Kelly Criterion the size of the bet is determined by the formula:  Edge/Odds = size of bet.  But here's the thing - according to the formula if your Edge = zero, then the size of bet is "Don't Bet!"

If most of us are truly honest with ourselves, do we really have an edge in picking an individual stock vs the market.  Do we think our guess as to where the "missing sub" is located is going to be more accurate and correct than the collective wisdom of a diverse group of actors with money on the line?   Are we really just punters when we think we are experts?

The good news is that unlike most gambling games (which are negative sum), the stock market is a positive sum game.  All we need is to go with the market via diversification in a group of 10-15 high quality businesses and we will do fine.  So my recommendation is to put away the Kelly position sizing stuff because if you think you have an edge then you are making big bets that you alone can locate a "missing sub" better than the market can.

wabuffo

p.s. the USS Scorpion story is also featured in James Surowiecki's "Wisdom of Crowds" book.

the problem is  for the wisdom of the crowed to operate in stock market you need diversity and independence which usually breakdown at market extreems.