Enjoy the charts and Excel file guys I'm off to do some shopping for stuff for my new home tomorrow. Going to try to get an early start. I'll try to check the thread for questions tomorrow evening. I hope the information I put out here helps.
I don't know if the Professor will approve of this or not but, Try this sequence: 160/640/2560/10240 Its basically the cool 4 square sequence (4,16,64,256,1024) but with an added zero!
I have spent the weekend playing around with nesting and placing categories into boxes. The common number is 7
ProfLogic, I think in this thread we have arrived at one interesting point of discussion that if we may - can discuss. You mention that you have stuck with the 7 square sequence and looking at it now on eurusd - it actually makes trading it a very different experience to the 4 square sequence. In the former, there are fewer trades, but usually the charts nail the trend more accurately and the trades seem to run longer too. In the later, it is catching alot of the same moves (plus a few more smaller ones), but mostly at different points in time due to the way the information is presented. Sometimes earlier, sometimes later than the former. On other products though (usdjpy) this sequence (7) simply doesn't show enough bars to make the trend chart work properly. Yet the 4 square approach makes it a wonderful scalpers market (i.e. going for the midsized moves). I realise some traders don't care how ticks form they understand stuff quicker than us mere mortals, however, for people who are obsessed with the "right look" on candlestick, oscillator patterns etc.- these matters are important in the way tick charts are presented. Finally, there is another related aspect to this issue that I feel is important to discuss. As we have seen elsewhere in this thread, some data feeds/settings will show the trend chart with too few bars (as MK pointed out) and other data feeds/settings will show it with too many. Are we not effectively curve fitting our whole trading strategy principally in the way our data feed is presenting the ticks? I could post charts from ten different data feeds (in spot fx there are 50+ different exchanges for the same product) and put in the same number i.e. 1024 ticks and get ten completely different results for the same product!. What happens if one of those exchanges/contributors goes out of business, damn it if I was getting good signals out of them
LMAO NZDspec Good question. Your comments about curve fitting is exactly what I was thinking of. However, it was a great way to consume the time on my weekend MK
The issue is perhaps even more fundamental than curve fitting. All approaches that claim some special properties for a certain number eg 14 bars for RSI, square of nines etc etc etc, seem to me to be based on the assumption that there is a hidden signal in the market that can be decoded by applying appropriate mathematical transformations to price/volume/time dataset. Whether any such signal exists or not is mightily difficult to prove, but what is certain is that it is an open question. The availablity of very very cheap computer power now and the ability to process vast search spaces very quickly must raise doubts.
I have been using ProfLogic's stuff for about a year in my own trading. Now I can say that exact volume increment is not sufficient. There is no "magic number" - 317V, 2401V, 777V etc. What is important is an order of increment size. 300V and 350V gives you the same picture but 3000V the different one. Tops&Bottoms and its classification (separate significant extremes from insignificant) IS important not the exact volume bar increment.