Hacker News Reader

> here's no outside investors in the Medallion fund anymore, so if it is a scam, they would only be scamming their own employees

The "it's a scam" theory that I heard had nothing to do with it being a Ponzi scheme, instead it was about laundering high-tax income into low-tax capital gains while maintaining a plausibly deniable investment cover story. It didn't have to beat the market, it just had to beat regulators.

I have nowhere near enough domain competence to comment on the plausibility of that scenario, other than to say that I suspect if it were that easy everyone would be doing it, so on those grounds alone I suspect the conspiracy theory I heard was incorrect. Someone more invested in the idea that it's a conspiracy would likely just argue that it isn't easy to be good at this or that there are economies of scale, so only one dominant player. Shrug.

> ... high-tax income into low-tax capital gains while maintaining a plausibly deniable investment cover story.

There are public swap-based ETFs (in Canada) formulated around this concept:

* https://www.pwlcapital.com/understanding-swap-based-etfs/

* https://en.wikipedia.org/wiki/Total_return_swap

But how exactly would this laundering work ?

I think he means that they funnel the income from fees into this "black box" fund and that's what creates the huge returns. My instinct would be that there's no way this would get past auditors, but then again, Madoff got away with cooked books, so it's plausible.

The book actually covers this. What they did is somewhat complex and involved obtaining leverage as much as the tax savings but basically they turned short-term capital gains into long-term gains. That saved something like $7bn, iirc, out of the $100bn in net profit they have generated...so it is not significant.

> What they did is somewhat complex and involved obtaining leverage as much as the tax savings but basically they turned short-term capital gains into long-term gains.

That sounds like misdirection. It's very common for hedge funds to use tons of leverage. It's not like getting this level of leverage is an unsolved problem.

On the other hand reducing your tax rate from 37% to 20% increases your returns by 27%. Even that would be valuable, but the real number is probably even more than 27%.

> A great book about RenTec and Jim Simons came out recently:

A podcast interview with the author:

* https://awealthofcommonsense.com/2019/11/talk-your-book-greg...

* https://awealthofcommonsense.com/2019/11/non-intuitive-lesso...

Sounds like you've assumed that rentech are running the same volatility as the S&P 500. That's very unlikely to be true - most systematic hedge funds ran crazy high risk in the '80s and '90s. But even if you assume it's a coin toss as to whether they perform well in any given year, twenty good years in a row is impressive.

My best guess is that it's a combination of luck, skill, and hindsight bias. Rentech probably had (has?) an edge. They were also partly lucky: in investment, you can do everything right and still lose money. There were also many other players: had LTCM not blown themselves up, we might be marvelling at their investment prowess now.

In the book (“The Man Who Solved the Markets”), the author actually provides the Sharpe ratios for their flagship fund (“Medallion”). Medallion’s Sharpe Ratio has almost never been below 2.0, and has been as high as 7.0 (!!!) in some years.

I’ll link to the Wikipedia page below for the math, but these figures would highly suggest that their returns are much less likely the result of luck, and more likely the result of some edge/alpha generation.

https://en.wikipedia.org/wiki/Sharpe_ratio

Well I'm saying it's both skill and luck, combined with hindsight bias.

They're famous because Medallion is the best performing quant fund. Even if quant funds were just chumps tossing coins, if you pick the best of a number of them, you'll get something that looks good.

Now that maths still doesn't stack up that they were just lucky - they probably had skill too. But I think it's more likely that they had a modest edge and were lucky than it is that they had a large edge and got an average result.

Thing is, with the amount of leverage they are running, it's more luck than skill.

Luck means making big returns, skill means doing it with low volatility and high Sharpe.

The point of Sharpe is that it’s invariant with respect to leverage. Anything with sharpe 3+ is extremely impressive - that’s 3 sigma, assuming normal returns that’s like 2% tail event.

Yes, but Sharpe calculated ex post is also subject to survivorship bias.

For example, you might make a (successful) bet that works only in a low-volatility environment; if it works, your ex post Sharpe ratio will look ingenious.

The point about survivorship bias is a fair one.

But, if a given fund (Medallion in this case) is able to thrive/survive during the 1998 Russian financial crisis (which felled LTCM), the 2000-2004 tech bubble bursting, the 2007 quant quake, the 2008 financial crisis, and 2010 flash crash, it would seem to suggest that their statistical arbitrage strategy performs well in all manner of environments.

After their very hefty 5/40 fees, their worst year between 2001 and 2013 was a 21% gain [1].

I understand that there are all manner of epistemic and mathematical problems that prevent us ever from completely disentangling luck and skill, but a conceptual framework like "The Superinvestors of Graham and Doddsville" would seem to apply here. I.e., after a long period of consistent overperformance, the case for skill starts to look much more likely.

[1] http://archive.is/JdQiw

Sure, but volatility is not, hence the "low volatility" part of my comment.

Sharpe isn’t invariant with respect to leverage. Volatility drag implies that a leverage will always decrease your Sharpe, unless your volatility is 0.

You really don’t know what you are saying. The leverage is to boost absolute returns. Sharpe isn’t changed by leveraging. They are automated black box strategies making 1000s of trades a day with massive undeniable statistically significant information content.

Sharpe is actually changed by leverage due to volatility drag. Your Sharpe ratio gets worse and worse the more leverage you apply

> You really don’t know what you are saying.

There's no need for the aggressive tone.

> Sharpe isn’t changed by leveraging.

Indeed, volatility is, however.

A change in volatility implies a change in Sharpe due to volatility drag. This only applies if you are compounding the returns though.

Given that they're selected for being the very best returns, it's probably a decent (but not completely exceptional) combination of every factor- skill, luck, laundering, connections, etc.

https://www.lesswrong.com/posts/dC7mP5nSwvpL65Qu5/why-the-ta...

That’s true I’ve assumed they have the same volatility. But since RenTec probably is market neutral for equities so I would imagine their Sharpe ratio is much higher than the S&P 500. And their Sortino ratio is most likely insanely high.

> There were also many other players: had LTCM not blown themselves up, we might be marvelling at their investment prowess now.

This is nonsense. Rentech makes 1000s+ of transactions per day across numerous asset classes. It is 99% skill. It’s black box automated.

I'd be curious what their risk-adjusted returns are, especially in a leverage free environment. Anybody know where to find such information for RenTec (or other hedge/mutual funds for that matter)?

idk about medallian but capacity-capped HFAT funds can be near zero risk. I remember one HFAT firm went public and their S1 filing revealed they had lost money on like 5 days out of 5 years. It's not really 'prediction' or 'investing' in the sense generally implied

According to the book, they were running a SR of 2 through the 90s and early 2000s. After overhauling their strategies, they started running a SR of around 7! I’m not sure where you can get information about other funds besides finding articles in BB or WSJ, but the HFRI Index publishes return information for different classes of funds. Regardless, I don’t think you’re going to be able to access SR unless you’re an investor or work at a portfolio analytics company.

Thank you. People cling to the efficient market hypothesis like it’s a law of nature.

But how do you know these return numbers? Esp since as you say the fund has no outside investors?

Who has audited these numbers and what is your source specifically?

I‘m aware of Rentec for many years and also awed by the numbers but I‘ve always wondered where do people actually get them?

Every five - 10 years the up to date track record is provided to reputable third parties to curate publicity. For example, Bloomberg and WSJ. Generally speaking though, most firms with a track record like RenTech's prefer to keep it quiet because they don't solicit outside investment.

In particular: RenTech likes to keep a handle on this publicity for the purposes of courting extremely good talent from academia and industry. That's the only reason the Medallion returns are ever intentionally publicized. Anyone investing in RenTech's other funds knows full well they're not getting the alpha powering Medallion's returns.

Simons would probably have preferred to stay entirely under the radar, but the cat's out of the bag already and has been for decades. The most successful hedge funds (like the Princeton Alpha offshoots) don't court publicity because they don't need to pool risk with outside investors. They're not even hedge funds in the common sense of the term; they're proprietary trading firms run on employee and partner capital. Medallion can be thought of a prop shop within a hedge fund in that way.

Medallion also had a few external investors until 2010ish, who could be a source for some of the numbers. IIRC, Bill Ziemba published their returns in the mid 2000s sometime.

Also, what Princeton Alpha offshoots? IIRC, Princeton Alpha was a PDT spinoff that closed due to poor performance. Maybe you mean Princeton/Newport offshoots?

Yes, I meant Newport. The Thorpe folks and such.

Just out of curiosity, are there any successful Princeton/Newport/Thorpe offshoots aside from the usual one in Irvine?

Just the one in Irvine I think. Apparently they kept most of their staff from P/NP.

He probably means tgs and co

> We get a P<1.64E-14. Therefore, I think we all can agree that we can reject the null hypothesis of RenTec having no alpha.

The probability of my particular shuffled card deck is around 1E-68: vanishingly less likely. Therefore I think we can all agree we can reject the null hypothesis that my card deck was obtained at random. /s

Highly unlikely things happen all the time. Their occurrence should not be ascribed to alpha due to occurrence. The probability of occurrence due to occurrence is of course 1.

Well if you prrdicated such a permutation, that would indeed reject the null hypothesis. If 1.64E-24 doesn’t convince you, how do you manage to believe in anything at all? Thr world view you are positing is essentially solipsistic: there’s nothing that exists or is true except your own existence. Hopefully we all can agree that at best, such a view results in paralysis, at worst, absolute destruction.

Moreover, I don’t think you or anyone else actually applies your logic in your own life. If you did, you would be dead because of your disbelief in reality. For example, you migh conclude that eating is unnecessary, because the people who starved to death were just unlucky and that in fact, there’s been no “proof” that people need food at all. Or even proof that people exist at all!

Hold on, it's not comparable to the ~10$ CAGR of the SP500 is it? RenTec's fund is capped so it's an average return of 71.9% but not an annualized return, since the returns are not compounded. Correct?

* ~10%

That's right, they dividend out the profits each year (or at least they started doing that once the fund got over ~$5b or so). Otherwise if you do the math (1.66^33) they would end up with most of the money in the world!

The "it's a scam" angle is they're insider trading.

Not necessarily, they could be money laundering instead.

Continuing the speculation, they could be just "picking" the stocks after the fact when the returns are already known, iff there is no external record of their positions and transactions

There is an external record, 13F filings. Every quarter funds above a certain AUM have to disclose to the public their top long equity holdings. I’ve personally analyzed these holdings at my old quant job, they definately trade.

They probably have significantly better raw data sets than even their competitors like AQR or Two Sigma. Satellite data on parking lots. Credit card swipe statistics. Parsed and structures full financials. Supplier consumer relationships. Analyst data. Short interest data. Options activity. Fund holdings. ... Everything you can possibly think of. All integrated into a single streamline analytical and modeling technological infrastructure.

So yes to us little people it is insider trading. But I’m 99% sure it is real alpha at the core.

SAC Capital says hi

> And before someone says that it's a scam, there's no outside investors in the Medallion fund anymore, so if it is a scam, they would only be scamming their own employees. And if that was the case, I think we would know.

There are other possibilities. For example shifting profitable trades that occurred in their external funds to their internal funds. In which case it is partly a 'scam' and they would not be scamming their own employees.

They have massive alpha. Why is this so confusing to outsiders. Huge effort on detailed historical data sets combined with top of the world modeling techniques creates alpha, it shouldn’t surprise anyone.

They are being sued for back taxes to the tune of $6.8 billion. So this may not be as legit as it seems.

That $6.8 billion is nowhere near the total returns generated by Medallion over its lifetime. I agree the basket option scheme with Deutsche was questionable financial engineering, but the tax implications could at best augment the returns of an already profitable strategy. You cannot account for ~40% annual average returns over 30 years (after 5% management and 44% performance fees) using clever tax evasion.

> What they've managed to do shouldn't really be possible and is out of this world.

We don’t live in a world of wizards. When impossible feats are delivered on schedule for years without others figuring out is a hint that something is up.

Someday, we’ll learn about whatever combination of corruption, grift or other bullshit that was lurking behind the curtain.

Have you read the book?

My own suspicion is that they have access to some raw information that others don't have.

An example of such information would be information gathered from insiders at other corporations or industrial espionage.

It's more likely, in my view, that they've created a world-class industrial espionage outfit than that they've "solved the market" for decades where virtually everyone else has failed.

On the other hand, if they really are cheating, I'm not sure why the US government (who doubtless has penetrated them with their own spies) has let them continue operating for all these years.

I guess all the math and science PhDs they hire are actually undercover spies that finished their dissertations in order to provide a good cover?

Without getting into the validity of the gp’s post, RenTec does seem to recruit from NSA/CCR, particularly in the early days. Again, not saying this lends credence to the gp’s post, just adding an interesting bit of info.

A perfectly innocuous connection could be that intelligence community folks are good at ... drawing conclusions from intelligence? It makes sense to recruit spooks to analyze and interpret mounds of proprietary but legally-gathered intelligence then.

NSA's core competence is signals, largely making and breaking ciphers.

Intel analysis is more CIA/DIA.

My understanding is that this was generally true in the 1960s-70s as well. E.g., MINARET:

https://en.wikipedia.org/wiki/Project_MINARET

RenTech recruited from NSA early on mostly due to Jim Simons' connections there. The recruiting they do nowadays is mostly because there are some excellent mathematicians, statisticians, information theorists, etc. there.

I have no doubt RenTech has some of the brightest out there. However, any consistently high-performing funds warrant suspicion, especially considering how many hedge funds engage in questionable activities (corporate espionage under the guise of "forensics", hacking, insider trading). Showtime's Billions isn't as far off from reality as people think.

Do you think there aren’t other highly educated smart people working anywhere else in the financial industry? They’re success can’t be explained by just hiring smart people.

You could say the same thing about Google and Facebook versus Twitter or Snapchat. Whatever strategies they did find and implement likely worked or were massively improved upon using the historic data the firm had on how to position quotes, what trades weren't liquidity constrained, what were the server response delays between exchanges, etc.

Think about the quality between Google Maps and Apple Maps when Apple maps was first launched. Apple even hired a bunch of ex-Google Maps employees to improve their product, so it likely was using similar algorithms, but with worse data. Apple though had the luxury of gathering every iPhone user's location data though to build up their dataset. In Rentech's case, a competing hedge fund would not only have to build out the team and infrastructure, they would only likely lose a lot of money at first to gather enough data to have a similar edge. There were a lot of HFT firms like Jump, Citadel's unit, Getco that generated very impressive risk-adjusted returns for a while, but I don't believe it was close to Rentech in terms of capital capacity and number of years.

> You could say the same thing about Google and Facebook versus Twitter or Snapchat.

You can't. Renaissance only hire ~100-150 people, their strategy is very much based on only hiring the best of the best. This is nothing like tech. Tech firms employ thousands of engineers, you have to employ average people if your demand is this high.

Jump and Citadel aren't trying to do what RenTech does either (it is very hard and very risky). They do a lot of HFT, afaik Jump is just a market maker, so they are more like tech companies in that they do not need geniuses. They just need competent bodies at scale who won't blow up the company.

There's almost no firm with a bar to entry as high as Rentec when it comes to hiring, and almost certainly none with the amount of people who have cleared such a bar. Some smaller firms still hire similar quality, but RenTec has been around for so long and always has attracted the best candidates. It's not as simple as saying "they hire smart candidates" it's that they've hired the outright best in the market when possible.