SOUNDS about right.
But looks like B Graham woud have preferred DOW+ SPYV \more value
Maybe less money than SPY + QQQ\ but more value\ lower PE
Volatility vs Risk
- Thread starter ironchef
- Start date
But looks like B Graham woud have preferred DOW+ SPYV \more value
Maybe less money than SPY + QQQ\ but more value\ lower PE
What risk? Volatility is not really risk of ruin, Graham's definition of risk.
I am on @wxytrader's side, SPY and QQQ are volatile but are almost risk free because they are "backed" by the US economy and their risk of ruin is almost none, as risk free as the USG's risk free rate.
Maybe I am dumb but for a Gen Z, putting money into SPY instead of T Bills for 40 years is quite risk free?
There are many types of volatility :
- big range (ie high volatility)
The market moves nicely / decisively / smoothly.
This is a low-risk market.
- big range (ie high volatility)
But the market moves chaotically / messily / jerkily.
This is a high-risk market.
- small range (ie low volatility)
The market doesn't move.
So you can't earn $$$$.
I am referring to risk free in the sense of zero risk of losing money. For example, putting money in a savings account or in treasuries will always grow. You can't lose money. From there if you have a risk of losing money at any point, you introduce risk, which should be fairly compensated. Why of course SPY and QQQ have risk (of which volatility is a way of measuring that risk... but agree is not a great indicator).
In terms of risk of ruin, fully agree that SPY and QQQ won't be going to zero. Also agree that with a 40 year horizon just parking in QQQ or SPY would likely yield a substantial amount of money and would have a zero percent chance of leading to ruin.
Its funny how people think QQQ is zero risk.
I thought that as well in 1999, 18 year bull market where Nasdaq had gone up over 20x in 18 years, Nasdaq, QQQ etc looked invincible, then QQQ has a 83% drawdown.
And you had to wait 16 years, until 2016 to get back to all time highs.
I thought that as well in 1999, 18 year bull market where Nasdaq had gone up over 20x in 18 years, Nasdaq, QQQ etc looked invincible, then QQQ has a 83% drawdown.
And you had to wait 16 years, until 2016 to get back to all time highs.
Bringing up 1999/2000 and 2007/2008 is a tired argument at this point. The market has completely changed and in no way reflects the market back then.
I still love how people fail to accept the reality of this market and why it's risk free. The market doesn't let a single red candle survive on any timeframe. There's too much money in the system, the market is way too important and on top of that too many derivates exists to ever let this thing have a real drop.
If you've follow the markets over the past 2 years you know this has unlimited pump in it and every while you may get a day or 2 red or even a rare week red. this thing is never actually going to go down.
Yes true. But if you hang on to it until today you are still way ahead of buying and holding risk free USG debts.
My point is when an index represents the US economy, it maybe volatile as hell but the risk of ruin is almost zero.
I totally agree.
The key for index is time. If you hold SPY or QQQ for 30 years, the risk of losing money is almost zero too.
IMHO, volatility is the std dev of returns, but risk is something different. Risk is the std dev of the residuals of your “model.” When you have no model, then volatility and risk end up being the same thing since the residuals become the returns around the mean (or 0 if there is no expected drift).
The above difference between vol and risk is based on the assumption that we can model the returns of an instrument. This shouldn’t be a stretch as we’ve been relying on models such as CAPM and the factor models from Fama-French, etc. for a long time. Let’s say that you’re using a single factor model (it’s easier to visualize), perhaps using the momentum factor. So you load up on positive momentum stocks and short negative momentum ones. In our idealized example, the positive momentum stocks may have high volatility due to the fact that their prices go up a lot and the negative momentum stocks may have high volatility because their price depreciates a lot. But, in this highly unrealistic and for illustrative purposes only scenario, the “risk” experienced by your model would probably be much lower than the volatility in the associated stocks. Your model would have nice (positive) returns since it went long the appreciating stocks and short the depreciating ones. There would be a distribution to the returns generated by your model over some time period (think daily returns over a 20 year span). And, these returns generated by your model would not match exactly the returns “expected” by your model; in some time periods, your model would expect higher returns than it produced and vice versa. The difference between the returns the model generated and expected is the residuals. The std dev of the distribution of these residuals gives us an idea of the “riskiness” of the model.
Of course, everything I’ve said above is easier said than done and is based on some assumptions that we know don’t really hold in practice. For one, it assumes that we can actually create an unbiased model of the underlying process; I’m not sure that I’ve actually ever seen this. Certainly it would mean that our model is free from any data snooping, look ahead and any other biases; no curve fitting
Another issue is that we’re using std dev, which really only works well on normal distributions and we know that market returns (even log returns) are not normal. But, IMO, despite practical implementation challenges, the above mental model is useful when thinking about the difference between volatility and risk.
The above difference between vol and risk is based on the assumption that we can model the returns of an instrument. This shouldn’t be a stretch as we’ve been relying on models such as CAPM and the factor models from Fama-French, etc. for a long time. Let’s say that you’re using a single factor model (it’s easier to visualize), perhaps using the momentum factor. So you load up on positive momentum stocks and short negative momentum ones. In our idealized example, the positive momentum stocks may have high volatility due to the fact that their prices go up a lot and the negative momentum stocks may have high volatility because their price depreciates a lot. But, in this highly unrealistic and for illustrative purposes only scenario, the “risk” experienced by your model would probably be much lower than the volatility in the associated stocks. Your model would have nice (positive) returns since it went long the appreciating stocks and short the depreciating ones. There would be a distribution to the returns generated by your model over some time period (think daily returns over a 20 year span). And, these returns generated by your model would not match exactly the returns “expected” by your model; in some time periods, your model would expect higher returns than it produced and vice versa. The difference between the returns the model generated and expected is the residuals. The std dev of the distribution of these residuals gives us an idea of the “riskiness” of the model.
Of course, everything I’ve said above is easier said than done and is based on some assumptions that we know don’t really hold in practice. For one, it assumes that we can actually create an unbiased model of the underlying process; I’m not sure that I’ve actually ever seen this. Certainly it would mean that our model is free from any data snooping, look ahead and any other biases; no curve fitting
Another issue is that we’re using std dev, which really only works well on normal distributions and we know that market returns (even log returns) are not normal. But, IMO, despite practical implementation challenges, the above mental model is useful when thinking about the difference between volatility and risk.