If it's backtestable basically, it can be said to have a reasonable basis. A lot of my business as a RIA rep is centered around suitability. If I can show the risk characteristics of a strategy, then I have my reasonable basis that acts as the foundation for taking any trades. It's required as a professional. If one could come up with a way to say these are support and resistance lines systematically, then, yes, I'd say that's valid. Discretionary trading lacks a reasonable basis, especially non-fundamentally based trading strategies. S/r lines I've not used in a really long time, but, in essence, the normalized volatility overbought and oversold levels are support and resistance lines, just a lot more mathematically intensive than drawing a straight line on a chart underneath or above the current price. ex-post edit: I actually think WL was created specifically for support and resistance lines. The floor broker pivots were the first published system to appear on the site. These are based on daily prices, and are different each day. This kind of trading has not been proven to be profitable on any site at any time frame. There is a weekly support and resistance system on a website that I think has a lot of potential but shall remain unnamed. A lot of the process into support and resistance has to do with extremum values, and, in the end, how do you decide what is a support and resistance level? The answer is always by looking at longer and longer time frames, starting from the shortest. In the end you'll find either the absolute lowest level or the absolute highest. Deciding what time frame is representative of the s/r lines is the main drag on these strategies.
Do you feel that fundamental considerations trump technical considerations? Do you feel that one must have a fundamental thesis, so to speak, underlying his or her trading decisions? So you would say that even technical analysis is valid only insofar as that analysis is the result of mathematical calculation rather than observable price behavior? Are broker floor pivots themselves mathematical derivitives obtained by manipulating various price points from the prior day? As such, Is it correct to say that these numbers represent actual observable price action? Is a support level a price region where declining price stopped declining? Is a resistance level a price region where rising price ceased to rise? If we choose to limit those price levels that we identify as support and resistance to levels that are empirically observable, then is choosing the time frame really such a conundrum? Would not an intraday trader will likely find the same support and resistance levels operative whether using a 1 minute or 5 minute chart? Would a particular support or resistance level be visible for a longer term swing trader whether viewed on a daily, weekly, or mmonthly timeframe? For example, is the price level on the S&P500 that has now been dubbed "the March Bottom" visible on a 5 minute chart, an hourly chart, a daily chart, a weekly chart, or a monthly chart as the very same price, or does that support level change when one changes the way one "packages" price into discreet units of time?
Fundamental analysis is generally required up until the point that you are calculating fair values as with options. Quantitative analysis requires examing a few different factors to arrive at a value estimate for the security. When this is not the goal, that you have formulas with which to base your decision, fundamental analysis takes over, but the real drawback is that it does not tell you when to buy but might only be useful insofar as it tells what to buy. Yes, it requires derivation if it's not fundamental They are price point manipulations, but it is not my opinion that they represent anything I would call price action. Price action to me is either momentum, support, or resistance followed by reversal. You can either chase, or wait for prices to come to you. I think that there is evidence that both work. Either way, this is always what we do in trading. By definition, that's correct, but the gotcha is really over what time frame, and even going forward you might find that price point changed. I don't believe they'd find the same 1 minute or 5 minute support levels if they were looking at different amounts of data. Whether we are talking about 1 or 5 minute, I would agree we might find the same price levels if we look at 20 days of data, but in most cases you can only fit so much observable data into the chart, so if you can only fit 20 days of intraday 1 minute data into a chart, there is always the question of whether you want to see 365 days of daily data, then 5 years of weekly, and in the end I haven't seen a mathematical relationship to building those support and resistance lines. As to the March Bottom, if you go back enough years by the same logic, you might not ever be sure that is the bottom. I found as my rational for holding with my clients through that period was more based on fundamentals, as nearly all stocks median book value had fallen to below 1, which means you could buy the whole market, sell if off, and keep the profit instantly. It changes as you pack different units of discrete time into them. As I said, you can only observe so much data in one chart. It is unlikely you could fit five minute data into as much space as a 5 year daily chart. That's not just a limitation when you go about setting your support and resistance lines by hand, manually, but also a limitation when you derive a formula to build those values. You must decide which time frame to look at, but you have to understand S&P below 750 is stupidly cheap on a fundamental basis. We might differ in how we determine our support and resistance lines, because absolute bottoms and tops are not how I draw support and resistance lines. I find where the data is clustered around a high or low level over long periods of time, rather than just placing it at the bottom. So, my experience has been whether you want to talk about the traditional floor broker pivots or the Camarilla equation that these systems do not work over long periods of time, and are very dependent on the period being tested in.
For those who may be observing, here are two people caught in a coomon basic problem. Establishing a system requires dealing logically and only deductively with the structure, the process going on in the system, and the results the system produces. All markets deal with a subject named non stationarity. there are two sides involved. These people have not dealt with either side. Instead, they look at two other sides which are not associated with non stationarity. I felt it was particularly interesting that a person running an ATS did not know it wasn't working. For those that have ATS's, what is a simple indicator that the ATS is working? Think in terms of non stationarity.
Jack, it's not that the system doesn't work, but that Fidelity's datafeed isn't working at the moment. The program runs fine when there's a datafeed. D asked about support and resistance lines, which isn't anything I have seen systematically implemented, and in the cases where I have, it has not been profitable. What's not logical about Scottd's system that you certified?
I do not believe you quite understand where I am going or what I am trying to discern from my discussion here. I certainly have no idea what you mean by "non stationarity." If you wish to be understood yourself, and if you wish further to interject yourself into a discussion in which you have heretofore been an observer, then, when you choose to use words to signify phenomena not conventionally used by those participating in that discussion, it seems to me that it is customary and polite to define precisely what the signifier is in turn signifying. So, unless or until you define non stationary, and how you believe the matter affects the discussion at hand, I would repectfully request that you remain an observer, and not a participant. If you do decide to define your terms in such a manner as to allow for a response, then by all means, contribute what you feel may be of value. Thank you
Hi, D, I agree it's unclear what he's discussing, but as it relates to mathematical instances of the use of non-stationarity it refers to the fact that variances across time frames are not stationary, and there's mathematical calculations to determine if a data series is covariance stationary. I'm not sure that's what he's referring to, but that's about the only thing I can think of, and that's only at Level II of the CFA Curriculum. What that misses is that this has nothing to do with regression. Sure we're fitting data to a time series, but as long as we don't try to minimize variances in fitting our data this has no relevance to the topic of either cash cow or PTQQS.
I think it does.... Your "pairs" system should be killing right now due to the *higher* variances, i.e. more deviation from value. Reasons why it is not are because either a) you haven't incorporated enough data in your analysis, or b) the model isn't able to deal with non-stationarity correctly. I suspect its a combination of both. However, as I've repeatedly made this concern known to you and you've consistently choosen to dismiss this opinion, I doubt you'll ever step back and re-assess your initial conclusions.
You and I are on the the same page. Further and obviously, it is the left and right sides of the data field that are more important. If you have an ATS and it is not getting real time data (right side), then do not expect to make money with it. LOL To get performance, the left side of the time non stationary interval becomes the most important single determinant for making money. The left side is not simply a matter of maintaining standardized lookup tables or just simple ma's, etc. Context determines the nonstationarity and this is done deductively and NOT inductively. A classic example is the nesting of fractals and the determinants that come from such. It is simply never allowable to trade with either side of the trading fractal not contained (or circumscribed) by at least a faster and slower fractal (each of whose non staionarity is taken intl consideration.). Certainty in trading is possible and therefore a requirement by deduction. In all logical considerations, order is maintained by reasoning with the facts. In trading, all the while closure is being reched with regard to various contributing constituants. wghen such have played their respective roles, irreversibility properties come into play. The ultimate results is that before the operating point moves, is that all other potential operating points are irreversibly eliminated by their respective closure being reached. I have repeated this statement for over 50 years. it is a defining statement with respect to what may be represented as the miagration of the market operating point. this statement also defines how logic is used to define precisely the left side of the non stationarity of the market's operation. My view on words that anyone uses in a post is this: anyone, at any time, is free to choose to look up something that they are unfamiliar with. There is also the notion of "seeding" in intellectual processes. If for no other reason, older persons have the odds in their favor for creating seminal observations. There is also "nobles Oblige"; people often feel responsible when others use their stuff in less than optimal intellectual ways. That is happening here, it turns out. My observation is that a lot of people on ET do not understand what I say. That is fine with me. There are poor people who do not understand what I say. On the other hand, those who wish to understand what I say do not have to be poor; it is their personal choice.
Just the opposite, Mike. More volatility gives a wider range of values that the market can take on. This leads to fewer trades, and even if the trades are just as profitable as the backtest, in the end there's fewer of them, making less profits. Do you have a formula for dealing with non-stationarity? What I've been told is just to normalize the values to eliminate non-stationarity, and this is the only way I've seen done in the literature on pairs systems.