Welcome back, Steve Just a quick question. Are you going to start a thread about data analysis? I've been playing around with a few ideas recently. I also ordered Sherry's book, but since I'm living in a distant corner of the world, it will take about a month for it to travel here. It's too damn long and I'm already running out of ideas
Hello Shel I will answer your PM question about "singles" here so that others can get that info too (if they are interested) Singles are simply "tpo's" or "time/price opportunities". They represent a single data point or "print" on a market profile chart. This happens when price moves from one area to another. If price does not retrace back to that (single) print, and a new bracket starts, that "single" tpo is called a "confirmed" single. In my opinion, confirmed singles represent the appearance of longer term participants in the market. They usually represent support or resistance. Unfortunately I cannot produce an image without changing my charts to mask proprietary data. You may be able to see examples if you go to Linnsoft's IRT website and looking at their videos. I hope this helps you. Steve
In answer to your question...no I am not going to start a thread on data analysis. I have a research project of my own that I have to complete and that for me is first priority.. If I have time at a later date I may start a thread on that subject. If you want to obtain more information on that subject I suggest doing more research on the Internet. For example, one could simply use "Google" and type in the question "How can I use Excel to analyze stock market data? Also you may want to investigate "YouTube" to see if there are similar resources of interest to you. Good luck Steve
Thanks for the response. I have no problem using Excel, so it's not an issue for me. I'm also a programmer and am able to code much more sophisticated stuff than it is possible to do with Excel (I've done at least 8 projects recently). I have some background in mathematics and statistics, but when I need something more subtle, I look for it and learn. Recently I caught an idea of stationarity which seems like a very important concept and you're one of the few people on ET who have mentioned it. Anyway, my interests are in stuff such as: what is the best way to measure stationarity, is it Chi-square or Kolmogorov-Smirnov or any other test; what is the best way to detect change points; how to segment data into stationary periods: this is what I've been doing last few days, tried graph clique problem approach, then brute force approach, but these are too damn complex problems and the algorithm takes days and weeks to complete calculations even after optimizations and only for 5 years worth of ES data (about 1200 sessions). Now I'm considering other, simpler approaches, but I know this isn't gonna end anytime soon. Were your experiences similar? If you're busy and don't have time, it's okay, I understand, you don't have to explain anything.
Indriomas A stationary time series is one whose statistical properties remain constant over time. What I do with intraday data is export it into Excel and use simple mathematical transformations to make the data either stationary or approximately stationary. Once you do that it is relatively easy to predict. That is because you can assume that the transformed data series will exhibit the same characteristics in the future that it did in the past. Once you have the skills to do this with a stationary time series, you can (if you want) work your way backwards out of the transformation process and obtain a prediction based on the original time series If a time series exhibits stable long term trend and it tends to revert to a mean after a "shock" (like the S&P500 does) it may be possible to make it stationary by "detrending". I make those assumptions when working with intraday data. What I do is export "intraday" data into Excel and "detrend" it by subtracting the the first price from the second, the second from the third, the third from the fourth and so on. I then divide the data into two parts (halves) and the result generates two (2) frequency histograms (probability density functions). If you compare them visually, you may see that they are similar. If that is the case, you can usually assume that the data is "stationary". If you are uncertain, you can test to confirm using Pearson's Chi-Square (sometimes called the quantile Chi-square test). I am gonna have to stop at this point because I am out of time Steve
Sorry folks, I have decided not to continue posting on this site. Good luck http://www.elitetrader.com/vb/showthread.php?s=&postid=2025015#post2025015 Steve
Steve, this was a good thread you had going here. Surely you're not gonna let the actions of one sponsor make you abandon all the people that were following you. Sure that sponsor was batty as a fruit cake, but you've been here longer than most. This is your home.... you helped build it! Now just take a day to think about it. You got your established thread here.... dont abandon it. When life gives you lemons, you make lemonade, when sponsor behave like that.... well that's when you sit back and enjoy the comedy hour that soon follows.