4 ms·
Pretty good article. It explains almost everything that is known about Medallion fund_ 1. Very research oriented 2. Their strategy does not scale. Fund has l
by nabla9 6y ago
Pretty good article. It explains almost everything that is known about Medallion fund_
1. Very research oriented
2. Their strategy does not scale. Fund has limited size.
3. They use some kind of arbitrage. No high-frequency trading, but longer.
4. Their current strategy must be kept secret for it to make money and it changes over time.
This is why "I have discovered fool proof way to beat the market" sales pitch is always a hoax. If someone has it, they keep it secret and make money. If everyone has it, it has no value.
- concreteblock 6y ago2. No strategy can scale indefinitely. 3. Every strategy can be viewed as a sort of 'arbitrage'. 4. Applies to every strategy.
- qeternity 6y ago> 3. Every strategy can be viewed as a sort of 'arbitrage'. No, it can’t. This is a bastardization resulting from the proliferation of “stat arb” to mean mispricing and is truly a misnomer.
- twic 6y agoWe have a running joke at work that our attempts at making money are "probabilistic arbs".
- concreteblock 6y agoYou're right. I was 100% thinking of 'stat arb' when I wrote that comment. What I meant to say that the way 'arb' is used is completely meaningless. Maybe a clearer comment would have been: "If rentech's strategy can be considered an arb, then every strategy can be considered an arb".
- qeternity 6y ago> "If rentech's strategy can be considered an arb, then every strategy can be considered an arb". Yep, agreed.
- fakedang 6y agoI don't know where the whole HFT thinking came from? Renaissance wasn't known for HFT, it was known for using erstwhile new technologies such as data analysis and some machine learning to find patterns between uncorrelated data (like the weather in Paris affecting LSE trades). I think all of those points are the same for all Quant shops. All of them are at stagnant AUMs now for a reason. I mean this article seems like even GPT-3 could have written it.
- bradleyjg 6y agoNot arbitrage in the literal sense. The article claims only a small edge over thousands of equal risk positions which means no trade, or pair of trades, is risk free.
- dcolkitt 6y agoI'll add that very few things, even those that are arbitrages in the theoretical sense, are truly risk-free. The classic example being the arbitrage between on-the-run/off-the-run treasury spreads. Bonds in the current series ("on-the-run") tend to trade at a slight premium to equivalent bonds that were issued earlier ("off-the-run"). Theoretically this is a perfect arbitrage. Short sell the expensive bonds and use the cash to buy the cheap bonds. Sit back and collect the spread. However, you have to understand why the OTR spread exists in the first place. Investors value the higher liquidity that comes with the current series. Consequently during liquidity crunches, the spread will blow out to multiple times higher losses. In these scenarios arbitrageurs face deep mark-to-market losses, margin calls and investors withdrawals. This very strategy was (one of) the "pure arbitrages" that blew up LTCM in 1998. The market can stay irrational longer than you can stay solvent. So, even what looks like a simple mechanical science in practice is an art requiring experience and judgement. You have to know the right leverage to use and at what times. You have to keep your finger on the pulse of the market and have a sense of when spreads are too tight or loose given macro conditions. You have to secure good funding relationships. You have to be smart about keeping powder dry so you can buy at cheap prices during dislocations. And so on.