|
Blinky2099 posted:Below what "average"? See my post above. Sequencing risk of returns is compatible with a 100% random stock market, and sequencing risk is a big reason you benefit from a low/negative return stock market during the first half or so of your retirement savings accumulation years. Its not worth getting into but you are also ignoring the fact that valuations (P/E, EV/EBITDA) drop and dividend yields rise all else equal when the stock market drops. Swingline fucked around with this message at 01:08 on Nov 10, 2016 |
#
?
Nov 10, 2016 01:05
|
|
|
|
| # ? May 27, 2024 06:14 |
|
|
Swingline posted:See my post above. Sequencing risk of returns is compatible with a 100% random stock market, and sequencing risk is the reason you benefit from a low/negative return stock market during the first half or so of your retirement savings accumulation years. Swingline posted:First to clarify, I don't advocate market timing. You should invest your retirement savings immediately as you get paychecks/other income, rather than save up a bunch of cash waiting for a dip. Swingline posted:2) The S&P 500 returns 8.16% after inflation per year for the first 15 years. Using this flawed logic, I'll create a more extreme example to show my point: 4) You predict an average of 4% annual gain over 30 years. The S&P 500 returns 0% after inflation per year for the first 29 years. Now we need to make a decision with our investments. Are you going to predict a 324% return on the last year (30 years of 4% gain = 324% return with compounding gains) to match your initial assumption of 4% annual gains? Of course not. Blinky2099 fucked around with this message at 03:30 on Nov 11, 2016 |
|
#
?
Nov 10, 2016 01:11
|
|
|
Blinky2099 posted:This is a very, very wrong/bad analysis. I'll explain why. There's no predictions involved. All that you have to assume is that 1) the stock market will have a geometric return of X% over a Y year period (doesn't matter what X and Y are) and 2) the annual returns during the Y period will vary around X% - some years higher, some years lower. Given that, you will end up with a much larger dollar amount if the worse years randomly happen to occur towards the earlier part of Y. What part of that involves any predictions?
|
|
#
?
Nov 10, 2016 01:17
|
|
|
To add to the above, you're way too focused on the illustrative numbers and the simple return sequence I threw out to make my life easy. The analysis works the same with nonsensical numbers in a completely random sequence, IE 500% geometric annualized return over 1000 years, year 1 = -55%, year 2 = +1500%, year 3 = -75%, year 4 = +2000%, etc etc for 996 more years. Its just math.
|
|
#
?
Nov 10, 2016 01:24
|
|
|
Swingline posted:There's no predictions involved. All that you have to assume is that 1) the stock market will have a geometric return of X% over a Y year period (doesn't matter what X and Y are) and 2) the annual returns during the Y period will vary around X% - some years higher, some years lower. Given that, you will end up with a much larger dollar amount if the worse years randomly happen to occur towards the earlier part of Y. Swingline posted:To add to the above, you're way too focused on the illustrative numbers and the simple return sequence I threw out to make my life easy. The analysis works the same with nonsensical numbers in a completely random sequence, IE 500% geometric annualized return over 1000 years, year 1 = -55%, year 2 = +1500%, year 3 = -75%, year 4 = +2000%, etc etc for 996 more years. Its just math. If your point is "it's good for the guy who invested when the market went up 8%" then sure, but I don't see why that is relevant. That has absolutely nothing to do with the question: "Is the market going down early in your career good for you?" In order to answer that question, you have to involve predictions. You can't answer that question otherwise. Yes, you can say "market yielded 0% and the guy didn't buy any and then he bought a lot and it went up 8%" but that's not somehow evidence of the 0% years having any effect on the 8% years. Unless you can accurately predict "market yielded 0% and the guy didnt buy any but now we have a reason to believe the market return will be 8% and not 4%", your examples are meaningless. What exactly is your point of giving these examples? In what situation, what is expected to happen? Can you summarize with like a sentence or two? Blinky2099 fucked around with this message at 01:35 on Nov 10, 2016 |
|
#
?
Nov 10, 2016 01:33
|
|
|
Swingline posted:There's no predictions involved. All that you have to assume is that 1) the stock market will have a geometric return of X% over a Y year period (doesn't matter what X and Y are) and 2) the annual returns during the Y period will vary around X% - some years higher, some years lower. Given that, you will end up with a much larger dollar amount if the worse years randomly happen to occur towards the earlier part of Y. Part 1. You're conditioning on the thing you're trying to predict. Blinky2099, to hand-wave towards an answer to your question, here's Bernstein: The Four Pillars of Investing, Chapter 7 posted:What makes recency such a killer is the fact that asset classes have a slight tendency to “mean-revert” over periods longer than three years. Mean reversion means that periods of relatively good performance tend to be followed by periods of relatively poor performance. The reverse also occurs; periods of relatively poor performance tend to be followed by periods of relatively good performance. Unfortunately, this is not a sure thing. Not by any means. But it makes buying the hot asset class of the past several years bad odds. The section containing that is titled "The Immediate Past Is Out to Get You". He's mostly warning against buying recent strong performers, not buying into a down market. But later: The Four Pillers of Investing, Chapter 10 posted:Finally, I can’t help but mention the Morningstar Unpopular Funds Strategy. Since Morningstar is located in Chicago and staffed by a sports-loving crowd, they have a special affinity for losers. Every year since 1987, they’ve used the fund money flows discussed above to identify the most popular and unpopular fund categories. They then follow the average performance of the three most popular and unpopular fund groups forward for three years. Eight out of nine times, the unpopular funds beat the popular funds, and seven out of nine times the unpopular funds beat the average equity fund. Most tellingly, the popular fund categories also lagged the average equity fund seven of nine times. I certainly don’t recommend this as an investment strategy, but it’s an excellent example of the dangers of chasing performance, because of the tendency for asset classes to “mean-revert,” that is, to follow good performance with bad, and vice versa. So I don't have any evidence to show you, but there's the claim. EDIT: Oh hey, I skipped right over all the good stuff from Chapter 1. quote:Paradoxically, in the long run, bonds are at least as risky as stocks. This is because stock returns are “mean reverting.” That is, a series of bad years is likely to be followed by a series of good ones, repairing some of the damage. Unfortunately, this is a two-edged sword, as a series of very good years is likely to be followed by bad ones, as investors have learned, to their chagrin, in the past few years. In Figure 1-13, I’ve plotted the annualized 30-year real (inflation-adjusted) returns of stocks. Note how placid this graph looks... Monokeros deAstris fucked around with this message at 01:41 on Nov 10, 2016 |
|
#
?
Nov 10, 2016 01:36
|
|
|
Blinky2099 posted:I'm having a hard time articulating why you're wrong but here I go. Maybe I'm bad at explaining things on an internet comedy forum, so I just opened up the first book in Bernstein's Investing for Adults series, which has the example I was trying to illustrate: "Uncle Fred is a mythical employer who offers a retirement scheme with a most curious set of annual returns, determined by a coin flip that yields either a +30% or -10% result. This simple paradigm provides a useful shorthand for the long-run behavior of equity investing: Its expected return is 10%, that is, the arithmetic average of +30% and -10%. But the more relevant number for most investors is 8.17%. This is the median return of a fair 50/50 sample of coin tosses, as well as the geometric average of +30% and -10%. Finally, the 20% standard deviation of the coin toss is similar to the volatility of the broad U.S. stock market. Imagine that on December 31 in each of 40 consecutive years, you contribute $1,000 to a retirement account, and that Uncle Fred's coin toss determines what happens to your account. Let's further assume that your yearly returns sequence defies the laws of probability, and you enjoy 20 consecutive years of positive returns first, followed by 20 consecutive years of negative returns. The 40-year result? Your account has an ending value of $159,960. Not too bad, you might say. But if we are calculating in nominal returns, your total return is not far from flat after adjusting your contributions for inflation. Now assume the opposite sequence: 20 consecutive years of negative returns, followed by 20 consecutive years of positive ones. Result? Your account grows to an ending value of $4,307,437. Figure 3 displays the simplest way to understand the superiority of the negative-results-first sequence, in which you bought at significantly lower average prices than with the positive-results-first sequence. If stocks sold for $1.00 per share at the beginning of the negative-results-first sequence, they declined in price to just $0.12 per share after 20 years, before rising over 20 years to $23.11 at the end of 40 years. Factoring the reinvestment of dividends into the share price, your average purchase price over the full 40 years was just $2.69 per share. With the good-results-first sequence, by contrast, the purchase price never fell below $1.00 per share, then rose as high as $190.05 at 20 years before falling back to the same final value of $23.11 at 40 years. The average purchase price over the whole period was $58.04.  There's an even more profound interpretation of why the second sequence was better: The investor had a higher dollar exposure to stocks during the 20 good years with the second sequence (bad returns first) than with the first. One lesson here is that bad returns may not always be bad - if they lead to higher good returns in the long run. But imagine another scenario, in which instead of investing a stream of savings, you have just inherited a lump sum of money at age 25 from Uncle Fred that is large enough to fund your retirement at age 65 if you invest it at the 8.17% return rate. Uncle Fred, of course, has forbidden you to spend any of the money until you reach age 65. In other words, you will neither add to nor withdraw from this investment for 40 years. What could be better? You won't have to save a dime. All you will have to do in the meantime is meet your living expenses with a working salary. Assume that the usual annual coin toss determines your returns. Assume also that the gods of chance treat you fairly, with 20 heads and 20 tails. Does the precise sequence of heads and tails matter with this lump sum? No, for without withdrawals from or additions to the portfolio, the commutative law of multiplication applies...What we have learned here is that investing all of your money up front in stocks avoids "sequence risk." That is, with a lump sum, a particularly good or bad sequence of returns will not affect you final result one bit. Of course, unless you began your adult life as a wealthy heir, you don't have that option. But the corollary still holds. Investing as much as possible in stocks and other risky asses up front maximally mitigates sequence risk. Inheriting money from your Uncle Fred is the optimal scenario - and not solely because you don't have to work and save. It's optimal for a second reason: The lump sum of an inheritance provides a much more level dollar exposure to equities through your investing life cycle than having to work and save, which gives you almost no exposure, relative to the size of your total capital, to stock when you're young."
|
|
#
?
Nov 10, 2016 02:10
|
|
|
I appreciate the post but his example has the exact same issue as yours; it assumes you already know the results. It's trying to explain a concept that is already assumed in this discussion. I get it, you would rather invest all your money first, and then have the market shoot right up. That's literally the only thing that this example gives. It's "market drops 20 years, then you invest most of your money, then market shoots up 20 years, this is a good thing and better than investing after the market has gone up and you receiving 0% returns" which is fairly obvious. That has absolutely nothing to do with "does market dropping 20 years = a predictor of market shooting up the next 20 years". To bring it back to the original point of the discussion, "should 25 year olds be happy when the market tanks and their retirement accounts drop by 20%", that example has no relevance, because we have no knowledge of the next 29 years, or whatever timeline you want to use. We are not some imaginary example where we know it's going to tank for our first 20 years and then shoot up the next 20 years. We know that it has gone down, and have no knowledge of what will happen in the future, thus we have to make a prediction. That prediction can be 7% returns per year, or "it dropped a lot so we expect 8.2% average over the next 10 years rather than 7%", but that prediction needs to have some sort of reasoning behind it. The null hypothesis should be "your future expected returns do not change based on last year's results." The alternative hypothesis then being "last year's results is a good predictor of future expected returns." Edit: Swingline posted:One lesson here is that bad returns may not always be bad - if they lead to higher good returns in the long run. Alhireth-Hotep posted:Blinky2099, to hand-wave towards an answer to your question, here's Bernstein: Blinky2099 fucked around with this message at 02:35 on Nov 10, 2016 |
|
#
?
Nov 10, 2016 02:19
|
|
|
I need some advice. My situation involves being unemployed grad student with no debt and a chunk of savings. ($50,000+) I'm looking at options to grow this money into the beginnings of a retirement fund. Without being able to contribute to an IRA, and being a complete novice to trading, I'm considering doing some laddering with CD accounts ($10k in 5 acounts laddering up to 60mo terms). Would this be a worthwhile use of my savings?
|
|
#
?
Nov 10, 2016 02:43
|
|
|
Most CD rates I see are garbage and don't seem worth the sacrifice in liquidity you could avoid with a 1% savings account. Also if you're talking retirement you probably want to get more into equities. Tax-advantaged doesn't sound possible for you, but you could still mirror a target-retirement fund's allocations with other funds.
|
|
#
?
Nov 10, 2016 02:47
|
|
|
eggyolk posted:I need some advice. My situation involves being unemployed grad student with no debt and a chunk of savings. ($50,000+) I'm looking at options to grow this money into the beginnings of a retirement fund. Without being able to contribute to an IRA, and being a complete novice to trading, I'm considering doing some laddering with CD accounts ($10k in 5 acounts laddering up to 60mo terms). Would this be a worthwhile use of my savings? Looking online at CD rates, they seem to be a little under 2% for a 5 year. You can get 1% in an online savings account. If I were you, I would put the money into an online savings account for the 1% and start buying stocks slowly over time until you are comfortable with the total risk you have. Alternatively, you could buy $10,000 worth of i-bonds per year. I wouldn't normally recommend this, but since you are describing a CD ladder I bring it up only because you might not want to invest in stocks at all. In that case, I think IBonds might be a better choice than CDs.
|
|
#
?
Nov 10, 2016 02:48
|
|
|
So does it help if we assure you that periods of slower than average growth are in fact followed by periods of higher than average growth, and vice-versa? I'm not trying to be a jerk but these discussions with you are really frustrating because you keep insisting on using analogies like coin flips and gambling that are completely inapplicable to the market, which doesn't behave randomly. It does in the short-term, but in the long-term we generally agree it fits a trend line. If you fundamentally disagree with that, then I think D&D is a better place for that because it's more of a debate. But to be clear I'm not a mod of this forum and this is just my personal opinion.
Alereon fucked around with this message at 07:38 on Nov 10, 2016 |
|
#
?
Nov 10, 2016 07:36
|
|
|
Alereon posted:So does it help if we assure you that periods of slower than average growth are in fact followed by periods of higher than average growth, and vice-versa? I'm not trying to be a jerk but these discussions with you are really frustrating because you keep insisting on using analogies like coin flips and gambling that are completely inapplicable to the market, which doesn't behave randomly. It does in the short-term, but in the long-term we generally agree it fits a trend line. If you fundamentally disagree with that, then I think D&D is a better place for that because it's more of a debate. But to be clear I'm not a mod of this forum and this is just my personal opinion. I don't think it was a terrible question, because people are bad at thinking about statistical models, and even if the market's random walk didn't mean-revert, they'd still be getting the (wrong) explanations people keep posting. Now in fact, according to Bernstein, the market does have mean-reverting tendencies -- or stocks do, at least. Bonds, apparently, don't. But most of the posts in response to them have missed the point, made bad statistical arguments in order to assure them that the market must necessarily have mean-reverting tendencies (which is not true; see bonds), or both. It's an empirical question, and nobody here including me has posted any actual evidence that it is true. I'm just quoting an authority I trust.
|
|
#
?
Nov 10, 2016 07:50
|
|
|
Alhireth-Hotep posted:I don't think it was a terrible question, because people are bad at thinking about statistical models, and even if the market's random walk didn't mean-revert, they'd still be getting the (wrong) explanations people keep posting. Now in fact, according to Bernstein, the market does have mean-reverting tendencies -- or stocks do, at least. Bonds, apparently, don't. But most of the posts in response to them have missed the point, made bad statistical arguments in order to assure them that the market must necessarily have mean-reverting tendencies (which is not true; see bonds), or both. It's an empirical question, and nobody here including me has posted any actual evidence that it is true. I'm just quoting an authority I trust. On bonds, though, you know exactly what your yield to maturity is (assuming you are buying government or highly-rated corporate debt.) Of course interest rates/debt ratings/whatever may make that bond worth more or less, but if you plan to buy-and-hold, it's written into the bond.
|
|
#
?
Nov 10, 2016 08:08
|
|
|
Oh I don't think it's bad to ask, I just felt like we had transitioned to a frustratingly circular argument about whether a principle of investing was true. E: And a late edit just to double-clarify that I am not a mod in this forum and any posts I make are my personal opinion and you can ignore them if you want! Alereon fucked around with this message at 17:03 on Nov 10, 2016 |
|
#
?
Nov 10, 2016 08:10
|
|
|
Blinky's question is really about mean-reversion, is it a thing or not. Some people seem to think so but the extent and timeframe are mysteries.
|
|
#
?
Nov 10, 2016 14:47
|
|
|
eggyolk posted:I need some advice. My situation involves being unemployed grad student with no debt and a chunk of savings. ($50,000+) I'm looking at options to grow this money into the beginnings of a retirement fund. Without being able to contribute to an IRA, and being a complete novice to trading, I'm considering doing some laddering with CD accounts ($10k in 5 acounts laddering up to 60mo terms). Would this be a worthwhile use of my savings? My credit union gives you 2.25% on the first 20k in your checking so long as you make one automatic payment or deposit and use the debit card as credit 10 times in a month averaging $5 a purchase. It's a little hoopy/jumpy but 2 and a quarter is a better return than almost any other "zero risk" investment vehicle.
|
|
#
?
Nov 10, 2016 15:27
|
|
|
EAT FASTER!!!!!! posted:My credit union gives you 2.25% on the first 20k in your checking so long as you make one automatic payment or deposit and use the debit card as credit 10 times in a month averaging $5 a purchase. It's a little hoopy/jumpy but 2 and a quarter is a better return than almost any other "zero risk" investment vehicle. my credit union had a similar thing before the '08 crash that was 4 or 5%, which seems unbelievable now in the age of sub-1% savings accounts
|
|
#
?
Nov 10, 2016 16:19
|
|
|
Alereon, hope you don't mind me chiming in here. But we've had this discussion before and Blinky hasn't been able to get the point across so I thought I'd help. Imagine we have this performance and the prediction that follows (arbitrary units):  But then there's a drop. Which of the following is a more reasonable prediction?   That is, what do we hold constant? The future rate of return, or the expected overall rate of return over some period that also includes the past (or equivalently, the expected price at the end of some time horizon)? To me, I agree with Blinky, that the second prediction seems like market timing and even gambler's fallacy. (it also brings in difficult questions like what time horizon holds constant? and do you behave reciprocally when the market does well?) Last time we had this discussion it boiled down whether or not market performance is negatively autocorrelated, and I believe the jury is out among experts. In any case, acting on the belief that markets are negatively autocorrelated certainly is market timing. Which maybe is valid, if you think that (e.g.) the current dip is the market reacting to Trump, but you think that he won't actually damage the economy and the market is being pessimistic. If you believe that then a recovery is likely and you should be happy the market dropped. But you can't argue that doesn't count as timing the market. SurgicalOntologist fucked around with this message at 16:52 on Nov 10, 2016 |
|
#
?
Nov 10, 2016 16:49
|
|
|
I've got a decent amount in an IRA that's been floating around ever since rolling it over after leaving my old job. Now that Trump will be running the show, I figure that it may be a good time to convert it to Roth with all of the talk about low taxes. Am I simply looking for the lowest rate possible on my highest tax bracket, or is there anything additional to keep an eye out for in the coming years?
|
|
#
?
Nov 10, 2016 17:18
|
|
|
Alereon posted:So does it help if we assure you that periods of slower than average growth are in fact followed by periods of higher than average growth, and vice-versa? I'm not trying to be a jerk but these discussions with you are really frustrating because you keep insisting on using analogies like coin flips and gambling that are completely inapplicable to the market, which doesn't behave randomly. It does in the short-term, but in the long-term we generally agree it fits a trend line. If you fundamentally disagree with that, then I think D&D is a better place for that because it's more of a debate. But to be clear I'm not a mod of this forum and this is just my personal opinion. If we're that confident that phenomenon exists then we should also be acting upon it. If we're not really sure whether it's true or not, the default should be that it does NOT exist, and just assume any loss is permanent and will not be "corrected" somehow. Regression to the mean is about your slope regressing (in a linear example) or your ROI regressing (market example), not that the dollar value of the market will magically go back to some setpoint previously created by historical data on top of your usual expected ROI gains. Alereon posted:Oh I don't think it's bad to ask, I just felt like we had transitioned to a frustratingly circular argument about whether a principle of investing was true. The problem is that people are misapplying "regression to the mean" as meaning this: When really, regression to the mean is this. And this type of attitude is dangerous to have without evidence of it being true. SurgicalOntologist posted:Alereon, hope you don't mind me chiming in here. But we've had this discussion before and Blinky hasn't been able to get the point across so I thought I'd help. SurgicalOntologist posted:But you can't argue that doesn't count as timing the market. Edit: The main point of me posting this was acknowledging "I don't know poo poo about the markets and maybe market timing is real and if anyone has any evidence or data it would be really interesting to see". The book copy/pastes were really interesting to see his opinion on it, even though there wasn't really any analysis given. I'm definitely not arguing that it doesn't or cannot exist, just that our first assumption in these situations should always be that it does not exist and that needs to be proven otherwise. I love talking about this sort of stuff which is why I post about it -- not meaning to have obnoxious circular arguments or anything. Blinky2099 fucked around with this message at 18:09 on Nov 10, 2016 |
|
#
?
Nov 10, 2016 17:41
|
|
|
Yeah this makes for some fascinating high level discussion. I think "active value investing" believes in the second type of graph (short term underperformers overperform relative to the market at large) and I know their data shows that certain managers who have long timelines and long levers are able to achieve exceptional results but I don't know that it holds for whole MARKETS. Thus I think index investors are better off NOT timing markets whatsoever, but if you're going to speculate stocks to buy and hold (and you've got billions to do it) you're better off picking relative dogs (2nd runners, etc.) That's the impression I get from Graham and his extensive works and historical data - who knows whether it will hold from this moment forward, however.
|
|
#
?
Nov 10, 2016 20:43
|
|
|
If you predict that, over a finite period (e.g. your life), you will see X% average gains, and then one day the market drops sharply, then yes you should predict greater than X% gains for the period following the drop, because two periods of X% average gains separated by a drop taken all together have lower than X% average gains. I'm not making any statement about whether that first expectation is reasonable, but if you do accept it, then the only consistent position is to accept its implications as well.
Ralith fucked around with this message at 02:39 on Nov 11, 2016 |
|
#
?
Nov 11, 2016 02:36
|
|
|
Ralith posted:If you predict that, over a finite period (e.g. your life), you will see X% average gains, and then one day the market drops sharply, then yes you should predict greater than X% gains for the period following the drop, because two periods of X% average gains separated by a drop taken all together have lower than X% average gains. I'm not making any statement about whether that first expectation is reasonable, but if you do accept it, then the only consistent position is to accept its implications as well. Sure, but I think when people say "over a finite period I expect X% average gains" they really mean, "lacking other information I expect X% gain over any finite period". Or at least they should mean that. If I know the market's value every single day in the period I sure as hell shouldn't expect the same average. If I know it every day except the last, I assert that you also should not expect the same average. Rather, that last day becomes your unknown period over which you expect the baseline X%.
|
|
#
?
Nov 11, 2016 03:03
|
|
|
Ralith posted:If you predict that, over a finite period (e.g. your life), you will see X% average gains, and then one day the market drops sharply, then yes you should predict greater than X% gains for the period following the drop, because two periods of X% average gains separated by a drop taken all together have lower than X% average gains. I'm not making any statement about whether that first expectation is reasonable, but if you do accept it, then the only consistent position is to accept its implications as well. see: Blinky2099 posted:4) You predict an average of 4% annual gain over 30 years. The S&P 500 returns 0% after inflation per year for the first 29 years. Now we need to make a decision with our investments. Are you going to predict a 324% return on the last year (30 years of 4% gain = 324% return with compounding gains) to match your initial assumption of 4% annual gains? Of course not. Prediction: flip a coin 10 times. I predict that heads and tails will both show up 5 times... which is true, on average. Half-way in, and we've seen heads show up 5 times already. The score is 5-0. Now to make our initial prediction correct, we have to assume that tails will most likely show up the next 5 flips...? Of course not, we change our prediction to (5 heads known + 5 flips unknown * [0.5 heads, 0.5 tails]) = 7.5 heads / 2.5 tails. Blinky2099 fucked around with this message at 03:31 on Nov 11, 2016 |
|
#
?
Nov 11, 2016 03:09
|
|
|
Sorry in advance for politics, but Trump is aiming to heavily modify or get rid of the upcoming Fiduciary Standard law for employee benefits/pensions. So I guess that's going to continue to be a thing: https://webcache.googleusercontent....n&ct=clnk&gl=us
|
|
#
?
Nov 11, 2016 03:15
|
|
|
Blinky2099 posted:and it makes absolutely no difference whether you're talking about stocks or the casino
|
|
#
?
Nov 11, 2016 03:36
|
|
|
Alereon posted:I don't think this is a reasonable assumption for you to make at all. Blinky2099 posted:4) You predict an average of 4% annual gain over 30 years. The S&P 500 returns 0% after inflation per year for the first 29 years. Now we need to make a decision with our investments. Are you going to predict a 324% return on the last year (30 years of 4% gain = 324% return with compounding gains) to match your initial assumption of 4% annual gains? Of course not. Blinky2099 fucked around with this message at 03:41 on Nov 11, 2016 |
|
#
?
Nov 11, 2016 03:38
|
|
|
The market is a real thing and recoveries happen though?
|
|
#
?
Nov 11, 2016 03:48
|
|
|
Alereon posted:The market is a real thing and recoveries happen though? You change your predictions after you witness results. If they're not changing, then you're doing something very wrong, no matter the topic. If market timing is real then yes, you can say "the market sucked recently, so instead of 4% gains I'm going to guess 5.5% gains this year". You cannot say "the market sucked recently, but I predicted 4% gains each of those years, so the market is likely to completely make up for the entirety of that the next few years" without being incredibly wrong (and applying a fallacy.) Also, coin flips are "a real thing", gambling is "a real thing", blackjack is "a real thing". Also, the third sentence of the wikipedia article says: quote:This fallacy can arise in many practical situations although it is most strongly associated with gambling where such mistakes are common among players.
|
|
#
?
Nov 11, 2016 03:58
|
|
|
Blink sees a series of stock returns and doesn't really understand what he is seeing and assumes they are random. Other people see stock returns and realize that they are a very complicated result of millions of factors, and while they seem random they really arent, they are just incredibly complicated and based on impossible to predict factors. He is not willing to accept that stock returns are different than the universe rolling dice, so he will never accept any other conclusions beyond that premise.
|
|
#
?
Nov 11, 2016 04:08
|
|
|
Droo posted:stock returns ... while they seem random they really arent, they are just incredibly complicated and based on impossible to predict factors.
|
|
#
?
Nov 11, 2016 04:14
|
|
|
Droo posted:Blink sees a series of stock returns and doesn't really understand what he is seeing and assumes they are random. Other people see stock returns and realize that they are a very complicated result of millions of factors, and while they seem random they really arent, they are just incredibly complicated and based on impossible to predict factors. 29 years of 0% returns. 30th year? I get it, stocks aren't totally random so you can make a prediction that the 30th year you're going to see a higher than 4% gains. You're adjusting for the fact that the previous 29 years have underperformed. This makes complete sense. But you make a new prediction after the 29th year about the 30th year, you do not keep your initial 4% per year prediction to magically be reversed by the 30th year. Initial prediction: 4% per year After seeing 29 years of 0% gain: You still assume 4% gain for the 30th year if the market is truly random. After seeing 29 years of 0% gain. You might assume 4.5%, or 5%, or 8%, or 10% gain for the 30th year if you think your returns will be higher because of recent under-performance. If you kept your initial prediction of 4% per year for 30 years, you would then have to assume a 324% gain for the 30th year. This is completely loving ridiculous even if you believe in market timing and higher gains due to recent losses. You make a new assumption based on previous years in both cases, whether you think it's truly random or "a complicated result of a million factors". You never, ever keep your initial 4% assumption and just blanket apply it to the 30th year to make up for the previous 29 years. Blinky2099 fucked around with this message at 04:27 on Nov 11, 2016 |
|
#
?
Nov 11, 2016 04:16
|
|
|
You don't forecast returns based on previous returns. You forecast returns based on future expectations of profits compared to current price, among other things. The 4% forecast from 30 years ago was irrelevant 29.99 years ago.
|
|
#
?
Nov 11, 2016 04:27
|
|
|
Droo posted:You don't forecast returns based on previous returns. You forecast returns based on future expectations of profits compared to current price, among other things. The 4% forecast from 30 years ago was irrelevant 29.99 years ago. but anyway, your last bit, "4% forecast from 30 years ago was irrelevant 29.99 years ago" is exactly my point, you're agreeing with me. You make new assumptions and predictions immediately. You're not locked into your initial 4% forecast after 5, 10, 20 years (or after 0.01 year like you mention.) Ralith posted:If you predict that, over a finite period (e.g. your life), you will see X% average gains, and then one day the market drops sharply, then yes you should predict greater than X% gains for the period following the drop, because two periods of X% average gains separated by a drop taken all together have lower than X% average gains. I'm not making any statement about whether that first expectation is reasonable, but if you do accept it, then the only consistent position is to accept its implications as well. He is literally saying "because we assumed 4% average returns through our life but recently saw a big market drop, we now have to predict higher gains to make sure we hit that 4% through our life."
|
|
#
?
Nov 11, 2016 04:33
|
|
|
I'll stop making GBS threads up the thread with this since it's more than run its course, but yes, this:Saint Fu posted:My take is he is just looking for some research/analysis/evidence that this is true.
|
|
#
?
Nov 11, 2016 04:40
|
|
|
Blinky2099 posted:It'd be cool to see "if you assume 4% gains but the market tanked recently, there's real evidence to suggest that we can now predict 5.5% gains for the following year" or something. The boglehead posts start getting to this but they're just generic statements that it probably exists but without enough confidence to suggest acting on it. I'm not saying it can't happen, I'm saying any type of example should always start with assuming random chance and then deviate based on evidence. There are lots of reasons the price of a stock can go down. Sometimes the price movement can be unfair, in which case it is fair to expect higher future returns. Sometimes the decline in price can be very fair, in which case it is reasonable to expect lower returns than you originally expected. It is very hard to tell which is the case, but certainly plenty of people make a good living by value investing when the market panic overreaches reality. Over a very long period of time the stock market has gone generally up, so for 200 years it has been fair to expect buying stocks during a market crash to have better returns over a long period of time than buying during a market that is making new highs on a daily basis. You are trying to turn it into a boolean case, where it is either completely one way or the other. It really doesn't work that way. The past performance MIGHT be correlated to future expectations, based on the underlying expectations, but not necessarily. And it is very hard to tell.
|
|
#
?
Nov 11, 2016 04:52
|
|
|
Imagine you have a crystal ball and know the "true" price of the market, x. Traders are good at their jobs so the market tends to price around x, with some small random swings. One day the "true" price drops 10%. What happens next is that traders get stopped out of positions, retail investors panic sell and the market drops 20%. You have no way of telling when this will happen in advance, nor do you know for sure that it is exactly what's happening, but your expected return increases whenever prices drop sharply because of this dynamic. It becomes a lot easier to accept if you realize that markets aren't perfectly rational. If you approach the problem like a card/dice game that is purely probabilistic you are neglecting the impact of all the people involved in the system.
|
|
#
?
Nov 11, 2016 05:24
|
|
|
Problem solved everyone. Stock returns are not negatively autocorrelated at a timeframe of up to 5 years. 1 lag = 1 day *IANAE SurgicalOntologist fucked around with this message at 06:44 on Nov 11, 2016 |
|
#
?
Nov 11, 2016 05:29
|
|
|
|
| # ? May 27, 2024 06:14 |
|
|
Xyven posted:Imagine you have a crystal ball and know the "true" price of the market, x. Traders are good at their jobs so the market tends to price around x, with some small random swings. One day the "true" price drops 10%. What happens next is that traders get stopped out of positions, retail investors panic sell and the market drops 20%. You have no way of telling when this will happen in advance, nor do you know for sure that it is exactly what's happening, but your expected return increases whenever prices drop sharply because of this dynamic. For the millionth time, I'm not arguing that the markets have to be rational and random chance only. I'm saying that without information convincing you otherwise, assuming random chance is the best starting point. You then adjust based on how you think the market is irrational. Your example is fine in a vacuum; true price drops 10%, market drops 20% irrationally. I get it. Possible. But if this were a regular, average occurrence, that the market is so insanely irrational that this is common and predictable, it would be incredibly obvious from some very basic market analysis and then insanely exploitable. I can make the exact same example on the other side. True price drops 10%, market drops 0%. or drops 10 and then quickly recovers to 0 because the market doesn't believe the true price really is 10% lower initially. Do you see why these individual made up examples don't really matter in the grand scheme? it only matters if they're predictable in nature, which is what I'd love to see. of course the market isn't perfectly rational, but if it's irrational in any meaningful way that we can predict solely based on the sp500 dropping 10% then you should really want to see evidence before making that assumption. 9 What are the inputs here? And what's the x axis? excuse my ignorance Blinky2099 fucked around with this message at 07:05 on Nov 11, 2016 |
|
#
?
Nov 11, 2016 06:17
|
|