[amongwhales]

When a previous article on RenTec/Jim Simons was posted, I saw a comment thread about how Renaissance couldn't possibly be making such consistent money from Machine learning and quantitative analysis, but instead the reason they are so secretive and consistent is that they have insider info they have to hide. Do others agree with that theory? It makes more sense to me, due to a conviction on efficient market hypothesis.

[jandrewrogers]

No, and the reality is more interesting.

Their success is based on an unusual type of information asymmetry -- they have better mathematics than everyone else. They invest heavily in development of novel mathematics that can be applied to find new types of patterns analytically that are not discoverable with existing mathematical methods and algorithms. This allows them to mine patterns that no one else has discovered yet because the barrier to discovery is anomalously high since it requires esoteric invention. Most firms just rearrange the existing methods in new ways or with new data.

It is an efficient market but they are unusually profitable because they have little competition for the patterns they exploit.

[soVeryTired]

As I’ve pointed out before, there are a lot of quant firms out there. I doubt Rentech have the massive edge that people seem to think they do. I think it’s much more likely that many firms had a small edge, and we notice the luckiest among them.

That said, I don’t doubt that they’ve had an edge. Maybe they were doing trend following in the 90’s, or HFT in the 2000s. Most firms’ strategies eventually leak via lateral hires, but rentech is a small shop and no-one has managed to nab one of their employees. So there’s no way to know for sure.

[vasilipupkin]

They do have massive edge, but I doubt it comes from novel mathematics. My guess is it comes from very well designed systems at every stage of the process, honed over decades.

[whatok]

You don't notice the other firms because they've likely not been in existence as long and do not run as much money as Rentech does. Rentech has public pension money in it which by itself will attract more scrutiny. Chances are if you aren't running more than a few hundred million it could be all internal capital or a HNW individual or two.

[1024core]

> Rentech has public pension money in it which by itself will attract more scrutiny.

I thought Rentech was closed to all outsiders, and only internal money was allowed?

[whatok]

Only the Medallion fund.

[valuearb]

It's called being a "market maker" and they are best in the world at it.

[dharmon]

I think you should really think about the old joke: "Two economists were walking down the street. One looked down and said, 'look, there's a $20 bill on the sidewalk!' The other, without even looking down said, 'No way, if there were, someone else would have already picked it up'"

RenTec has been successful for decades, but the are regularly shifting strategies as the $20 bills are "picked up" by the market.

A sibling post asked you which variant of EMH you are "convicted" about. You should know that if it is strong or semi-strong, you have a conviction about the financial equivalent of tequila being good for teething babies or using leeches to bleed out various diseases. And the weak form is so useless I am not sure it really qualifies as a hypothesis.

You will have to search out the old fogies in economics depts to find the true EMH adherents. Even Fama professes it, but if you listen to the substance of his speech and read his papers, doesn't really seem to believe it. (maybe he thinks he'd have to return his Nobel prize if he denounced it?)

[QML]

So TL;DR the market isn’t as efficient as most would believe it is.

That makes sense since the market is ran by humans; but what if it was solely managed by a computer.

While reviewing for a final, I learned that computers are extremely fast at detecting negative cycles. An application of this? Arbitrage.

If you graph the foreign exchange market with each vertex being a currency and a directed edge being an exchange rate, applying the logarithm operator to each edge and then running a DFS transversal, it would be possible to find paths of arbitration.

[jdietrich]

What RenTec do is more akin to gold mining than finding $20 bills on the sidewalk. They expend substantial computational and intellectual resources in the pursuit of alpha. They're finding alpha in places that are non-obvious to the rest of the hedge fund industry, let alone the average investor. They're not eating a free lunch.

[avvt4avaw]

Quantitative trading is certainly an information game, but that does not mean that the information being used is insider information (which has a specific definition - privileged access to material, nonpublic information).

There is a large amount of public information about a company, and the source of edge in quantitative trading is

  - Identifying what kind of information is likely to move prices
  - Acquiring and processing the information as quickly and efficiently as possible
  - Making good forecasts
  - Combining the forecasts with a model for risk, financing and trading costs to form a portfolio with a high likelihood of positive returns

These are not trivial tasks. The third and fourth bullets are where traditional quant finance (what you read about in textbooks) has focused, and the difficulty of forecasting and portfolio construction should not be underestimated, but increasingly what differentiates a good from a merely average quant investment process is a relentless focus on uncovering and processing new information.

In the past it was enough to use highly structured, numerical data (market prices and volumes, accounting data, analyst forecasts) but the profit available from these sources has been diminishing, and the most important sources of information for forecasting stock prices are unstructured and non-numerical (e.g. text, speech, images).

Renaissance is a leader in its field, and has been for decades. It is perfectly possible for them to have produced the returns that they have, without using insider information.

[mlevental]

very few people on hn actually know about quantitative finance - so they co-opt folk wisdom in order to be dismissive. all quant firms are secretive - it's not like de shaw or jane street or two sigma publish white papers. rentech has been successful for decades. if they were doing it using inside info they would've been caught a long time ago.

[ordinaryperson]

Bernie Madoff began his Ponzi scheme in the early 1970s (source: https://www.forbes.com/sites/johnwasik/2012/10/03/inside-the...)

He ran a scam for almost 40 years right under the nose of regulators and Wall Street.

Rentech may 100% legit, but absence of criminal convictions is not proof.

[dsacco]

I disagree with respect to Renaissance in particular, but I think others will probably comment on that point adequately. Instead, I’ll address this:

> It makes more sense to me, due to a conviction on efficient market hypothesis.

I encourage you to walk back from that position, or at least reexamine which particular form of the hypothesis you believe in. Only one form of the hypothesis (the “strong” EMH) dictates that consistently beating the market is infeasible. Fama himself has since stepped away from that level of conviction, because there is no demonstrable mechanism by which the markets can price in all nonpublic information (and frankly, it’s silly to assert that the market does not have nontrivial inefficiencies at any given time, except for the most conceptually academic of cases). It’s a very hard pill to swallow, and it gets harder when you look at the number of funds that ostensibly beat the market without insider information (RenTech is the most infamous, but there are other firms comparable to it that like to stay under the radar).

The weak and semi-strong forms of EMH are much “neater” theories, both in intuition and empirical mapping to the real world. You can consider the weak and even semi-strong versions to be correct without precluding the capacity for certain exceptional firms to consistently earn better returns than the market average. It’s intuitively and demonstrably correct that public information is quickly “absorbed” into price consensus, but this is not at all clear for nonpublic information (though if insider trading becomes significant enough, it starts getting obvious to the market).

In contrast, accepting the strong form of EMH requires you to resolve two difficulties: 1) how do funds like Renaissance beat the market if strong EMH is true?; 2) what mechanism can empirically demonstrate that the strong EMH is in effect, as opposed to “merely” the weak or semi-strong EMH? You can try to solve difficulty 1 by asserting that any given successful firm is actually illegally trading on confidential information, but you still have difficulty 2. Further, you have the obvious follow up: if these firms are successful only by acting illegally, why has it gone on so long? Apply Occam’s Razor: is your explanation more reasonable than the alternative, which is that a minority of organizations are capable of accomplishing something very difficult but not impossible?

To circle back to your point about secrecy - it’s tempting to assume there is a decades-long conspiracy in play whenever rich companies are being extremely secretive, but it’s also a poor heuristic. A very simple and “Occams-compliant” explanation is that hedge funds are hyper-competitive, and go to great lengths to keep their intellectual property out of competitors’ hands.

[cepth]

It depends on which form of the EMH you subscribe to. I.e. weak, semi-strong, or strong.

As described in this article by Cliff Asness, https://www.institutionalinvestor.com/article/b14zbgrj5pflsc..., empirical research by Asness and others has shown there is strong evidence of “momentum” in markets.

RenTech and various other quant shops employ a number of strategies, among them momentum, arbitrage, etc.

Edit: also worth nothing that Asness was a doctoral student of Eugene Fama, perhaps the most famous proponent of the EMH.

[mjfl]

With all due respect to Asness, the returns from his fund and quant "factors" in general are complete crap compared to what the Medallion fund has consistently experienced over the past 20 years.

Most of Asness-type funds run "rotating" strategies whereby they have 30 or so different "flavors" (one invests in small cap value, another in US low-vol, etc..). Due to probability, one of the "flavors" is usually doing very well and they put that on the front of flip books. If any of them do badly, they "fire" that strategy and therefore ensure that all their running strategies have decent histories. It's basically engineering selection bias.

Quant factor funds are barely related to Renaissance.

[cepth]

I’ll grant you that AQR’s actual performance is middling. This is doubly true next to the consistency of Renaissance.

However, I’d stand by the academic research they put out. Especially because Renaissance famously eschews finance PhDs, preferring pure math and physics, AQR is still an industry leader in terms of the empirical studies they do.

The original comment in this thread asked about the EMH, I don’t think Asness’s returns as an investor should necessarily reflect on the academic research he’s done.

[mjfl]

perhaps the ratio of their returns to their academic output explains why Renaissance doesn't hire Finance PhDs.

[whatok]

Think about how much material insider info they would need to get their returns on their AUM. Then think about that on the timeframe the Medallion fund has been in existence.

[1024core]

For people who may not know the figures (these are from memory): In 30 years, Medallion has had only 1 down year, when they were down by 4.5% (this was around 1988 or 1989). In their entire 30 years of existence, they have averaged over 34.5% annual returns. Averaged.

[allenz]

I don't think that you would need that much insider information--you can just use more leverage. As a thought experiment, it's possible to earn similar returns by wiretapping any large business newspaper.

[whatok]

More leverage = much higher chance of getting caught.

[55555]

I dont think this is the case re: RenTec, but it does seem to frequently be the case for hedge funds more generally. Look into SAC Capital.