Spyder, My bad, shortly after posting I continued my reading and found your new method. I am a bit confused about exactly what you're doing. I"ve look over the posts that talk about this and cant figure out what data series the mean and std dev are based on. Are you taking a series of DU numbers to create the mean and the bands or the entire daily volume series? Depending on the answer, I may have an idea or two.
The Latest Version of our Hershey Chartscript, seeks the lowest volume occurring within a given cycle. The chartscript calculates this low volume bar for each cycle of 20% occurring with an individual equity. In order to calculate the mean, the Chartscript adds the low volume within each cycle and then divides by the number of cycles. In order to reach the High and Low Bands of Dry Up Volume Range, I used a multiplier equal to 3 standard deviations in an attempt to capture 99.7% of available data. Adding the product of the multiplier to Dry Up Average yields "High Band" Dry Up Volume. Subtracting the product of the multiplier from Dry Up Average yields the Low Band. I hope you find the above information useful. - Spydertrader
Thanks for the explanation Spyder. Let me explain what I intend to experiment with. It may take me a bit to do it, so if anyone else wants to try this out that would be great too. I have spent the past two years studying auction market theory (what market profile comes from). What I have found is that Jack's method fits auction theory perfectly. "DU" is, according to AMT, the market coming to agreed value. The break out is a directional move caused by a surge of disagreement on value. In AMT, the tool of choice is using the bell curve to measure value. My hypothesis is that it may be more useful to consider a volume data series of ALL days in the ranking period (ie. 6 months). Find the mean for volume and the standard deviation. DU would be defined as some number of standard deviations below the mean, and FRV would be some number of standard deviations above the mean. There is a certain logic to considering all the days, instead of just the DU's on the 20% cycle launch points. My thought is that DU in theory should be related to the whole picture of the current situation. If you consider DU as being derived from all the volume data, it becomes self adjusting if avg volume has shifted one way or another, for instance. As it stand right now, the idea that FRV = 3 X DU is somewhat arbitrary (not saying it's not useful). Also, the current method that you just described, uses only a handful of data points, which may or may not be distributed close to normally (which means the whole idea behind std dev's doesn't hold up). Please take this post only as an idea to play with. I'm not in any way suggesting what you have done is wrong or bad. Let's just say I have a hunch.
2005-05-24, Tuesday - SIGM Update Family obligations required I leave the computer before 11:30 AM this morning. At that time, SIGM traded at the $8.55 price point. Knowing SIGM planned a post market earnings release, I placed a very tight stop in order to automatically exit the trade while away. As a result of my tight stop, I exited the trade at the $8.50 price point. The resulting trade provided a .42 per share profit or roughly 5%. Our 2000 shares yielded a gross profit of $840.00 USD. http://tinyurl.com/aotru The above URL shows the earnings for SIGM and indicates a net LOSS for the quarter of -.03 per share. - Spydertrader
Bundlemaker, I was just thinking that your proposed method may not take into account the varying probability distributions of different stocks and you might need to factor that in somehow. Determining an appropriate confidence interval would be easy if all stocks followed the same distribution, but since some might be more fat-tailed than others, finding a confidence interval applicable to all might pose problems. I'm not a statistician so correct me if I'm wrong, but the current method seems to reduce the impact of that by focusing solely on the lows.
Capa, From my very preliminary investigation, I think you're right. I've manually gone through a handful of stocks. Summary statistics show very high kurtosis and very fat tails on the high side of the mean. Clearly, it aint' gonna be simple; but I'll continue playing with this a bit. One possible solution is to just look at the subset of data below the mean of the whole set or some variation on that.
2005-05-25, Wednesday - Lists Hershey Wealth-Lab Chartscript Culling Methodology Hershey Chartscript Scans / Qcharts Culling / Stocktables.com Sort Hot List DCAI ELOS FORD JUPM MCRI MFLX PARL SNDA USG Dry Up Stocks BCSI EXM EZPW MCRI TRGL Hot List Stocks Scores DCAI - 2 ELOS - 7 FORD - 7 JUPM - 2 MCRI - 1 MFLX - 5 PARL - 7 SNDA - 7 USG - 7 Dry Up Stocks Scores BCSI - 7 EXM - 0 EZPW - 4 MCRI - 1 TRGL - 5 Keep an Eye on These Stocks EXM (Attached) <img src=http://www.elitetrader.com/vb/attachment.php?s=&postid=755107>
Final Universe as of 05-25-2005: BCSI DCAI ELOS EXM EZPW FORD GDP HANS JUPM MCRI MFLX PARL SIGM SNDA TASR TRGL UBET USG - Spydertrader