Thanks very much lj. As it turns out, I have read that doc in its entirety; it's from stockcharts.com and supports what I say above with regard to the middle parameter, and so on. The author introduces a concept of a "full" stochastic, which is simply the ability to define a middle parameter (other than the default of 3) for smoothing within SlowK. I thought the author could have been a little less obtuse about that aspect. Otherwise, a very good doc.
This is a very good reference and it brings out about 5 reasons why the STOCH has been so helpful for many years. I started trading in '57 a few years after Lane did his creating. Cisco was a typical stock for position trading as you can see. We, meaning the folks I traded with at that time, did 6 to eight day trades and the moves were in the 20% range and we tried to knock off about 10% in the move. We used the WSJ to get high beta lists for plotting by pencil on brownlines. We blueprinted the brownlines using the IBM blueprint machines and interoffice mailing envelopes. B size blue print paper was used and it all was more or less automatically done. In that era this type of thing ws called a 'gov't job"; IBM, at th time had 80% of the world market so things were sort of pleasant in that rapid growth period. stock bonuses wre really spread around and many people had ownership succeeeding from the Standard Time Division (punched card time clocks). You can search an ATS record from a slightly later period (after PC's were invented) that showed the average results to be 11.1 % average return for 6.6 day average holds. It was out of Sunnyvale, Calif. Today, we are on the 5 min ES e mini insturment which ws not avaialbe in those days. Lane started with a stock tuned periodicity, the 14, 1, 3. Trading the DJXX became popular when information became available. When the satelites when up you could use a 36 inch disk as long a trees, etc weren't in the way and you mechanically adjusted the disk into position using single intensity meters. 30 and 15 min charts were popular and you just plotted as the day went along. I did two concurrent 30 minute charts that were overlapped by 15 minutes. All color coded and with projections for the next bar's H/L range (volatility). Cycling ws very apparent during the day. On our test day chart, you can see these cycles on the indicators. With the addition of the STOCH (5, 2, 3) and using it for Sentiment, we get an earlier signal and then use the slower indicators to sustain the HOLD as a way to have a Beginner level cycling trading going on. Here is a shape that is common to the trading cycle. B is the sentiment change signal part of the waveform. C is the "trending HOLD" signal part of the waveform. D is the profit taking "end effect" of the waveform. We will take the Excel and convert it to a binary signal chart that shows these three signals. Logic is used for this as a "certainty" practicality that takes the Signal to noise ratio to its limit by eliminating "noise". As has been pointed out, I am a poor communicator. Most people see that we have spent a lot of time and the ATS is not complete not profitable. Obviously, I am the weak link do to poor skills. Dealing with logic may be new to some of you. What is important is understanding stuff. Getting the understanding of the arithematic we are using is not a simple thing. By using the 80 and 20 on the STOCH (14, 1, 3) fast line we either have or don't have signal C or signal D. D is an action signal and C is a hold signal. On the STOCH (14, 1, 3) the B signal is just a transient during our present STOCH (5, 2, 3) 50% cross over signal. we can use a binary before/after indicating column on this indicator by setting up a column in the excel. going further you can see the A, B, C, D in the MACD (5, 13, 6) indicator waveform. yesterday I was unsucessful in communicating a relationship of MACD as a supressor for some STOCH signals. I will work through this again so it is more clear how the various ongoing cycles in trading ES can be differentiated. This four part waveform may be such that it enables us to go from analog to binary shortly. I tried to get the columns set up by "signal" in Excel in my first pass and that did not work. As may be seen by the dialog here, some people are learning or working together and others are not. In the end, everyone will get what they want. The code can be written in english as a rule set for the level of trading that any given person thinks in. We have three convenient levels showing now with the data and it's arithmetic processing. I will call these Beginner, Intermediate and Expert. With the advent of binary columns for the indicators along side the analog columns, we can just use logic to differentiate the three levels of trading. For volume, I failed to communicate the levels of PACE which are commonly posted on ATS lookup tables we are using. reinventing the wheeel of these scripts is not necessary, I believe. Do do want to set up some quick money making for people not interested in sitting at computers. So I need volume differentiated for that. Also I want to introduce risk control for people that cannot get out of the "freakout syndrome" of CW--PA trading. So I need volume differentiation for that. Some people have difficulty diffierentiating retraces from reversals. I would like to have that as an output a well. So I need volume differentiating for that. Two common lookup tables are: 1. PACE vs volatility, and 2. PACE vs overlap. Commonly, we have used a 1600 bar lookup table. Lookup tables in ATS's are maintained. Table maintenance is a common programmer coding skill. I feel it is possible to work at this level in ET, especially for those who are working together and learning or growing in trading skills. I know all of this is very old hat and went out of vogue many many years ago BUT it does provide a terrific way to become mentally differentiated with repsect to things like "tape reading" and trading with "sports memory". There is also a relationship to APL modelling and design. Software languages did not just come out of the blue nor did the computer hardware configurations that handle software languages. In old vacuum tube design day there was the same kind of flack you see in ET. a soldering iron and/or a wirewrap gun sloved a lot of testing problems by just rewiring cable shoe connectors to show, in test cells, bad hardware design and bad languaging approaches. Changing a mainframe provein from 2 1/2 days to 2 1/2 hours happened by wiring workarounds for bad science. We need to get the Excel columns to read binary so we can connect the wires from the new columns to software logic. To get software outputs to set up singals and trading platform feeds.
Latest CashCow automated trading system attached as a text file. Changes in this version New functionality: Context Supression Logic. Diagnostic Module. New enhancements: Better organized code structure for natural reading. Bug fixes: None. I'll post the corresponding chart next. (20081216430#30)
Here's the updated chart for 10 Dec 2008, which is the date we focus on to maintain consistent time series and to understand the effect of changes against a standard day. This chart shows the effect of changes of version 0.3 of the automated system. We do get some useful improvement and avoid some chop.
As a side note, on our test day, you see three "waveforms of the A, B, C, D components. Th go from one to another simply extend the D to meet A. This cycling of the market represents dominant advances (A, B C and D and non dominant retraces on a beginner level. When PACE is insufficient (V DU, DU and maybe Low levels) then time is stretched between the dominants by having a non dominant drift. When markets are "slow" this drift is usual the 4 o'clock drift.
I annotated it to look it over. I noted A through G stuff And I didn't get how you did suppression in you coing post. If you just snag the pertinent lines for my convenience and also tell me if they are generic (apply all the time where applicable). A For the open, Gapping is more significant for messing up indicators than the pre synch condition. Lets do this: Take any STOCH (5, 2 3) signal on the close of bar 2. And lets put up the last two days to check out the open. B Did the volume value @ D prevent taking the signals @ B? When we get the volume PACE lookup in the picture I will tune volume. C Using the 2 on smoothing may turn these three signals into one signal. We will see. That low volume (E) in the picture will go away too. Let's check and see if smoothing turns the F and G action into just one long Beginner level trade. General: Lets see if we can get 10 days up soon after version .6 or so.
You'd be surprised... depending on what you're trading and the time frame, the difference can be huge. My point is, just be careful about doing things in a back test that you can't do in real time. Not to mention that a system worth trading will still work well if action is taken after signals/conditions are met.
I'll add a PM BO next the complete the beginner 4 to 7 trades a day. Then we can do some hiking Regarding dominant and non dominant price movement considerations) on the Volume to pick up the Intermediate level. As we get into the Expert level we will want to consider the intrabar aspect of the indicators. This is not a fractal shift but instead using the coded software lapping frequency to take off STOCH, MACD, and PRV volume PACE. We will also have to have C (convergence) and D (divergence) binary indicator columns when we do intrabar or just before we get there.