|
| Early on the systems were based on predicting the short-term trajectory of futures prices. Then came along market neutral strategies - a very short term one akin to market making and a longer one with 6 month holding time (the closest to stat arb) |
|
|
|
 |
Jurassic
|
|
| Total Posts: 415 |
| Joined: Mar 2018 |
| |
|
6 months holding period for stat arb?
|
|
|
 |
|
| yes, that's the Equimetrics strategy. Not sure about it now but at the time it was based on empirical analysis of the relative performance of the minimum variance portfolio and the S&P 500. |
|
|
|
|
|
| any chance you can share those letters? |
|
|
|
Azx
|
|
| Total Posts: 46 |
| Joined: Sep 2009 |
| |
|
Up 76%
https://www.institutionalinvestor.com/article/b1q3fndg77d0tg/Renaissance-s-Medallion-Fund-Surged-76-in-2020-But-Funds-Open-to-Outsiders-Tanked |
|
|
|
|
zee4
|
|
| Total Posts: 90 |
| Joined: May 2010 |
| |
|
The only thing they have in common is that they are operated using the same software and have the same senior management team. Everything else about them is uncorrelated.
 |
Бухарский |
|
|
deeds
|
|
| Total Posts: 521 |
| Joined: Dec 2008 |
| |
| |
|
pj
|
|
| Total Posts: 3670 |
| Joined: Jun 2004 |
| |
|
> same software?
Excel 32 |
The older I grow, the more I distrust the familiar doctrine that age brings wisdom
Henry L. Mencken |
|
|
errrb
|
|
| Total Posts: 12 |
| Joined: Oct 2007 |
| |
|
Taking the risks to sound silly, I assume that in this context "using the same software" means using similar framework to forecast returns/construct portfolios.
It's clear from what we know that time horizons for Medallion fund and big institutional funds are very different.
One of the questions which comes to mind is: Is there a proper separation wall between different funds? Is Medallion privy to the information about bigger trades coming from big institutional funds? Can it "frontrun" other funds?
|
|
|
|
|
|
My 2c.
I do not really know how does it work in such big institutions, but assume you run a strategy based on earnings results/estimates and offer short holding period (say, <= weekly) for small fund and 3-month holding for other (big) funds. Their predictions are not the same, but I would say there is a significant overlap. The shorter strategy naturally "front-runs" the larger one and gets juiced in overlapping positions, but there is no direct information leakage between the two. Moreover, those people could be naturally separated by a wall, but still trade "similar" things.
I can think of many similar examples tho. |
|
|
|
Jurassic
|
|
| Total Posts: 415 |
| Joined: Mar 2018 |
| |
|
https://twoquants.com/decoding-rentec/
best analysis ive seen of the medallion fund |
|
|
|
|
dlwlrma
|
|
| Total Posts: 15 |
| Joined: Jul 2019 |
| |
|
Anyone know about Enlightenment Research? Apparently they are run by two ex-RenTec guys who tried to leave to Millennium a while ago but that was blocked for a few years.
Now their website says they are in an agreement with a larger hedge fund - is it Millennium again? |
|
|
|
dlwlrma
|
|
| Total Posts: 15 |
| Joined: Jul 2019 |
| |
| |
|
errrb
|
|
| Total Posts: 12 |
| Joined: Oct 2007 |
| |
|
dlwrma,
I believe one of the guys who runs "Enlightenment Research" is a former rentec Russian guy, who ended up in court fighting Jim Simmons. He is bright and technically competent (to say the least). I believe python software which is wrapped around kdb was developed by his team? Not sure if the Millenium is supporting them though |
|
|
|
jslade
|
|
| Total Posts: 1250 |
| Joined: Feb 2007 |
| |
|
> https://twoquants.com/decoding-rentec/
> best analysis ive seen of the medallion fund
It's a shitty summary of the Simons biography and widely misses the mark, IMO. This thread reigns supreme. You're a good 'bot Jargon, but you show the classic failings of machine learning.
The book is quite interesting FWIIW, but more for Simons life and management style, and for the origins of some of the shitty internal drama. One of my biggest takeaways is "computer science" people are useful for reigning in the excesses of sloppy Ph.D.s writing S-Plus spaghetti, but if you allow their sperdo culture to take over your company, it's going to turn from an apparently fairly pleasant academic seminar feel to a dystopian tech company hellscape. I also thought Simons suggestion (I think mentioned in an interview) of "The Captain" by Jan de Hartog was pretty good for management and leadership hints for making heterogeneous talent work well together.
One thing which was obvious from the book: a great deal of the "information" gathered by the author was from axe-grinding disgruntled employees or former employees who had personal beefs with others (natural in any firm). Lots of the contributions of certain figures were obviously minimized, and some figures were just shit all over. I mean, that's what you'd expect: the ones who talk are pissed off, but an actual journalist would have done a better job of removing the obvious biases of his interlocutors. |
"Learning, n. The kind of ignorance distinguishing the studious." |
|
|
|
elf
|
|
| Total Posts: 38 |
| Joined: Mar 2009 |
| |
|
| Bleed continues: RIDGE down 5% in Jan |
|
|
|
Jurassic
|
|
| Total Posts: 415 |
| Joined: Mar 2018 |
| |
|
| @jslade im not sure ive seen anything better, even in this thread theres no ideas as to how better execution could work |
|
|
|
|
rickyvic
|
|
| Total Posts: 250 |
| Joined: Jul 2013 |
| |
|
What makes me laugh is the use of the word AI.... hehe
The stories that Simons tells MIT graduates are online, very entertaining.
My take on what I gathered from it and what I heard from other quants on them is that they were 20 years ahead of the competition and probably still are.
Looking at their returns they look more like market makers these days or hfts. They definitely do a lot of market making and they built some serious infrastructure.
Instead when I look at other highly profitable operations I see a lot of situations where people sit of their results or they move away from the firm, the firm stops or gets messed up.
R&D does not continue while probably rentech is rigorous on that.
|
"amicus Plato sed magis amica Veritas" |
|
|
|
New here. I've read this entire thread twice in a row, found it very interesting so thumbs up for that. I'm really not into conspiracy theories and even less so when it comes to RenTec, but I couldn't help but think of one. I'd like to share it with you guys and get your opinion on the feasibility (and not on the likelihood) of such a scheme. So here is the story...
Imagine that after 10 years of phenomenal returns, your edge starts to fade off and you start searching for new strategies. In those few years though, as success drew attention on your firm, you remained extremely secretive which created a mystique around you, and which in turn you exacerbated by portraying your firm as a place where only the brightest minds collaborate. Naturally, people start worshipping you and they all want to stack their money into your fund. So a brilliant idea comes to your mind, the new sought-after strategy you've been looking for. Instead of taking their money, you redeem all exisiting investors and close it to new ones. You do this because you will create a new fund which you will market as a high capacity smart beta style fund. This one, let's call it RIEF, will be open to institutional investors which will rush to invest because hell, it's Renaissance. So, you have your fund, Medallion, with only your money, and another one with others' money. Now, RIEF's true strategy is not important, all that matters is its size. But what is Medallion's new strategy? In two words: market making... But only for RIEF. Let me explain. Imagine RIEF wants to buy x shares of stock A for a holding period of one month. Medallion goes on and buys x shares of stock A at $P and sells them to RIEF at $P+s, s>0. RIEF doesn't care about the little spread Medallion pockets because they have long horizons. Thing is, they can't do it this openly and directly. So, whenever RIEF wants to trade, Medallion needs to trade at market and then position itself on the other side in the book. This immediately gets more complex for two reasons. First, they're now making markets so they need a good price discovery model (constantly adjusting orders to not get filled at a price worse than what they got in). Second, they want to get filled only by RIEF, which involves understanding and mastering order routing (provided that's even possible). So you still need lots of very smart guys to run this "strategy". |
|
|
|
|
cherk
|
|
| Total Posts: 5 |
| Joined: Aug 2019 |
| |
|
| This is likely what they are doing. They obviously can't directly cross between the two funds, but most of Medallion's profitability likely comes from picking off flows from slower moving asset managers/ hedge funds, of which RIDGE and RIDA are. |
|
|
|
rickyvic
|
|
| Total Posts: 250 |
| Joined: Jul 2013 |
| |
|
| This is what the financial industry does since the beginning of time |
"amicus Plato sed magis amica Veritas" |
|
|
|
chiral3
|
| Founding Member |
| Total Posts: 5233 |
| Joined: Mar 2004 |
| |
|
The stories I've heard, and these go back decades, have been fairly consistent, although I am just an old man talking about the past these days. I've mentioned them in this thread I believe. There were some hires back in the 90's involving building modules to optimize tx cost. Low latency -ive feedback high volume. This idea that there are a huge ensemble of positions that have been identified as having pxing anomalies. So this becomes this age-old issue that all of us have seen whereby this would backtest beautifully in-sample but looks like shit OOS. Really large N with a super small edge that collapses with too much money or velocity of money over small time horizons. So the large N gets you convergence on this small edge, the strat is capped, but large number of tx collapses the edge so it becomes a matter optimizing that aspect.
Whether or not Medallion scalps REIF or RIDA over different time scales... who knows? Even if that happens it would make sense that it's incidental. What's nuts is the amount of discussion over the purported returns of a $10B closed fund. I've heard other things through the years: order-book mining, tax arb with international trades, leverage distorting return figures... The fact that Medallion reports insane returns through this past year is almost another datapoint that the secret sauce isn't as market based as one would think, especially when you consider it's three highest earning years were 2000, 2008, and 2020. |
Русский военный корабль, иди на хуй! |
|
|
dlwlrma
|
|
| Total Posts: 15 |
| Joined: Jul 2019 |
| |
|
| @chiral - Well to be fair, those years are at the very least the best markets for HFT / market making strategies right? To me that's the biggest tell. |
|
|
|
|
dlwlrma
|
|
| Total Posts: 15 |
| Joined: Jul 2019 |
| |
|
| @chiral - Well to be fair, those years are at the very least the best markets for HFT / market making strategies right? To me that's the biggest tell. |
|
|
|
leftskew
|
|
| Total Posts: 25 |
| Joined: Sep 2019 |
| |
|
| Does anyone have an idea what RIDGE/rentec open funds did for Feb? Given the factor moves I saw it might be rough if they still have any exposure to them. |
|
|
|
|
