All of what you mention is distilled into a price and volume chart. Crowd sourced intelligence. It has the final say, so why not just observe it directly? It displays the shifting balance of sentiment in real time, multiple fractals in any timeframe. Another way of perceiving it, "When the market is up then the sentiment is up". Which is more true? When I'm in a great mood, the market goes my way. or When the market goes my way, I'm in a great mood.
Wow, I have to think hard on this one for a while. I don't have a ready reply. But I think it would have something to do with predicting the prices before they post. Seems like there is the implication in what you say that the price now is a predictor of future price, using patterns, but I'm not sure if you mean that. And that having current price you don't need the sentiment analysis or whatever to predict a future price. A good read, especially the part about how machine learning yielded factors nobody else knows about. Maybe NLP, or anything you can think of, could yield new predictive factors. But I'm too much of a newbie, I have to give some thought to what you said. https://www.bloomberg.com/news/arti...y-netflix-run-two-of-world-s-best-quant-funds EDIT: I'm going to have to figure out what you mean by this to really understand the implications of your post: "multiple fractals in any timeframe," and if it means price/volume at time x is related to p/v at time y.
There might be dozens of funds using it, but they're mostly very small, or a small proportion of the assets being managed by more conventional techniques inside a big fund. They generate a disproportionate amount of PR but this isn't a big part of the industry. I would wager that less than 10% of systematic hedge fund assets (about $100 billion out of $1 trillion) are managed using this stuff. Most of that is Rentech ($60 billion). When you take them out you're talking about an insignificant amount. GAT
let me give you an example that directly applies to trading: I trained a neural network to learn to detect when natural disasters strike. The idea is to be alerted as soon as a natural disaster strikes anywhere on earth. The detection is via tweets and the job of the algorithm is to detect with a high probability that the event is actually taking place (opposed to, for example, just someone talking or mentioning it). Next, the algorithm was trained to extract names of manufactured products that would be either more demanded or less demanded as function of the type of natural disaster. Next I built a database that links up those products with companies with the constraint that those companies and products must be affected not just by the disaster itself but specifically in the region it strikes. So, for example, when the Fukushima Tsunami struck and nuclear reactors were affected, that would not necessarily impact prices of bottled water or iodine tablets in France but it would definitely affect the prices of such products in Japan. As function of the above as soon as my algorithm detects a natural disaster it would attempt at finding products that will be affected and generate a long and short list of companies whose share prices might be affected by changing demand for such products which such companies produce or trade. The result is not good enough to have it generate orders autonomously but it is an awesome tool in my toolbox that assists my discretionary trading. Edit: I like to add that there is an edge to deriving this sort of information as fast as possible. One only has to go back to the tsunami example, it took many companies minutes to start reacting to the breaking news even though the Nikkei futures moved instantaneously. In fact I know first-hand that it was one of the best days in the lives of many index arb traders that made a living arbing nikkei futures vs index constituent names.
This could be any timeframe. These three fractal forms are always present. Trends overlap and are interlocking. The chart is incomplete. Volume gaussians complete the dataset.
That is not true at all. Please read my above post. Another example that directly contradicts your statements is the inclusion of Chinese companies in the MSCI Emerging market indexes just now. The inclusion benefits A-share listed Chinese companies whose names will be considered for inclusion. Their share price did not just jump but were in demand all day (Singapore listed A50 futures up nearly 2%, in a very tradable fashion, A-share names, impacted, are up a lot more). One would not pay attention nor understand the source of those price moves without knowing about this MSCI event. Obviously, if you just generated a long-short list of names that move up or down purely as function of price action then you will see that you will not be able to generate a profitable strategy when you include transaction related cost. But with knowledge of this event and with knowledge of share price behavior after such index events one has much more confidence that the upside momentum will continue with a high degree of confidence.
and you base all those numbers on what exactly? Hunches? What tells you that Rentech even puts a whole lot of funding into those strategies? For what its worth it may just be a few hundred million or so. By the way, Rentech by far does not have the same edge over its competitors anymore than it had before. In fact they closed down several divisions (among others, options market making) for lack of profitability. They are very equally affected than other shops. The times are long gone where Rentech commanded an almost magical competitive advantage over others.
You're right I have no idea what Rentech is doing, and nor does anyone else. However, there are good reasons to believe that they are or at least were running more in this space than the average fund is. The figure for the rest of the fund space is my own estimate, since I know quite a few people in the systematic industry, enough to give me a reasonable sample size (weighted by AUM). So 10% is probably an upper bound, and the real figure is likely to be lower if Rentech aren't, in fact, big players here. Like I said the PR around this stuff is massively out of proportion to it's impact. GAT
You mean in financial trading space? Because elsewhere it is shockingly underestimated and not feared enough. Ai will most likely impact our lives much more pervasively than the entire internet revolution.
This comment supports what I assert. I'm not exactly clear on what you are referring to? That a PV chart does not have the final say? When observing the markets the day starts, the price either goes up, goes down or stays the same. That is an effect observed. Much rational goes into explaining this behavior from day to day - all searching for a cause. The cause is experienced after an effect. It is an attempt to explain what is happening now. An analogy would be walking into a room, seeing a broken window, a baseball on the floor and a view of kids playing in the field. One experiences the broken window first and puts pieces together to formulate a cause that makes rational or emotional sense. If one looks at cause first, then each time a kid swings the bat, we are expecting a broken window. In consensus reality, although the above can be conceptually challenging, it is in fact how we process the world through our perceptions. Perception has latency as one of it's characteristics.