Bloomberg On the Markets
The more I am in this game, the more I realize it is all about relationships, organization skills, lawyers and information. When people start talking about AI and algorithms and other quant bull shit, my eyes glaze over. How is a computer going to call up company XYZ and ask them what their strategy is to make EBITDA increase, and why their capex spend is 3 MM this year on projects, and do they have revenue locked up for those projects, and are the 1st lien holders organizing. The human brain is a magnificent machine, but maybe even more important than that is the society aspect of being a human. Anyone that trades on computer data, in my mind, is a moron. You can show me returns from Stark and others for the last 10 Million years, and I will still not be persuaded. I sort of believe in the efficient market, but as Gordo says information is the most valuable commodity, and I do not see a computer coming up with the information. AI is dead. Computers fail at the worst possible times.
posted by MattKelly54 at
10:22 AM
13 Comments:
We put a quant trade on this year. It worked quite well. Of course, sample size=1 is not ideal.
Jim Simons seems to be getting the computer to do ok.
I think that the best investors of our time will be both quant and fundamental.
I think the best investors will be fundamental. Quants are investors but asset allocators, quantitative trading is based on first and second order regression models developed by humans based on historical returns. Problem with that is that I believe, as most do, that you can't make money off past trading patterns unless they offer arbitrage opportunities, and the problem with arbitrage opportunities, which do exist (SEE Case of the Penny), is that they tend to be arbitraged away and leave the arbitrageur lookinng for new arbs.
As competition and numbers of hedge funds trying to do this have drastically increased in recent years historically great strategies like convertilble arb have seen its annual alpha whittled away.
I have found companies like RGC, FFEX and TRBR - companies that the investing public forgot about - as great alpha producers for my job and my PA in the last year. I think the best way to make money is to do your HW (whether that is understanding the industry, the macro, the idiosyncratic, etc and its relationship to the stock value) and then take a position. I am not a big fan of quantitative trading or quantitative investing as a LT viable strategy as I think that the world is dynamic and that regression models based on the past will break or see the alpha driven down so that they are not effective LT vehicles.
What is the situation in ASAP? How Many don't have jobs? Who has the best job? Who is scrambling?
i think the best investors of our time will post returns and earn a track record before hijacking and self aggrandizing on chat boards.
I think Paul is a funny dude.
I have this crazy view of the ASAP students cranking paper in to a box, and waiting for 30 minutes for the Cray esque computer to spit out one name.
Matt,
Your comment reminds me of Mark's CB Predictor - hit the CB predictor and it will tell you where the stock is going to go. These guys are bozos.
When the computer can file a 13D like Loeb that is when I will go long AI.
1 guy is still scrambling a bit, the rest of us have a home. Two in NYC, Two in KC, Two in Madison then LA, and myself in Cleveland. The first-years had a crazy strong year - a few had 5 offers. Kevin got some stuff done and the internship hunt went pretty well.
Why are you guys so quick to reject this type of analysis? We know that 80% of PM's don't beat their benchmark (or whatever the stat may be...it's a majority). So chances are you are in that pool too. Me too. So maybe these guys that you're so quick to dismiss are laughing all the way to the bank??? I'm not sure...but one thing I've learned is that being dismissive is generally dangerous.
When a computer can negotiate a four way stop, I will go long AI.
Rich, I agree with everything you have said. The hedge fund that I interned at destroyed the market with quantitative strategies mixed with hard nosed research. Really in my mind "quant" is a way to hardwire efficiency into your process, but I do not think it should become your process. Maybe people can program networks to know when markets are going to turn or strategies are going to work. But I am skeptical that numbers are anything more than numbers and the more information that you add to a model the more likely it is to fail.
The computer cannot tell you that sunspots really do not cause the market to go up or down, it just crunches the data. I think possibly my original post did not explain my thoughts well, I was more talking about AI and not "quant". I stand by my original statements however, I do not believe that a computer any time soon will crunch through a conference call and tell you if the stock is a buy or not.
Maybe I am too old, but War Games was probably my favourite movie as a kid. In that movie all the computer did was crunch numbers. It did not realize that the tactical decisions that were accuratley computed could destroy the whole population given choices that the other side makes. It is like the Cuban Missile crisis. McNammara (the computer in this example) wanted to nuke the Cubans. Castro wanted Russia to nuke the U.S. knowing that he would die. They both thought that they had the right data given that they were in the fog of war. In the end the Russians and Kennedy took the less "logical" route and saved lives. I do not think a computer could solve that, and I do not think that a computer knows when to buy OSB.
There are two sides to every trade though, so I guess someone is always wrong. When will computers buy from other computers?
Also portfolio insurance was a great quant strategy? LTCM? And Buffet is a great quant guy? You can find any example to match your opinion. Maybe everyone is always right or wrong and we are biased to find examples that match our opinions. We should find unbiased computers to solve this debate.
I am not dissmissing anything. I would just like to see some proof before I buy your snake oil on AI.
If 80% of Portfolio Managers fail to beat their benchmarks it is because either 1) the definition of Portfolio Manager is being applied far too liberally or 2) whomever drew that conclusion is failing to adjust for risk.
For what it is worth, it is impossible to invest in a benchmark. Even index funds (open- or close-ended) have management fees, which means that by definition they underperform their benchmarks.
Finally, this is based totally on anecdotal evidence, but I don't think the "miss" is distributed normally. Rather, you have a helluva lot of guys who "miss" the benchmark by 50 basis points or less, but on the other end you have guys who are absolutely crushing the benchmark. Almost no one underperforms the S&P by 20%.
Good points, guys. Love the wargames reference...it WAS a good movie and I think the analogy is about as perfect as one can get. So, I completely agree, and I would also be short AI because I agree, computers can't parse through calls. Quant strategies, I'm not short, but agree that they shouldn't be your process...they're a tool.
Say hello to Jim Albers and Tom Utula when you get to Cleveland.
Post a Comment
Subscribe to Post Comments [Atom]
<< Home