speaking of correlations...I read somewhere that valuations have absolutely zero predictive ability over one year. Longer than that, sure, but not over one year.
keep that in mind next time you try to time the market, thinking it is expensive.
You need to be much more clear about what you think is autocorrelated, and remind me what bearing that has on whether the hypothetical retiree should put their entire $1M life savings into US equities, because I've lost the thread.
Your simulation allows for runaway scenarios, ie where the market tanks multiple years in a row, sometimes 3+ years in a row. In reality, that doesn't happen. I explained it in my first response to your simulation. You can simply look at a plot of average annual returns or 3 year moving average returns. After a big positive data point, the next data point is likely to be something that brings the data back to the average. Example: If the market returned 30% one year, your model (which assumes independence) would say there's a just as high of a probability of seeing another 30% year in t+1, and again in t+2. That is simply wrong. After a 30% year, we are much more likely to see a smaller gain the following year to bring the overall average down to 10%. This fact greatly reduces variability in possible (likely) outcomes of your simulation. The chances of going broke reduce tremendously, as do the chances of making obscene amounts of money.
OK, admittedly don’t follow all the shop talk. It appears to me that analyzing index price levels from a statistical view is not much better than technical analysis.
After a big positive data point, the next data point is likely to be something that brings the data back to the average. Example: If the market returned 30% one year, your model (which assumes independence) would say there's a just as high of a probability of seeing another 30% year in t+1, and again in t+2. That is simply wrong. After a 30% year, we are much more likely to see a smaller gain the following year to bring the overall average down to 10%. This fact greatly reduces variability in possible (likely) outcomes of your simulation. The chances of going broke reduce tremendously, as do the chances of making obscene amounts of money.
OK, I see what you believe to be true. Sorry to break the news, but you are completely incorrect. No shame in that, we all have pre-conceived notions and unsupported beliefs.
Feel free to dive into the data yourself and prove me wrong.
speaking of correlations...I read somewhere that valuations have absolutely zero predictive ability over one year. Longer than that, sure, but not over one year.
keep that in mind next time you try to time the market, thinking it is expensive.
Well I stated the market was expensive in March 2015 (S&P 500 2,200). Interestingly it was back at that same level March 2020. The Fed and politicians took market to new highs by early 2022. What was the cost? $5 Trillion Fed QE and another $3 Trillion in government stimulus. Where is the market today? Back to where it was May 2021. So, all that money bought 2,000 S&P 500 points in a 13 months. More expensive today then when I first posted. Inflation is so strong it suddenly choked the “valuation doesn’t matter” argument.
BTW, here is a plot of 40 years of evolution of the hypothetical $1M stake, showing the average (expected value) assuming 100% in a SP500-tracking fund, the average plus and minus one standard deviation, and a theoretical "safe" 6% per year investment (assuming such a thing were to exist; I include this because it aligns with the worst case graphs Igy shared yesterday):
The mean minus one standard deviation leads to bankruptcy well before death (this lady is a survivor).
Of course it is MOST LIKELY that the estate grows in value faster than the safe investment. But it's not guaranteed, whereas the "safe" investment (supposedly) is.
OK, I see what you believe to be true. Sorry to break the news, but you are completely incorrect. No shame in that, we all have pre-conceived notions and unsupported beliefs.
Feel free to dive into the data yourself and prove me wrong.
I have. You don't even need to do any analysis on the data. Just look at the numbers year to year or plot them on a graph. Or just think about it the way I've already described. There is essentially no realistic scenario in real life where the market would drop 30% three years in a row. Probability is essentially zero (unless the entire US financial system collapses, in which case cash and bonds will be worthless also).
Wish we could talk more about this in real life. FWIW, I'm nearly finished with an MS in statistics. Autocorrelation is brought up in any statistics class that mentions time-series at all. If I wasn't so busy with end of semester tasks I'd run the Likelihood of your losing scenarios myself to get specific values. Maybe in two weeks...
There is essentially no realistic scenario in real life where the market would drop 30% three years in a row. Probability is essentially zero (unless the entire US financial system collapses, in which case cash and bonds will be worthless also).
Your objection carries no weight. None of my 200 model runs contain such a 3-year scenario.
It's great you are studying statistics at the graduate level, and maybe feeling your oats a little bit, trying out your budding knowledge anonymously on the internet. Please recognize that when you engage anonymously on the internet, you have no idea the background of the person(s) (or bots?) you are arguing with. Maybe I'm a complete moron shouting into the internet, or maybe I've got graduate degrees and a career built on working with statistics. Either way, I'm right and you're wrong, so there! :-D
Feel free to ask your professor to read this thread and weigh in. That would be fun!
Tom Lee, Fundstrat Global Advisors managing partner and head of research, joins 'Closing Bell: Overtime' to discuss what we're likely to see in the markets a...
BTW, here is a plot of 40 years of evolution of the hypothetical $1M stake, showing the average (expected value) assuming 100% in a SP500-tracking fund, the average plus and minus one standard deviation, and a theoretical "safe" 6% per year investment (assuming such a thing were to exist; I include this because it aligns with the worst case graphs Igy shared yesterday):
The mean minus one standard deviation leads to bankruptcy well before death (this lady is a survivor).
Of course it is MOST LIKELY that the estate grows in value faster than the safe investment. But it's not guaranteed, whereas the "safe" investment (supposedly) is.
Your argument hinges on your assumption of independence. My position is that there is not independence. So, seeing -1 sigma returns year after year is far less probable than what you're assuming, as is +1 sigma returns year after year. Annual rates of return bounce up and down along the 10% line. They do not ride substantially above or below that line for years on end.
Your objection carries no weight. None of my 200 model runs contain such a 3-year scenario.
It's great you are studying statistics at the graduate level, and maybe feeling your oats a little bit, trying out your budding knowledge anonymously on the internet. Please recognize that when you engage anonymously on the internet, you have no idea the background of the person(s) (or bots?) you are arguing with. Maybe I'm a complete moron shouting into the internet, or maybe I've got graduate degrees and a career built on working with statistics. Either way, I'm right and you're wrong, so there! :-D
Feel free to ask your professor to read this thread and weigh in. That would be fun!
It was just an example to illustrate a point. The fact you even know what a monte carlo simulation is tells me you're more knowledgeable about statistics than basically every other person on this thread. Still wrong though. ;)
You really need to dive into the data instead of performing a flawed mental experiment and imagining what they may say. You've dug yourself into a little hole and it's not too late to climb out.
Your objection carries no weight. None of my 200 model runs contain such a 3-year scenario.
It's great you are studying statistics at the graduate level, and maybe feeling your oats a little bit, trying out your budding knowledge anonymously on the internet. Please recognize that when you engage anonymously on the internet, you have no idea the background of the person(s) (or bots?) you are arguing with. Maybe I'm a complete moron shouting into the internet, or maybe I've got graduate degrees and a career built on working with statistics. Either way, I'm right and you're wrong, so there! :-D
Feel free to ask your professor to read this thread and weigh in. That would be fun!
You are really a nice character. Funny yet polite retort; as always.
In comment to your hypothetical retiree, I was eligible for one of the first self-directed employer retirement accounts in 1982, some 40 years ago. Fortunately I can afford to be a bear— I worked until age 70.
My best guess, is we settle in at the pre-pandemic high of S&P 500 3,380. That would be by mid-summer, with further downside risk as we approach the fall.
“There is essentially no realistic scenario in real life where the market would drop 30% three years in a row. Probability is essentially zero (unless the entire US financial system collapses, in which case cash and bonds will be worthless also).”
From 3/2000 to 10/2002 the NASDAQ collapsed 83% and did not reach a new durable high until spring 2015. Not 30% three years in a row, but significant drop. At the high point tech was about 30% of the market cap of the S&P 500, and dominated the index EPS. Most here discount the market risk using statistics that have little to do with market valuations and current to future economic conditions. The financial industry doesn’t help, encouraging excessive risk. IMHO.
speaking of correlations...I read somewhere that valuations have absolutely zero predictive ability over one year. Longer than that, sure, but not over one year.
keep that in mind next time you try to time the market, thinking it is expensive.
Well I stated the market was expensive in March 2015 (S&P 500 2,200). Interestingly it was back at that same level March 2020. The Fed and politicians took market to new highs by early 2022. What was the cost? $5 Trillion Fed QE and another $3 Trillion in government stimulus. Where is the market today? Back to where it was May 2021. So, all that money bought 2,000 S&P 500 points in a 13 months. More expensive today then when I first posted. Inflation is so strong it suddenly choked the “valuation doesn’t matter” argument.
2015 SP500 EPS: $100/share
2022 SP500 EPS: $225/share
Corporate profits have grown tremendously....betting that the increase in earnings is because of the government....is a difficult assertion to hang on to. I choose not to go with that one.