I liked The Big Short a lot, but Flash Boys was so bad that I found myself wondering how trustworthy The Big Short was.
You can see part of what's problematic about the latter book just from this Guardian summary of it. The summary would have us believe that HFT is a trick employed by giant Wall Street banks to screw over smaller players. But Katsuyama, the protagonist of Lewis' book, was a big-bank trader, paid exorbitant amounts of money to tax large block orders by pension and hedge funds, and outraged by his inability to compete with algorithms that were bidding that tax down.
I'm uncomfortable calling any giant financial firm "trustworthy", but if you were going to pick one firm that has come close to earning that label, it's Vanguard. Vanguard's chief investment officer has repeatedly and unequivocally stated that HFT firms have lowered Vanguard's costs and improve outcomes for its investors.
Here's Matt Levine writing entertainingly about how the "Flash Boys" in Lewis' book ought to have been, based on his previous books, the protagonists, and Katsuyama the villain:
I mean, I understand (and even slightly sympathise with) the way big hedge funds and banks hate the way HFT are competing with them and driving their margins down.
...what I'm confused by is why so many journalists and authors are taking the side of the hedge funds over the interests of retail investors. With apologies to Goldman, but when you have Goldman on one side and Vanguard on the other, my default assumption is that the Goldman side is probably doing something shady. Which is, to be sure, a rebuttable presumption, but not one I've seen rebutted.
I am a big fan of Lewis & I have read multiple accounts / variations of the sub-prime crisis. The Big Short tells only half the story but nevertheless a very important one especially by covering Michael Burry as patient zero for identifying the problem. However, the book does a big dis-service by not writing more extensively about John Paulson. He made the most amount ($15BN) shorting sub-prime mortgages & when you have a book called the Big Short but don't cover the biggest short, it seems incomplete. Gregory Zuckerman's "Greatest Trade Ever" should be read along with "Big Short" to get a more complete view of the crisis.
What I found most curious about "The Big Short" was its (almost exclusive, IIRC) focus on telling the story of the folks who were short these ultimately toxic derivatives: Why they thought this was a smart move, the finacial engineering and negotiation done so they could actually make the move, the usual Michael Lewis quirky portraiture.
What it was missing was any explanation of the folks who were long. Presumably many of them were at least almost as smart as the shorts, but I got no sense from "The Big Short" why they bet in the other direction.
Genuinely curious if there's a mirror of "The Big Short" out there that could fill in that gap.
It does explain it. They were long because they thought the housing market could never crash. It doesn't really matter how bad someone's credit is, if the value of homes keeps going up the banks will not lose money. They might put nothing down and then pay their mortgage for 5 years, then default, but the home's gone up 15 or 20% in value so the bank will resell it and be made whole.
For a longer (and fairly technical) answer, see "Slapped by the Invisible Hand".
Essentially, a subprime mortgage had the price-will-rise-forever view baked in, as a way for the borrower to passively increase equity in the house while making very low payments. When the prices stopped rising, these borrowers went underwater fast and could not refinance when they needed to.
There isn't but that is largely because the [general points] are popularly known:
1) Politically, at the Federal level, steps were taken to regularly increase subsidies to the housing market. It was a populist position in both parties so it seemed safe from serious intervention that would depress prices.
2) Housing had been increasing for some time and this upward momentum "seemed" safe in the general case where the borrowers were credit worthy. In the long term, housing has always gone up decade-over-decade at least as fast as inflation.
3) Housing demand is rock solid, even in times of crisis, so the fluctuations would be less severe than the surrounding environment. In other words, they had always assumed they could find buyers for housing in any market environment so the hits they took in a recession would be small.
4) Subprime being a vector for a contagion that would affect middle class homeowners and cripple the economy didn't occur to them. As such, risk was effectively mispriced even tho the banks [frankly] were and still are aware that the Credit Agencies are really captured by the large financial institutions. They had assumed that "AAA" would remain relatively safe.
5) The institutions themselves didn't really understand that the grading mechanisms for credit were so badly corroded by corruption to be essentially worthless in the housing market until the damage was so severe that all they could do was unload the toxic assets before other people realized they were toxic. The only people who could buy these at scale were other banks and financial institutions...
Parts of that is right, but I think you mis-state a lot of the ratings stuff.
The fundamental idea behind the ratings is (simplifying a lot) that you could take a lot of crap mortgages and then slice them into tranches, assigning losses to the lowest tranches first, and to the highest tranches last. So if you have 1000 mortgages, and you slice it into 10 tranches of 100 each, and 250 mortgages go bad, then the lowest two tranches lose 100%, the next tranche loses 50%, and the top 7 tranches lose 0%. What the ratings agencies were doing by assigning AAA grades to the top tranches was estimating the likelihood that EVERYTHING would go bad was very low.
And what I think a lot of people miss about the financial crisis is that this was not wrong. Not only was it correct in principle, but actual losses on AAA rated securities was extremely low because even among subprime a lot kept on being serviced. So the ratings agencies said "hell, we could have a massive crisis and these top tranches will still mostly pay out", and then we did have a massive crisis, and those top tranches still mostly paid out.
There were a lot of issues and problems and corners cut and bad paperwork and robosigning and a hundred and one other scandals, some quite significant. However when you say:
> They had assumed that "AAA" would remain relatively safe.
It's worth remembering that this assumption was shown to be correct! From the financial crisis inquiry report:
"Overall, for 2005 to 2007 vintage tranches of mortgage-backed securities originally rated triple-A, despite the mass downgrades, only [...] 4% of subprime securities had [...] losses imminent or had already been suffered by the end of 2009."
On the other hand, something like 95% of lower rated tranches ended up suffering losses. The tranches were a mechanism for concentrating credit risk in the lower tranches; the problem isn't that it didn't work, but that 1) it worked really well and 2) everyone was badly underestimating the total about of credit risk in the system. AAA was mostly fine; everything short of AAA was absolute toast.
(Also, I think you're missing how huge the decline in home prices was, both as cause and vector of the crisis. A vast amount of paper wealth nominally owned by the middle class vanished in a puff of smoke; most of the crisis was the result of that. But the impacts of an asset price bubble should properly be attributed to its inflation, not it's popping.)
> "Overall, for 2005 to 2007 vintage tranches of mortgage-backed securities originally rated triple-A, despite the mass downgrades, only [...] 4% of subprime securities had [...] losses imminent or had already been suffered by the end of 2009."
> (Also, I think you're missing how huge the decline in home prices was, both as cause and vector of the crisis. A vast amount of paper wealth nominally owned by the middle class vanished in a puff of smoke; most of the crisis was the result of that. But the impacts of an asset price bubble should properly be attributed to its inflation, not it's popping.)
Its inflation exists because of how corroded the mortgage ratings were. None of them were correct except relative to each other within the mortgage bond market. You don't rate things AAA when they are really BBB. I understand you might feel that is correct but I sure as hell do not.
That is the expected cumulative default rate for a AAA corporate bond. The mean is .10% with standard deviations of .07%. So when you say "4%" that is absurdly high in comparison yet they were marketed and sold as if they were equivalent risks.
> The extent and speed of rating downgrades of asset-backed
securities. Investors have learned to their cost that the credit
ratings of structured credit securities are more likely to suffer
rapid and severe downgrades than are corporate bonds (see
chart). Not only have downgrades occurred more frequently
to subprime securities, they have often been reduced several
notches at once because of the sensitivity of such securities’
ratings to increases in assumed credit losses. As a result,
investors’ faith in rating agency opinions has been shaken—
credit spreads on AAA-rated U.S. residential MBSs have been
priced at about the same level as BBB-rated corporate bonds
since August 2007.
The real value of the AAA-rated residential MBSs were the equivalent of BBB for other securities and had been for years preceding the event. The majority of the market was blinded by the assurances of the credit bureaus which were valuing BBB assets as AAA assets.
A credit bureau that doesn't consistently and accurately estimate risk across securities is in many ways worse than useless because it gives a false sense of security where none exists.
Similarly, captured referees / regulators / credit bureaus cannot be trusted if you are not the sole organization in control.
However, I can see how this is likely more in the realm of opinion so I'll just leave it here but I think the comparison to corporate bonds is relevant.
1) It's worth reiterating that there was a huge housing bubble and a massive devaluation. You say 4% losses is terrible for AAA rated securities (which is true), but you could also say it's pretty amazing for subprime mortgages during a massive housing bubble popping. It's like building a levee that's meant to stand up to all but the most massive floods, and then having a massive flood; the fact that the levee failed it not itself proof that it wasn't engineered correctly.
2) The point I was originally trying to make and got sidetracked from is that if (and seems likely to me) there was an extremely widespread systematic belief that there was no housing bubble and that a coordinated, nationwide drop in housing prices was effectively impossible, then you would see the exact outcomes we saw without the corruption or "corrosion" you posit.
In short, you're suggesting that people knew the loans were going to go bad, and put high grades on it knowing there was a real risk of default. I would put it to you that actually no one inside the system believed the loans were going to go bad, and the high grades were put on it in the incorrect belief that there was no real risk of default.
There's been a lot of research on this subject, but to me the most compelling is research showing that "insiders" did not diversify away from the crap they were selling. See facts 9 and 10 from this paper[1] for example, which notes that Lehman, for example, was using internal models which estimated the odds of a bursting bubble at 5%, whereas they assigned 30% odds to scenarios that assumed absolutely impossible and historically unprecedented house price growth. Bear Stearns executives invested heavily in their own bank's funds which in turn were invested in MBS. Similarly, many loan originators got into trouble not because they bundled the loans up and sold them on (as their risk models required them to do), but because they kept too many of them, thinking they were great investments. If Lehman and Bear Stearns and Bank of America were the suckers, who was doing the exploiting? (And the biggest winners were people like John Paulson, who had no clue how the mortgage market worked, and had never traded mortgages before.)
From where I stand, there is a compelling, coherent story of people being idiots, but the cynical, blatant corruption seems missing. If your model says there's a 5% chance of housing prices dropping, then your model is wrong, but your conclusion about the risk of impairment of AAA tranches makes a lot of sense. So the next question is, well, did you really believe that there was a 5% chance of housing prices dropping? And the people involved bet their careers and personal fortunes on that, so...apparently they did?
Edit: I highly recommend the entire paper, and in particular you may find "fact 12" of interest, as it gets into the weeds on how the ratings were actually calculated, why the ratings on MBSs ended up being mostly accurate, why the ratings on CDOs ended up being completely wrong, and why there was a difference.
> From where I stand, there is a compelling, coherent story of people being idiots, but the cynical, blatant corruption seems missing.
Yeah, like I said, its closer to opinion than verifiable fact.
However, the paper basically admits the facts I use to conclude that are correct but draws the opposite conclusion by comparing incestuously rather than across verticals.
Yes, relative to other mortgage related things, the ratings were correct but such ratings are published as absolute probabilities and not as relative assurances that AAA > AA > A > BBB. The fact the absolutes were in error for a single portion of a single vertical among multiple firms implies its an incentive problem which the paper agrees. It just draws the line in a different place because its analyzing a different question [Insiders vs. Outsiders rather than Insiders vs. Insiders].
> The answer may well involve the information and incentive structures present inside Wall Street firms. Employees who could recognize the iceberg looming in front of the ship may not have been listened to, or they may not have had the right incentives to speak up. If so, then the information and incentive problems giving rise to the crisis would not have existed between mortgage industry insiders and outsiders, as the inside job story suggests. Rather, these problems would have existed between different floors of the same Wall Street firm.
The paper believes its internal to the firms but the uniformity of the problem between firms implies it was the culture of all the large firms and likely exported to the credit bureaus due to heavy competition in a single vertical. The difference between what the paper is arguing vs. what I'm arguing is likely impossible to determine because of the incestuous movements between those firms and even the credit bureaus of the human capital involved.
Discussing the merits of HFT is similar to discussing the merits of different approaches to cryptography. Because there's such a vast amount of prerequisite knowledge required for analysis, one either has to slog through the learning curve or look for expert opinions.
So who are these experts in HFT that should be referenced? Certainly not me or Michael Lewis.
One source I like to follow is
Eric Hunsader. One of his claims to fame is catching a leak in FOMC data that led to a policy change by the Federal Reserve [1].
My own take on the HFT debate: HFT is here to stay; the method of trading is not the problem, the danger is that no regulator can monitor the massive stream of quotes and trades and accurately sift out bad actors. What's the solution -- I have no clue, but arbitrarily slowing down trades is not useful.
Apparently no true Scotsman would follow Nanex as a source. Duly noted. :) I'd be interested to hear of other sources that at least attempt in depth analysis like Nanex.
Because there's such a vast amount of prerequisite knowledge required for analysis, one either has to slog through the learning curve or look for expert opinions.
Thomas, Erin, and I possess voluminous experimental evidence that suggests generalist engineers can know more about market making than Michael Lewis (and, if one assumes Lewis is not committing malpractice with regards to framing quotes, Brad Katsuyama) in under two hours.
This is not a commandingly high bar. An order book is just a data structure. It tends to have a lot of money associated with it, but that is also true of at least some hash tables, and you understand hash tables.
I am frustrated with HN discussion of HFT occasionally for the same reason I get frustrated with HN discussions of some privacy technologies: there exists a political valence to the discussion and some participants treat that political valence as carte blanche for just ignoring that we are talking about actual codebases which have actual rules which can be described in words, including words which definitively falsify some claims about them which feel right to people due to political valence.
> Thomas, Erin, and I possess voluminous experimental evidence that suggests generalist engineers can know more about market making than Michael Lewis (and, if one assumes Lewis is not committing malpractice with regards to framing quotes, Brad Katsuyama) in under two hours.
After which level in Stockfighter is it likely that I know more about HFT than Lewis?
Depending on how exactly you decide to solve it and how good of a mental model you have about what your code is doing, level three. Again, a high bar Lewis' level of understanding is not.
I know it feels like this is an abrasively combatative statement from me, and it's quite uncharacteristic about any topic other than Bitcoin, but if you've coded a trading system and then read Flash Boys it's like several hundred pages of "I used a GUI to query optimize the traceroute with MongoDB and a red/black tree." Those words, individually they have meaning, but in that configuration it is resoundingly unclear that the speaker understands what is going on. It is resoundingly unclear that Lewis understands what an order book is, what order types are (other than (paraphrase) a mechanism by which sophisticated operators cheat mom-and-pop hedge funds out of their hard-earned 2 and 20), etc.
If you'd like this critique at literally book length, read Flash Boys: Not So Fast. It's brutal. It's a point-by-point and page-by-page refutation which brings in lots of crunchy detail (which Lewis scrupulously avoids), quotes extensively from experts (including ones who would be incentivized to say the opposite thing except for a respect for the truth), comports with the understanding of our informal advisors, and does not require me to suspend belief in core principles of math, physics, or computer science, which Flash Boys does multiple times.
Pretty much any of them. He interviewed zero HFTs for his book and seems to repeatedly misunderstand how the basic concepts of order books work, or things like "someone trying to sell a million shares is going to drop the price of the stock", which you learn on level 3 at the very latest.
The second order problem is how to select which "experts" to believe. In my opinion, Hunsader sees conspiracies where none exist and interprets all unusual or hard-to-explain behavior as evidence for his existing beliefs. (Not to say he's never right, of course). But you have no reason to believe me, some random guy on the internet, and in general most experts on the subject are going to have some kind of conflict of interest.
Yeah....most people in the know think Hunsader is full of crap.
He certainly sounds smart to people outside of the business of HFT, but he's not quite the droid everyone seems to think he is. His opinions of HFT and the various market microstructures that evolve are quite hilarious to anyone with a trading background.
You can see part of what's problematic about the latter book just from this Guardian summary of it. The summary would have us believe that HFT is a trick employed by giant Wall Street banks to screw over smaller players. But Katsuyama, the protagonist of Lewis' book, was a big-bank trader, paid exorbitant amounts of money to tax large block orders by pension and hedge funds, and outraged by his inability to compete with algorithms that were bidding that tax down.
I'm uncomfortable calling any giant financial firm "trustworthy", but if you were going to pick one firm that has come close to earning that label, it's Vanguard. Vanguard's chief investment officer has repeatedly and unequivocally stated that HFT firms have lowered Vanguard's costs and improve outcomes for its investors.
Here's Matt Levine writing entertainingly about how the "Flash Boys" in Lewis' book ought to have been, based on his previous books, the protagonists, and Katsuyama the villain:
http://www.bloombergview.com/articles/2014-03-31/michael-lew...