Al_B on 5/1/2013 at 09:01
Interesting link but I'm afraid it comes across as extremely biased. Sounds like the guy has been very successful but I have to wonder when it includes:
Quote:
Using the seemingly magical number of pi, after realising that the 8.6 year cycle he’d discovered was exactly pi times one thousand, he was able to predict future significant events in the markets.
That comes across more as financial witch-doctor logic than rational economist thinking.
scarykitties on 5/1/2013 at 11:47
Quote Posted by Al_B
That comes across more as financial witch-doctor logic than rational economist thinking.
Agreed. I mean, apparently his stuff is correct, but the idea that it's calculated by pi like some kind of horoscope is... odd.
faetal on 7/1/2013 at 13:17
If the pi thing matched enough times, you could hypothesise that financial markets undergo some kind of pseudo-natural algorithm, but it'd need to be thoroughly tested. All evidence I've seen thus far are that markets are totally unpredictable.
Some interesting studies by Daniel Kahnemann for example show that in 20 years of records given to him by Goldman Sachs, the correlation between investment bankers and successful investment decisions is something like 0.001, e.g nothing at all. So you could replace all investment bankers with computer programs and not see any difference. Would save a lot of money in ridiculous bonuses as rewards for gambling I'd bet.
Al_B on 7/1/2013 at 19:00
Have you seen just how many upgrades computers demand as it is? Don't start giving them ideas or the economy will be totally wiped out.
demagogue on 8/1/2013 at 03:02
I saw some news stories that a lot of investors are using programs and just tweaking their algorithms. I think the catch wasn't just that the performance could be just as good, but with a program they can do a few 1000 transactions at the same time & time them by the second, and they get the first jump when there's new information because they can respond to it within tenths of a second or like within the opening second of the new trading day. I should probably get in on that. Pays better than poker bots.
faetal on 8/1/2013 at 12:11
The point is that there is no relationship between variation in human skill (assumed to exist in a large number of investment bankers) and investment performance. So even tweaking algorithms is unlikely to have much effect, since markets cannot be predicted.
Another study, I forget which, showed that poor investment performance is also positively correlated with (not sure if causative, but seems likely given simplicity of variables) switching frequency. So it would seem that the best investment strategy would involve picking some random funds and staying invested in them as long as possible, while obviously switching out if e.g. one of them is gong to nose-dive FOR SURE. But even then, you have a good chance of losing money. I would never subject my money to the market personally. All of the evidence points to it essentially being a more austere form of gambling.
demagogue on 8/1/2013 at 13:59
Some things can be predicted if you could get a time-advantage before the equilibrium settles. If there's some disaster, you know the price of gold will spike before people flood to it and it goes back down. But if you had a computer that could make a transaction in the first tenths of a second, you can cash in that way. The article explained it in those kinds of terms.
faetal on 8/1/2013 at 14:09
That's a very specific scenario. I'm talking about investment banking in general. If you could reliably make money every time there was a natural disaster by having a computer program buy as much gold as you can and then selling it for the higher price, then eventually, you'd reach a situation where the price differential would end up being dampened because of its predictability, as the price would increase far faster the more people exploited this, meaning fewer people would buy, which would shift the balance towards most of the sales being to people hoping to exploit it and then realising that they were operating in a buyer's market, which forces prices down again.
CCCToad on 9/1/2013 at 01:21
Quote Posted by faetal
Some interesting studies by Daniel Kahnemann for example show that in 20 years of records given to him by Goldman Sachs, the correlation between investment bankers and successful investment decisions is something like 0.001, e.g nothing at all. So you could replace all investment bankers with computer programs and not see any difference. Would save a lot of money in ridiculous bonuses as rewards for gambling I'd bet.
The question I have is does "successful investment decisions" mean good for the banker or good for the investor? As Michael Lewis put it in
Liar's Poker, "What was good for the bank was seldom what was good for the client".
faetal on 9/1/2013 at 11:48
It means simply that the amount invested gave a return and it measured the magnitude of that return. So the sum of all net change was compared to who was managing the accounts. There is no such thing as a skilled investor.
