Here’s an amusing chart showing the percent of stocks that sell-side analysts have rated “sells”, on average:
There’s a million junk-chart bloggers who will tell you how much is wrong with this graph (myself included) – starting with the left hand scale, which should go up to 10% rather than 100%. But in a rare twist, the left hand scale is the graph. The message here is that “sells” are a minuscule component of the whole universe, and that the SarbOx legislation did little to affect that proportion. The tall left hand scale and bold “Enactment” line highlight the incongruity of the rest of the graph.
(via Paul Kedrosky)
I don’t believe that a trader error kicked off yesterday’s crash. I do believe that computers exacerbated it. I also believe that there were transactional errors following the initial collapse. I would like to point out that when I say “computers exacerbated it,” I include both HFT, algorithmic trading, and every retail investor’s trailing stop-loss.
I’m ready to eat those words if the record shows me to be wrong.
But here’s my evidence (besides how nonsensical it is that a single sell order could trigger that collapse):
That’s a tick-by-tick of Accenture in red – banner stock of the “trading mistakes killed the market” crowd – overlaid on the S&P 500, Dow 30, and SPY. You’ll notice that it hits its bottom one minute after the indices, which are actually well on their way back up when the (clearly erroneous) print occurs. If there’s causality here, it involves all sorts of time travel.
But I know better than anyone that graphs with multiple vertical axes can be misleading, so here’s the exact same data, normalized to be 100 at 2:30 PM:
Here you can plainly see ACN lagging the general indices… until they start to turn back up, at which point its decline steepens. Again, there’s no correlation between the ACN low and the index performance.
Anyway, I don’t know what sparked the selloff. I’m willing to believe it was a groupthink collapse. We saw the same thing – though much worse – almost two years ago, though we were able to find a news story which satisfied us as a catalyst in Lehman’s collapse. Here, there was no news on the wire or obvious cause – but that doesn’t mean it was malicious or erroneous. It just means some people (or computers) started to sell, and then others followed suit… and before you know it, the whole thing is compounded by errors on the major exchanges (but not caused!).
But come on, isn’t it a little ridiculous when major news outlets believe that hitting the letter “b” sank the markets? Fat finger mistakes have happened before (disproportionately in Tokyo), but they usually result in losses for the firm that makes them, not the entire world.
Type carefully, my friends.
The WSJ has crunched the numbers and concluded that:
In nearly 200 years of recorded stock-market history, no calendar decade has seen such a dismal performance as the 2000s.
Investors would have been better off investing in pretty much anything else, from bonds to gold or even just stuffing money under a mattress. Since the end of 1999, stocks traded on the New York Stock Exchange have lost an average of 0.5% a year thanks to the twin bear markets this decade.
I think the authors should have stopped right there, scratched their heads and wondered how the words “twin bear markets” and a tiny loss of just 0.5% a year could appear in the same sentence. But they did not, they plowed on with such statements as, “From 2000 through November 2009, investors would have been far better off owning bonds.”
The answer, of course, was a bubble-fueled bull market of low volatility and cheap credit coupled with a decade of unprecedented volatility in general. If we assume for a moment that instead of being a horde of mindless idiots, investors are shrewd market timers, they could have returned 100% just by buying the S&P 500 basket from 2003 to 2007. They would have had not one but two (“twin”) opportunities to reap the same profit shorting the market in 2001 and 2008.
This is why the exercise of boiling arbitrary time periods down to a first order statistic – the average return – is such a meaningless exercise if performed without context. It fails to capture the full texture of the decade – the ups, the downs, the in-betweens. The richness and texture of the last 10 years are to some extent by the S&P’s chart. It’s not useful, instructive, or even a proper comparison to compress the decade. Show me an investor who purchased the index at the start of the decade and hasn’t traded since, and I’ll show you someone who actually might find long term bonds a more suitable alternative.
The second order measure is the real story of the decade: in 2008, the VIX hit 80!! And as late as January 2007, it was under 10!
But enough with the decade retrospectives, with the sensationalist headlines comparing this decade’s average return to that of the 1930′s (here’s a hint: Black Friday was in 1929). You want a story? Focus on the twin bear markets – look at how amazing it was that you could have two in such a short span and still only lose 0.5% a year on average!
But the next time I find someone who only invests on January 1 of years ending in zeros, I’ll be sure to pass this along.
The WSJ’s top story this morning was one titled “The Cruel Math of Big Losses” – an article written as if it were an eye-opening expose into a little-known piece of financial wisdom rather than a blatantly obvious restatement of basic math: when you lose X%, it takes a gain of more than X% to get back to even.
Nonetheless, I’ve seen the mistake time and time again – and it is exactly what leads people to chase levered ETFs despite their downward bias. It is especially amusing that many of the comments on the article fall in two camps: either the article is a waste of time because percentages don’t matter, only dollar amounts, or it’s a waste of time because its a boilerplate nascent recovery article.
To the first point, percentages are all that matter. Yes, if you lose $50 you need to make back $50. But what if you started with $100? What if you started with $1,000 – or $51? The risk (both future and encountered) on those three portfolios is very different and must be considered as such. It remains beyond me why we insist on reporting stock changes in terms of dollars or points. Is it relevant to me that the S&P is up 12.8 and the Dow is up 124.2? Of course not! – but saying that both of them are up 1.26% is actually useful information. At least with individual stocks you could do some mental arithmetic, taking the numbers of shares you hold times the dollar change – but that doesn’t work with the indices. Let’s beware this “dollar price” fallacy: if you don’t have a spreadsheet (and even if you do!) dollar prices should not be your primary concern.
To the second point, a wise investor would keep in mind that this math works both ways (neither of them helpful): gains can quickly evaporate when the market drops a relatively smaller amount.
This article serves a useful purpose: it reminds us of the basic temperament of the market. It forces us to confront the difficult nature of the investing exercise. But it should never come as a surprise.
Via Alea, a very interesting econometric study on the impact of false news on stock prices.
In September 2008, an article on United Airlines’ 2002 bankruptcy resurfaced and was distributed as if it were new information. The company’s stock plummeted immediately, but bounced back and by the end of the day was off only 11% - which is still surprising, given that the news was in no way relevant. The paper uses a factor model to examine how the stock returned to its “correct” level following the false idiosyncratic shock – a process which took almost a full week, as illustrated here:
The factor model only has an R2 of 40%, but it appears to fit the data quite well at a glance – I’ll hazard the guess that with a stock like this, enough outlying idiosyncratic moves exist to distort the R2 even as the model preserves most of the behavior. This is an interesting opportunity to examine the interplay of idiosyncratic and systemic information – the alpha and beta moves are quite apparant (assuming we accept the factor model, but given its out-of-sample performance I’m inclined to do so in this case). Following the initial shock, the stock tracks the lower error band, though I’m not convinced there is significance in it tracking that particular line. Over the next few days, a deliberately positive alpha move carries it back to the factor model baseline, but the beta sensitivities remain apparant. Finally, alpha dissipates and the stock resumes its factor-predicted path.
Unfortunately, such cases are so hard to observe and one example does not data make. I’d be curious to see a similar analysis of Apple’s stock following the false announcement of Steve Jobs having a heart attack.
I firmly hold that all financial contracts are zero-sum games. Recently, however, I have heard many arguments premised on the idea that the stock market is positive sum because economic growth creates wealth, which is reflected in universally rising stock prices.
But in this scenario, you purchase $1 of stock on Day 1. On Day 100, thanks to robust economic growth, you sell your stock for $100. Are you wealthier? Yes. Has the amount of wealth in the world increased? No. All that’s happened is that someone else has given you $100. Wealth has been transferred but not invented.
In a more general sense: when you buy a stock, you give cash to someone in return for the stock certificate. Later, when you sell the stock, you hand the certificate to someone else in return for their cash. It does not matter whether the amount of cash at the end is more than that at the beginning or less – no cash has been created by this process.
This is not surprising, as it is well known that stock trades in the “secondary market” – which is to say, the market not directly related to the issuing corporation. But even IPO’s do not create wealth – they merely gather cash from many investors who are (suddenly) willing to hand it over to a single company. True, the company may use that cash to generate new wealth – but that is now outside the realm of the stock certificate, which remains a zero sum game in the secondary market.
There is one possible exception. Dividends provide a mechanism by which money is transferred directly from the wealth-generating company to the holder of an otherwise zero-sum game. In other words, an entitiy outside the zero-sum realm is giving money to someone within it – shouldn’t that necessitate dividend-paying stocks as positive sum games? I believe it does not. When dividends are paid, the stock drops in value by the amount of the dividend, ensuring that the holder does not get paid twice for the same dividend. (Incidentally, the same logic led Modigliani and Miller to conclude that dividends are irrelevant in a frictionless world, because any investor can create his own dividends by selling shares of stock). Since the sum of an investor’s cash and the value of a stock he holds is identical immediately before and after the dividend is paid, I conclude that dividends do not create a positive-sum environment.
Thus, stocks do uphold the premise of a zero-sum game: no wealth is created, it is merely transferred. To the extent that wealth is created and injected into the system via dividends, the value of the stock drops by an equal amount, so the net value remains unchanged. The search for positive-sum finance continues…
The market shot up late this afternoon because Paul Krugman stated:
“I would not be surprised if the official end of the U.S. recession ends up being, in retrospect, dated sometime this summer,” he said in a lecture today at the London School of Economics. “Things seem to be getting worse more slowly. There’s some reason to think that we’re stabilizing.”
In an exceptional display of reverse-causality, Krugman clarified today’s remarks with ones he made last Friday:
Nobel Prize-winning economist Paul Krugman said the world’s economy is showing “not a hint” of a “V-shaped” recovery marked by a swift decline and revival.
The economy is “stabilizing, not recovering,” Krugman, an economics professor at Princeton University in New Jersey, said today at a conference in Dublin. “Things are getting worse more slowly.”
“We have made the transition from sheer panic to chronic anxiety,” Krugman said, adding he’s has a “hard time” seeing what might drive a “full” economic recovery.
Indeed, anyone following Krugman for the last year is certainly aware that he doesn’t see the end of the recession as anything more than a semantic difference. An L-shaped recovery is hardly a “recovery” at all. But don’t tell the day-traders.
Those naive financial journalists at The Atlantic are back! Andrew Gelman pointed me toward this misguided look at the latest auto bankruptcy (you know the one I mean). Key quote:
Purists — and virtually every academic economist one happens to encounter — wonder what happened to the once inviolate principle of rewarding risk-takers.
You’ll have to excuse me because I’ve never heard of this sacred principle of rewarding risk-takers. In fact, I always had this funny idea that “risk” represents the chance that an investor won’t receive any “reward.”
Incidentally, the last time I checked in with the Atlantic I discovered a story with the line “when people are investing, they should get a return.” I’m usually joking with the whole naive journalist thing, but this is ridiculous.
Immediately after his “inviolate principle”, the author actually goes on to note that “you can forget about poor unadorned stockholders” as if it’s surprising that equity gets wiped out in bankruptcy. If that’s news to anyone, then I have some Lehman shares and a bridge for sale.
If this quote is indicative of real investor sentiment then the lessons of 2008 will have been sorely wasted.
And I apologize but if we can just back up for a second, the opening line of this article reads:
Bondholders are kicking and screaming, but it appears as if General Motors Corp. is headed for an orderly bankruptcy, and the Obama administration is about to be handed the keys to a venerable corporate institution. Again.
Woah, Silver. Are we talking about the same Obama who demonized the people that “forced” Chrysler into bankruptcy and characterized them as unpatriotic? I don’t think this is a positive development at all. Moreover I don’t think GM’s bankruptcy is going to be remotely “orderly”, especially after seeing how few bondholders agreed to GM’s restructuring plan. The Chrysler debacle shows that a minority group of investors can wreak all kinds of havoc. This isn’t going to be pretty.
I’ll bet my Lehman shares on it.
FT Alphaville has a post up regarding new research from Citi on how analysts make recommendations. It is accompnaied by this graph:
The graph shows the average recommendation across all analyst-covered stocks, for the last 15 years. A stock gets a 1 if every analyst recommends buying it; a 5 is given to a universal sell; 3 is neutral. The graph shows the average rating over time. Alphaville notes:
Basically it shows that over the last 15 years analysts have been moderately positive on stocks with an average score of circa 2.47. They were the most bullish at the end of 1999, then spent the next three years becoming increasingly bearish. Bullishness was on the rise between 2002 and mid-2007, before the analysts became bearish again.
I think the graph is somewhat misleading. Firstly, the range of values here – roughly 2.25 to 2.65 – only encompass 10% of the total possible range from 1 to 5. Thus, it’s a stretch to consider any of these values much more “bullish” or “bearish” than any other as they are concentrated in a very tight cluster. Without some knowledge of the distribution of ratings, it’s hard to draw a conclusion. If all stocks were 2′s, 3′s, and 4′s, then the graph has more validity than, say, if stocks only had 1′s or 5′s. An extreme example to be sure, but the variance of the ratings distribution is critical to interpreting any change in the mean rating.
Second, Alphaville notes that the average of 2.47 is slightly bullish. What’s more, the minimum and maximum ratings are both bullish as well. I think the graph is misleadingly labeled when it says “bullish” and “bearish” along the right axis. Those labels, taken together with the horizontal average line, appear to indicate that ratings above the line are bearish. In fact, none of the average ratings are bearish – ones above the line are merely more bearish than ones below, though still bullish themselves. “More bullish” and “more bearish” would have been more appropriate labels; “More bullish” and “less bullish” would have been even better.
At the end of the day, is anyone surprised by these findings?
Russell Napier wrote the book on identifying bear market bottoms (really) and Alphaville provides selections from an interview he recently gave to the FT. Fascinating stuff, if you like reading your news through Roubini-tinted glasses.
The concept of identifying any sort of local minimum without the benefit of hindsight is inherently questionable, but the use of indicators (such as P/E or q ratios) to identify what is most likely not a bottom may hold more water.
But to sum it all up, I can’t do any better than www.isthisthebottom.com.
