Felix Salmon

Market reports are hurting America

By Felix Salmon
October 10, 2011

Last week, Chao Deng published her “memoirs of a market reporter” at CJR.

Critics say markets reporters must suffer from A.D.D., because short-term fluctuations in stock indices really don’t matter much in the long run. They say it’s absurd to pin a single narrative on spot news involving countless individual decisions, many of them made by robots. Too often, coverage favors one slant if stocks are up and another if stocks are down when, in fact, nobody really knows.

And yet, the bigger the swing in the Dow, the more urgent the need to chase down an explanation, even if it’s a short-term one. Indeed, larger swings actually predict greater reader interest, which, in turn, validates the coverage.

She’s half right. It is absurd to tie narratives to intraday market moves. On the other hand, it’s even more absurd to chase down an explanation, especially when most of the time there is no explanation. Yes, readers demand such things. But they only harm themselves by doing so.

Here’s Chao criticizing one of her own headlines:

In response to one of my pre-market stories headlined “Futures Gain on Obama Jobs Plan,” for example, a reader had commented:

Can you prove that Obama’s $300b plan which has no chance of passage is the reason for futures being up today. There have been many days where futures rebounded after 300+ points of losses. Pretty sloppy reporting.

In retrospect, I wish the headline had been that futures were gaining “ahead” of the President’s jobs speech. Then I would have been laying out a possible reason for the gains in futures but not definitively pinning down on one. It’s a word game, sure, but words matter, and a small tweak would have resulted in a more accurate headline.

This kind of ridiculous Clintonian language-parsing helps no one, except insofar as it applies a wafer-thin layer of CYA to a practice which is fundamentally absurd. “I never said that stocks rose because of the jobs plan, I said they rose ahead of the jobs plan! So, I’m not saying that there is a causal relationship there, just that there might be! And you can’t deny that!”

Chao does try another tack, which is closer to David Gaffen’s argument in the video above:

All this might lead you to conclude that the market moves randomly most of the time, and we shouldn’t even try to find out why. But wait. Throwing our hands up is just as extreme an overreaction as pinning a day’s move on a single event. For one thing, it’s a sure way to lose readers, who are grasping for an explanation. For another thing, there are ways to do it reasonably without falling into the over-simplification trap.

Actually, faced with inexplicable moves in the stock market — and the vast majority of intraday moves are inexplicable, in that no one really has a clue why they happen, or whether there’s a reason for them at all — throwing our hands up is an entirely rational reaction, and not an overreaction at all.

There will always be readers who want some ersatz explanation of what the market did today; there will always be news organizations pandering to those readers. But if you’re a media organization which stands for reporting the truth in a high-quality manner, then market reports are a dangerous place to go, because they’re all built on quicksand.

Is it possible to do market reports “reasonably, without falling into the over-simplification trap”? I think it probably is, if you report on what the markets have been doing over the course of a few weeks or months, and you do a lot of reporting and thinking. But if you’re writing a dozen market reports a day? No. None of those are going to have real value.

Market reports should not be an everyday staple of news coverage. Sometimes, occasionally, there are stories in the markets. And then those stories can be reported. But when there aren’t any stories, there’s no point in trying to invent them. And so the daily report — let alone the intra-day report — is at heart a stupid piece of journalism. Some are better than others, to be sure. But none of them are any good.

20 comments so far | RSS Comments RSS

A relatively easy fix would be to insist that market reports include the weekly, monthly and annual moves in addition to the daily ones. The longer term trends are more important and smooth out a lot of the noise.

That still doesn’t fix the problem of building elaborate explanations for noise to fill the 24 hour news cycle. But that is only a problem if you still labor under the mistaken impression that the goal of the news is to inform rather than to entertain.

Posted by ralmond | Report as abusive

“Throwing our hands up is just as extreme an overreaction as pinning a day’s move on a single event. For one thing, it’s a sure way to lose readers, who are grasping for an explanation.”

Note that she never said that throwing our hands up is extreme *because* it’s a sure way to lose readers; she just placed them next to each other in a way that a human speaker of the English language would use to *imply* that. And besides, you can’t prove that reality isn’t driven by the dictates of reader demand. At least not if you have a journalism degree.

Posted by dWj | Report as abusive

Weekly, monthly and annual moves don’t give you a trend, they can still be very skewed by one very good or very bad day at one of the two end dates. Much better to take a moving average, or plot a chart through the peaks and troughs and see which way it’s going. But that’s probably beyond the abilities of the average TV presenter, or even news journalist. As you say, modern journalism is moving more toward the entertain category than the seek out the truth method we probably expect.

Posted by FifthDecade | Report as abusive

FifthDecade, why would you expect journalists to be able to do the [mostly worthless] first year undergrad maths you mentioned when they don’t know that Central banks are MEANT to be lenders of last resort, what custodian bank business models are, what the difference is between different categories in the capital structure etc.

It is a bit like complaining a science journalist can’t solve two body equations when the same journalist doesn’t even know what gravity is.

Posted by Danny_Black | Report as abusive

I strongly disagree with the premise that “most of the time there is no explanation”.

Sure for the last couple of months the moves have been much harder to explain (as is typical in moments of high volatility and low volumes).

But USUALLY there are very good explanations for exactly WHY a whole bunch of people (and bots) are driving the market in one direction or the other.

As a fundamental stock picker it is pretty rare that I have to say – oh, such-and-such a stock is down because the market is down. It is much more likely that NEWS has driven the stock down.
The aggregation of all these new items of information are what drive the overall market since that is after all what the market is… a bunch of stocks.

Posted by TinyTim1 | Report as abusive

Technically speaking, markets are complex (not complicated) systems. A complicated system is like a fine Swiss watch; there are a lot of pieces, but you can make accurate predictions because you have an understanding of EVERY piece in the whole watch system. A complex system is one where not every component (or its behavior) is known and where the components interact in unpredictable ways. Furthermore, the system itself may be changing over time (i.e., complex adaptive system). This last point is what differentiates weather from markets and where Felix should have gone with his debate video.
The weather is influenced entirely by natural laws, but a market is based on decisions that people make. Markets do not exist in some objective reality, they are emergent behavior. The weather does care what you think of it, but markets do. Market reporting has both beneficial and negative effects on the market. When they report good news, it serves to bolster people’s confidence and they are more willing to invest. When bad news is reported, confidence drops (i.e., self-fulfilling prophecy). The journalism itself is a part of the market system. Because of the delay in news and attempts to both explain past behavior and forecast future behavior, journalists are creating both leads and lags in the market system. These leads and lags (among a whole host of other things that have and have not been identified) lead to instability (read: unpredictability) in market behavior. Because of the butterfly effect, traditional cause-and-effect must be thrown out the window, especially in the case of markets.
In summary, markets are emergent behaviors of a complex adaptive system that itself is partially determined by journalism reporting.

Posted by cptcodfish | Report as abusive

It’s precisely because the market is an aggregate that it can be so hard to tease out exactly what is driving some particular daily blip as to be impossible, for practical purposes. “A storm in Latin America is now expected to drive up the prices of flowers and coffee, denting this quarter’s margins of SBUX 1% and FLWS 8%, because SBUX has better alternate sources of supply and beans are less perishable than flowers. This depresses expected earnings by about 0.2% for SBUX and 5% for FLWS, and explains the 0.001% dip we just saw in the Nasdaq.”

Posted by SelenesMom | Report as abusive

TinyTim1, depends on what you mean by “no explanation”. There is usually no underlying fundamental factor driving the short term price movements. It is mostly a mix of noise and various relationships between different financial products driving the prices. How well A is doing is unlikely to drive price movements short term.

cptcodfish, with all due respect, you missed the whole point of a complex system. That is that there is a large population and the local interactions are simple and usually deterministic but that the global behaviour is not predictable. For at least 40 years now we have known that even if you do know 100% the behaviour of the individuals and the interactions and even if they are deterministic then there can be a strong limit on predicting the behaviour of individuals in the system. Note this does not mean you cannot make any predictions. Also Butterfly effect has nothing to do with cause and effect. Again these are technical terms that have very specific meanings and you mangled them.

As for journalists, they add to the noise except that they work in a feedback loop. One journalists write about A, then the others just cut and paste and before you know it they are having a short-term market impact – one with possibly large long-term consequences.

Posted by Danny_Black | Report as abusive

I just think “usually” and “most of the time” are too strong.

Markets tanked for what 5 days after the S&P US downgrade.
There are 5 days happily explained, for example.

If journalists had the access they could happily call some major brokers who would be able to at least describe the flows.
Those guys could then ask their major accounts on the day WHY they had chosen to buy/sell at that particular moment.

Sure some of it is technical or momentum or HFT driven but I still believe LOTS is fundamental.

Felix and Mr Black have not shown me any EMPIRICAL evidence that >50% of the trading days have a market move that cannot be explained.

This isn’t a question of chaos theory or “noise” – it is a question of discovering the underlying motives behind market participant action which is surely possible.
Perhaps not possible for CNBC though!

Posted by TinyTim1 | Report as abusive

TinyTim1, I didn’t say they are no explanations just they are no ones based on the fundamentals of the company. They are usually driven by either flows of money into and out of certain funds or technical linkages.

Check how much value funds manage as a percentage of assets under management and check how much of the short-term trade flow they are part of. If journalists called major brokers, then a) the broker would put the phone down on him/her and if not then b) they will say what i just said.

In your example if the markets tanked after the S&P downgrade for 5 days did they bounce back up again on the 6th?

Posted by Danny_Black | Report as abusive

Wow, are we trying to re-proof efficient market theory here ? Wow, please look up Brealey and Myers basic stuff. Stock charts are statistically all noise. Has been examined over and over again 50 years back.
@cpcodfish: if you find even one butterfly phase diagram in any of the thousands available sets of stock data you are one lucky guy. What do you think ?

Posted by FBreughel1 | Report as abusive

The primary issue here – that is, with my colleague Felix Salmon, who is as handsome as he is misguided – is that market reports, written badly, will indeed fail to inform, and serve no more utility than an incomplete box score about a baseball game.

But the sports analogy is apt here. You don’t draw conclusions on the state of a baseball team from one game (ok, the Mets are an exception), just as you don’t do the same from the market. Which is where context comes in. Boring, repeated, restated, and then repeated context. Again and again.

But the job of a market reporter – and I haven’t written any confessionals, though I’ve been doing it 15 years – is to understand short-term events, medium-term events, and long-term events and the changes, as they happen, by reporting on it. Constantly.

Because if there’s one thing Felix can agree with me on, it’s that muscle memory means a lot. The more you write, the more you write, and so it is with reporting. Never covering a market, only to wake up and try to discern just what a big drop in stocks means at a given moment, is not an easy task. And it would rob the reader of the expertise one gleans from doing this on a daily basis.

Does that mean that every market report is one that investors should act upon? Hardly. Does that mean every market report is going to be a humdinger that we hold up as a prime example of amazing journalism? No. But there are reasons for this day-by-day understanding, the blocking and tackling, so to speak.

But Felix is English, so he doesn’t get American football analogies.

Posted by davidgaffen | Report as abusive

@SelenesMom, sounds like you should be writing these market reports.

Posted by Curmudgeon | Report as abusive

davidgaffen, the probability that a journalist is going give actionable investment news is around 0. Financial reporters – and i don’t think i have ever read anything of yours – tend to have a very very shallow understanding of the markets and frankly by the time it is widely known public information most of the market participants will have long acted on it.

FBreughel1, if it was all noise then that would be a proof of the efficient markets hypothesis which states exactly that. Secondly, again to be pedantic, butterfly phase diagram and what the other guy said are completely different mathematical concepts.

SelenesMom, that is because fundamentals don’t drive day to day price movements, market microstructure and noise do.

Posted by Danny_Black | Report as abusive

Danny_Black, I disagree on that point (unless you are using a very restricted definition of “actionable news”).

Plenty of situations which build slowly. Understanding the growing context within which the individual stocks trade can help to anticipate risk factors — risks that often are not yet fully recognized by the market.

Might somebody reading the financial pages have recognized the risks to the financial system in the summer of 2007? Perhaps taken that as reason to stay away from the apparent bargains in LEH? Or reading about the European turmoil, decided to step away from those stocks for a few months?

The trick, naturally, is reading what “might be” rather than what has already happened.

Posted by TFF | Report as abusive

@ Danny_Black: No, I think cpcodfish meant it: “Because of the butterfly effect, traditional cause-and-effect must be thrown out the window, especially in the case of markets”.
In any complex deterministic system an initial infinetismal different starting variable would not grow so much for the system to become unpredictable. It would be difficult, but still predictable. A complex chaotic system would not be predicable but since its pattern is stable, money could be made. I thought his distinction between complex and complicated was a particular give-away. Too bad cpcodfish didn’t reply !

Posted by FBreughel1 | Report as abusive

Market should move on news, so explanations that involve things that should already be well in the price (most sovereign rating downgrades, for instance) are pretty useless to start with.

What’s often missing, from a practitioner’s point of view, are the explanations of why the market doesn’t move when some piece of news upon which you have built a strategy happens, and yet you reap no benefit from your correct expectation.

Posted by Th.M | Report as abusive

TFF, I think I do have a rather restricted definition of actionable. It is a bit like porno but I will try and come up with something which approximates my idea. I think actionable news is something you can use within your analytic framework to have a greater than 50/50 chance of turning a profit that you would not have without said news and without the benefit of hindsight.

FBreughel1, as may be obvious I am a bit pedantic so if I am about to proceed to rewrite what you meant in slightly different language I apologise in advance.

Deterministic means that there is no random component, ie if I put the same input in I will ALWAYS get the same output. Random is non-deterministic. Predictable means that given knowledge of the initial variables with a certain error, at any future time I can state the output with a certain error. Deterministic and predictable have nothing to do with each other. A system can be random and have predictable outputs and deterministic and non-predictable. Chaotic systems is another name for non-predictable. A Chaotic system can consist of a small number of variables and a typically rather complicated interactions – for instance the Lorenz equation or logistic equation. A complex system is one with a large number of components each with simple local behaviour but which exhibits complicated – possibly chaotic – global behaviour.

The sense in which “traditional cause and effect” is chucked out is that people assume if there is a big effect that the cause has to be large and that these jumps are not simply inherent in the system. For instance if 5 days the market tanks then it must because of X whereas it can be lots of little things feeding back on each other.

Posted by Danny_Black | Report as abusive

If salmon are obsrtucted from returning to the place of their birth with they spawn in fresh water anyway?Please reply thank you.

Posted by roy13miller24 | Report as abusive

@Danny_Black I appreciate your adherence to the strict definitions and I agree I mangled them. Your comments show me that I may have muddied the waters for the sake of simplicity. However, this discussion was initiated to talk about why I thought market reporting is rubbish, not to discuss how I may have muddied the waters. To go back to my original point, market reporting is garbage for two reasons. (1) We will always be limited with our our predictive capability because outside of toy-problems found in textbooks, we never know all factors and we can never measure them with enough precision. There is always one more decimal place that we cannot attain. As long as we cannot know all factors, we will have an imperfect prediction. As long as we cannot measure with infinite precision, we will always have an imperfect prediction. Just as with the butterfly effect, a small imprecision (the decimal place we cannot attain), there is the potential that our prediction will be rubbish. (2) Markets are complex adaptive systems. As mentioned, small things can feed back (or forward) into each other to result in large effects. Consumer confidence is one of the variables that influences market behavior. It has a circular relationship with market reporting. That circular relationship means that whatever coefficient you use in your regression equation will always be changing and adapting. These feed off of each other. Given that we cannot measure these with infinite precision and that they are constantly changing, we find ourselves in the situation of Point 1, that the predictions will likely be rubbish because small differences can have substantial effects.

Posted by cptcodfish | Report as abusive

Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/