Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
My personal trading intention: Create a reasonably simple, robust strategy to scalp the market.
Of course, every word or concept has to be defined to understand what I mean, i.e. what does "resonably simple" mean? But that is not the point of this thread.
The reason I state my intention is because my understanding is many, or the majority, of traders who post here are pursuing a similar short-term intraday trading strategy, so we should have several who are like-minded in this effort.
What I would appreciate sharing ideas about, is an aspect of trading that is at least as critical as when to enter the market. In fact, if capital preservation and risk management are a priority in your trading, it may likely be more critical than entry:
When NOT to trade.
(yea, the title kinda kills the impact of the moment...)
So to start off this thread, I will suggest the following trader perspective from which to evaluate the concepts discussed below:
As a trader, assume that price action will continue as it has in the recent past several bars, and if it has not been price action that is useful for you to enter and profit from, then wait for the personality of price action to change before considering an entry.
I will suggest 3 concepts that may be useful to help identify price action to avoid:
1) Relatively low volume market conditions. I need to know what 'normal' volume is for my market that provides sufficient opportunity for useful price action to profit from. If volume is low relative to the same time on other days (perhaps due to pre/post-holiday lack of participation), or if volume is low because it is the typical time of day for it to be low (lunch lull), consider avoiding entries until acceptable volume resumes.
2) Narrow range consolidation. Perhaps one's strategy can participate in wider consolidation swings and profit in the middle, but at some point, the current trading range may be too small to expect to get the price action required to identify/confirm an entry plus achieve the number of ticks necessary to profit from based on your trade management. Narrow range consolidation is revealed by the obvious horizontal support and resistance channel it is in. Also, the price bar formations tend to be more overlapping (no strong moves that directioanlly progress from close to open) and alternating (up bar/down bar), not exhibiting any conviction of an ongoing directional commitment from the market participants.
3) The grinding directional move. The market is showing a directional bias, say making higher lows and higher highs with moving averages and other favorite indicators pointing up, but...price extensions are minimal, before stalling and retracing, then resuming the initial direction again. This one gives the hope of an entry, but price action that chews up tighter initial/trailing stops. It is really a subset of 2) narrow range consolidation, in that there is an upward (as opposed to a horizontal) channel of support and resistance, but minimal price extensions, which represent as the overlapping and alternating bars.
These are just a few ideas which require some analysis of relative volume and price action. Again, it seems to me that I can only expect the market to continue to do what it is currently doing now, and if I expect it to explode in my favor after it has been listless, just because my indicator or entry rule says it should, I may be placing my capital and psyche in harms way unnecessarily.
I welcome your critique and ideas.
Did today's ZN price action have any bearing on this post...?
Excellent post, Neal. Too many traders focus on when to trade or when to enter and not nearly enough on when NOT to trade, as you've pointed out.
Today for instance, I was concerned that as soon as I stopped trading (after being in chop all morning) that there would be a big break out move and I'd miss out on making my daily goal on the ZN. But, I also recognized, from past experience, that it was likely to just chop around again over and over.
So what is the safer course of action? The one with higher probability for not losing money? Simply not to trade when things don't go right. There are plenty of times when they do go right... so why force it.
The rules vary depending on the instrument you are trading and whether or not you are trying to automate the process or not.
Here is a brief overview of my automated system which you may find interesting. It has 3 different state classifications: trending, ranging, consolidating
It uses multiple timeframes to determine the current state of the system. Ie if signal to noise ratio is above a threshold in higher timeframe then market is trending. If it is below threshold on the higher timeframe but above threshold on a lower timeframe then it is either ranging or consolidating. To differentiate between those two I check the slope and width of a channel.
In this way I am able to trade with different parameters on a lower timeframe even if the market is not trending. Basically it is three totally different strategies depending on which state the market is in.
which has a screenshot of what i use for the highest timeframe, when the SNR indicator line with cyan and magenta dots is underneath the threshold then it is classified as nontrending and the system then checks against two lower timeframes to see if it can trade on either of them. Any signals generated when the SNR is under the threshold are considered void (unless it is seen as valid on a lower timeframe).
The SNR indicator is actually based on the channel in the top pane, it essentially calculates the variance about the median of the channel. The channel itself is somewhat complex, based on a nonparametric kernel density estimation but a similar system could be devised based on keltner bands with pretty good results.
My system is still a work in progress but already it is profitable across most all market conditions (in live trading not sim). Getting to that point has been quite a journey and like most people involved blowing up a few accounts, around $35k in all (much of which was lost in stress-induced discretionary trading before I got my psychology and money management in check). Based on my what I have learned, knowing when not to trade is far more important than how you enter trades. I would say it is second only to good money management.
BTW.. these are the same type of ideas behind Mark Jurik's JMA also. You can get basically the same signals I have from his indicator (they both closely approach the bayesian optimal estimate after a given number of samples)
The signals are not the point though, you can get similar signals from any number of indicators. The power of this kind of nonparametric technique is that they are 'measure preserving' transforms, in other words they give you a smooth estimate of the signal without introducing any distortion in the smoothing process (like most indicators do).
The lack of distortion is very important because it allows you to accurately asses the extent of one movement compared to another. And with density estimation you also get a framework which you can use to assign concrete numbers (ie probabilities) to different events. From there you can record statistics and do all kinds of interesting things with the data, such as define with high accuracy when the market is trending and when it is not =)
I haven't looked in great detail at the files there but in general the problem with Kalman filter is that they are usually based on the gaussian assumption and thus not as good at capturing nonlinearity and ergodic behavior.
However in that pdf it appears they are using nonparametric spline regression to make kalman filter adaptive. While I'm sure that is a powerful technique it appears to be much more complex to implement than the kalman smoother based on condensation algorithm I linked to above. Also by using spline (polynomial based) regression you lose the major benefit of the Monte Carlo based techniques which is the probability density approximation.
Here is an image which illustrates what I mentioned before about measure preserving transforms. You can see how the Woodies CCI shows nearly identical signals to the particle oscillator but it is far less smooth and it distorts the magnitude of movements.