Greetings! I had some further thoughts on this question, so here I am replying to - myself? Ok. Well anyway... I gave the equity market short shrift in the last post, and didn't really address the research in individual stocks at all. So I'd like to indulge in a little idle speculation about why different researchers may come up with different conclusions about whether individual issues tend to trend or revert. Keep in mind that this isn't based on any research of mine - my work has mainly been with the Forex market, not equities. So as I said, it's just speculation on my part. So I'll contrast two researchers who have done statistical work in individual equity issues - Perry Kaufman as discussed above, and William J. O’Neil, the author of several books on stock investing and the founder of Investor's Business Daily. Kaufman came to the conclusion that individual issues tend to revert to the mean, while O'Neil's research led him to a momentum based approach - the "buy high and sell higher" approach of the CANSLIM method. So why did these two come to opposite conclusions? I know I've said this twice already, but I'm just speculating here. However, I think this is a very plausible explanation. I'm guessing that these two researchers were looking at qualitatively different sets of issues. I'm familiar with O'Neil's approach, and he was concerned mainly with fundamentally strong stocks - that's what CANSLIM screens for. Kaufman, on the other hand, may have based his research on a much wider set of issues - including OTC small caps, bulletin board issues, and maybe even pink sheets. So O'Neil found that his universe of stocks tended to trend. That's because he was looking at issues that were fundamentally strong - good current & annual earnings, new exciting products and/or management, institutional sponsorship, and so forth. For obvious reasons, a momentum based trend following approach works better here. But if Kaufman was indeed working with the complete universe of stocks, including the lower quality issues and even so-called "penny stocks," why would this tend to skew the result more toward a mean reversion behavior? Well, almost 20 years ago, I spent a year trading these kinds of markets full time, and I have a pretty good idea of why they tend much more toward mean reversion. I learned during this time that the way to succeed with these lower quality stocks wasn't with fundamental analysis or with technical analysis (at least not exclusively). It was with what I used to refer to as "scam analysis." Many of these stocks are often subject to "pump and dump" schemes whereby operators first accumulate large positions in these relatively unknown issues, and then (illegally I believe) manipulate price upwards through the spreading of rumors and hype (the pump). As the general public gets excited about the stock, driving its price higher, the pump and dump crowd is now selling out to them, profiting from the pump and leaving the public holding the bag after the dump. So the way we (legally) made money in this market was to understand that these schemes were going on, try to identify them in the early stages (using TA), and then ride the coattails of the scammers. We wouldn't get married to any of these positions or get caught up in the great sounding stories. To us, each of these penny stocks was a POS (Piece Of ...ummm, Sewage!! yeah, that's it). So back to mean reversion behavior. That's the kind of price action you see all the time with lower quality issues. Not just because of the pump and dump kinds of schemes where stocks would rise spectacularly then crash back. But also because of the more general psychology of speculators in this market. Even when there's no actual pump & dump scheme, small cap traders often get caught up in the "story" of the latest flavor of the week stock. As this catches on, the hype drives the stock up. But as time goes on and the "story" doesn't pan out (the wells come up dry, the mines produce nothing, the big invention doesn't sell well, or whatever), these stocks tend to come crashing down. And there you have it - big runs followed by spectacular crashes equals mean reversion. So there you have it. One researcher confines himself to a subset of stocks that meet certain criteria, and finds trending behavior. Another researcher looks at the whole market, including the lower quality stocks that tend to soar and then crash, and sees mean reversion. Just a guess, but it's a plausible one.
Fundamental analysis are undoubtedly significant in the Forex trading, but we can just agree with your opinions that technical are not reliable.
Hi, and welcome! If I gave that impression, it's my fault - I didn't really clarify in this thread that I'm not against TA at all. I posted links to a couple of my articles in the first post, but I really can't blame people for not reading through them - they're kind of long (especially the second one). Anyway, let me clarify. Here's an extract from the first article: "A final word about Technical Analysis I want to make it clear that I’m not one of these fundy snobs that dismisses technical analysis out of hand. I think the arguments for technical analysis make sense – one of the key indicators that FXTW uses to identify short-term momentum is Tick Density, a TA tool based on tick volume. The FXTW Model account also makes use of technical analysis for trade management – setting stops, levels at which to scale in, etc. So I’m a TA fan too – I just don’t think it gives the whole picture." So if you're saying that technical analysis and fundies are both useful, then we agree. The thing is, most traders have no problem finding and interpreting technical analysis - there's information on it everywhere. But! But when it comes to fundamentals, many traders just throw up their hands and give up trying to use it for a couple of reasons. It's too vague - it's hard to determine a directional bias Information overload - it's just too much to analyze. So traders know that fundamental analysis would give them a better picture of the market, but it's just too unwieldy and time consuming for them to use. That's the problem that FXTW tries to solve by taking large amounts of fundamental data and turning it into an easy to digest visual form. Like the site says - it's a funnel. Hope this helps, sorry about giving you the wrong impression about my view on TA, and as always...keep pipping up!
STP, I ran across this post from MrAgi1 from days gone by. It sounds like an interesting question that you may have good insight into.
There is nothing to debate about if we assume that price series is not stationary process which has no memory, we can find patterns in random walks however if underlying process is random our wins that support our trading system is just a random streak that may soon change to a losing row of trades with same probability.
Cool! Do you happen to remember if they were talking about movement in real time after a single news release? That would be my guess, as that's how "news traders" tend to operate. The answer is yes - it's something that a trader could research and compile stats on. For instance, one could look at the past 120 (10 years worth of) NFP releases and see what happened to the EUR/USD each time. Plot the data on a scatter chart and see if there's any correlation. The complexity comes in with the independent variable (the news release) though. There's the raw value, the difference between that value and the previous value, and the difference between that and the expected (economist's consensus) value. So which of these is most important? Which one to use? I've done some research on this, and if I remember correctly, I found that the biggest movements occurred when the actual number was in the expected direction, but even further in that direction than the consensus number. So for example, if PPI was previously 3.4% and the consensus was 3.6%, you'd get more movement if the number came in at like 3.8% or something. But if it came in at 3.0% you wouldn't get as much movement, even though (counterintuitively) this would be a bigger surprise. Nobody trade on this!! LOL This is my vague memory of studies I did years ago, and of course these are just statistical averages, which may or may not be significant enough to provide a trading edge. Readers of FX Trader's Weekly know that there's a whole section in each issue called "News In A Nutshell" which crunches each week's news releases into a single number for each currency, then tracks that by currency and by pair as the "News Release Index (NRI)." The NRI algorithm makes use of the actual, previous, and forecast values of each release as well. But it's an aggregate of the entire week of releases, not based on single releases in isolation. I'll be starting threads on a couple of FXTW's features including that one, and I fully intend to totally geek out on research on this type of stuff in those threads. It'll take some time though, as I just started a journal on here as well, so...so much to post and so little time in each day! But anyway, hope that helped and wasn't too much of a ramble. As always, keep pipping up!
The price series may have no memory, but emotional traders do. In a nutshell, that's the theory behind why some chart patterns seem to work.
Hello Scott, I don't that I have much use for your report at this moment but I do have a question, can monthly subscribers leave at any time?
Hi Bill, and welcome! Yes, a monthly subscriber can cancel at any time just by sending me an email at FXTW. The address is on the site and in each email that I send out with the latest issue (if the subscriber opted in for email delivery). The cancellation generally takes place by the next day. Also, since I'm an actual person and not just an automated system, I can take care of any issues that may come up. For instance, if someone cancels right before their next recurring payment occurs, that payment may still go through before the cancellation is complete. In a case like that, I'd see it and just reverse the payment as a courtesy. Hope this helps, and as always...keep pipping up!