https://www.bloomberg.com/news/arti...-as-crisis-of-confidence-looms-for-quant-land https://www.elitetrader.com/et/threads/ii-blames-style-drift-for-cta-woes.322601/page-3 https://www.elitetrader.com/et/thre...-15-from-main-hedge-fund.313489/#post-4521361 https://www.elitetrader.com/et/thre...ghts-industry-woes.317672/page-2#post-4596931
“If sticking with them were easy, the threat of them being ‘arbitraged away’ would indeed be much greater, and nobody would take the other side,” he wrote. I call bull on that, sounds like famous last words... Those that bought in can look forward to di-worsification now
Looks like quants, algorithms and automated trading systems didn't do so well in October. Yet, I've been seeing these types of articles since 2015. wrbtrader
My personal opinion is, ( Excluding the bid/ask frontrun algo that is running on the switches hardware itself) a lot of the algos step in front of trains to make money. That didn't really worked in October
Well, this time it does feel different as you do see a fair number of stat arb players (which is what most people mean when they say quant) having several poor years in a row and then have October as a final kick in the nuts. The only star arb guys that seem to have done more or less OK are the ones concentrating outside the developed markets or doing something very niche. Stat arb is facing several serious problems and it's unclear what's next for that business. First of all, factor/style etc is a know game now and it is becoming increasingly crowded. Even more niche stat arb based on alternative data is becoming very commoditized by virtue of the data vendors hitting every fund on the street. I recall some head of a multi-billion multi-manager fund telling me that correlation between the stat arb groups in his firm is something like 30-40% on a regular day, which probably means that in a liquidity crisis that correlation will go to unity rather quick. The other issue is the amount of money in these funds. Stat arb trading is predatory by nature - they are not producing original ideas, rather they are trying to preempt or react to the trading done by other market participants. Once you have a large percentage of the assets driven in these funds, there are more stat arb money chasing preying on a smaller pool of original flows. It is getting to the point where faster stat arb guys are preying on the slower stat arb guys. Is it a dying business? Unlikely, though if we go through a proper quant meltdown like in 2007, that sector will probably shrink a lot, like the credit funds did after the blowup of 2008. Are the quant shops the canaries and the whole market is about to suffocate? Well, they certainly have been the last time around (i.e. 2007) so it might be prudent to pay attention.
Sle, it's more than that: the whole delusion underlying the quant bubble is the idea that data = information. People feel smart because they have sudden access to a large supply of eager programmers who can work with ocean vast data 9 ways to Sunday. The problem however is they can do that all they want, but what really matters is finding true edge/information, not data manipulation, and that takes intellectual honesty, which few have, too busy being blinded and seduced by the voluminous amount of bullship being thrown at the wall. The odds that these young programmers will suddenly stumble upon an edge are basically very close to zero. So they all crowd into simplistic ideas like factor investing, correlation type trades etc that stand no chance of surviving in the intermediate run, especially when the cycle turns
First of all, it's not really true, as statistical arbitrage in it's classical form pre-dates the whole data science revolution. This said, plenty of quant PMs/shops have jumped onto the big data train once their core strategies started to falter. As to data not being information - well, some of it is and some of it is not. People in other information-driven activities (best example is the intelligence services) have discovered it long time ago. The true art and science is determining what is noise and what's information in all that data (in finance and elsewhere). The crowding makes even real information less valuable. Is there information in knowing that more customers have shopped at Walmart based on the satellite images of the parking lots? Sure. Is it a solid trading signal once everyone else knows about it? My prior would be a "no". It's not any different than any other subdivision of finance, few actually have an edge and most are striving to find it. In fairness, these guys have originally "stumbled upon an edge" and have exploited it successfully for many years - even with the recent flat years and painful October the amount of money that has been made in stat arb over the recent past is staggering. It's is a mature field, there are books written about it. Multiple near-free vehicles allow people to invest in these strategies. People see that it works or at least it worked for a while, so these positions become further crowded which means it becomes further correlated with the market. PS. I've never done stat arb, but it's an important component of the market so I follow it as much as I follow the rest of the financial landscape (plus, I am envious of the 5th Ave or CPW apartments some of the stat arb PMs bought).
Most quants today are not doing stat arb, so that's not relevant to my post. By the way, the only firm that really knows how to do it is Renaissance, maybe one or two other names but that's it. This is the quants' problem (and the funds that hire them with high hopes) in short: Epistemological Fail
I am pretty sure you do mean the people I mean As I said, I am not a stat arb trader but understand the space well enough. A lot of people seem to have misconceptions about what stat arb is and how it's been transformed. How about this - let's not call it "stat arb", let's call it quant equity? If I had to guess, 60-70% of quant money is allocated in various forms of quant equity (which is what most people would call stat arb these days), another 15-20% chunk is in quant CTAs and the rest is evenly spread across the rest of the systematic space. I can ask a friend of mine who is an FoF manager for exact figures, but I'd imagine I am not far off. The non-HFT CTAs have been eating sh*t for a while now, mostly due to overcrowding. The quant equity guys are the ones that have been hit hard lately. The rest is insignificant enough to consider. Pray, do tell me even vaguely what you think Renaissance is doing? or any other quant shop, for that matter...
I am interested to know if human active managers outperform machine-driven algorithmic factor strategies in Oct 2018.