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)
I think your dream makes sense, a simple gauge that tells you the market's high value price, and low value price, much like "Bollinger Bands" but that works better. That's actually something I came up with after many years of research. I was thinking of packaging up some code and giving away a simple gauge, but haven't had the time yet.
I can describe something that may help here though... basically it's like Boll. Bands but different in a few ways:
1) Use an average deviation instead of standard deviation, because standard distorts the true data.
2) Calculate the high band and the low band separately -- the high band is calculated from the bar highs, and the low band from the bar lows, and the high & low bands' deviations are calculated separately IOW the high band is calculated from itself, whereas B.Bands uses the deviation of both sides for each side but that's technically wrong.
3) Use an EMA, do not use an average, this is actually very important, I won't go into why here.
4) Always use 2 avg deviations from the EMA. Anything other than this is just mathematically wrong. I won't go into why here.
5) Set the lookback to TWO wavelengths, the wavelength changes all the time, but take your best guess at the average. Don't use any other lookback, I won't go into why here, but this will give you the true value price. The market runs on swings, not bars, so you want two swings lookback, and since we're using bars for data to approximate the swing data, that is exactly two wavelengths.
This indicator is a gauge as to what the value-balance price of the market is, there's two values, or bands, the high band and the low band. So ideally you want to buy at/near the low band and sell at/near the high band (not in the middle). The middle is used to tell you momentum -- if the market price started at the high band (it was last/most-recently at the high band) and then crosses the middle, then there's a super high probability that the market price will reach the low band. And vice versa. You could wait for the middle to be crossed before entering, but again ideally you want to enter at/near one of the bands.
"In range" & "out of range":
If the market price passed above the high band, or below the low band, then that means the market price is "out of range" on this current scale. On larger scales the market price will still be "in range" (below the high band and above the low band). Once out of range, a reversal is a retracement plus crossing back into range. While in range, a reversal is reaching the high or low band, then having a retracement (but the main scale doesn't need a retracement, only your smaller signal scale).
I'll let you figure out what scales to use, but basically you have your main scale which is the minimum scale at which you can make money swing trading. For example, I use a ~12 tick range chart for the E-mini 500. Then you have your smaller 'signal scale' which you use to get signals off of, for example I use ~6 to 7 tick range chart, but you might also use even smaller scales than that for some other types of signals. I won't discuss signals here. Then, you have larger scales which are your main trend -- the net composite trend of all market participants. This is harder to determine because it changes all the time, based on what all market participants are doing. There is a way to figure this out, but it's too much to discuss here now. At least establish your main scale and signal scale(s).
And use range charts, because that's mathematically correct. Time charts don't make any sense IMO. Volume or tick charts are much better than time, but again don't make sense. Renko is okay, except that most/all software has it coded wrong -- it's supposed to be exactly the same as range charts but it eliminates some of the trending bars because it doesn't show the overlapping bars... so just use range charts because that's the correct way to plot data.
So, that's a basic concept of what I figured out. Noone will probably care LOL. Everyone has their own methods and is set in their ways, which is fine, but what I do is mathematically sound and works. But again, if I tried to sell what I know, I'd probably not make $1 because of all the crappy scammy stuff that's been circulating for decades, so everyone is skeptical for good reason.
Trading: Primarily Energy but also a little Equities, Fixed Income, Metals, U308 and Crypto.
Frequency: Many times daily
Duration: Never
Posts: 5,059 since Dec 2013
Thanks Given: 4,410
Thanks Received: 10,226
@thoughtful I've always wondered why we use standard deviation so much over average deviation. Definitely more effected by outliers. I think the answer is probably related to the fact that calculus doesn't like absolute value functions.
Would you also calculate exponential average deviations rather than simple ones?
Also whats the mathematical reason to use 2 avg deviations from the EMA? Obviously understand the relationship of standard deviations to the normal distribution but this is obviously different.
Thank you @thoughtful - I can see why you chose your name! I really do appreciate you taking the time to answer this. I am going to study this religiously and maybe I will have more questions to post, being so new. You have been a godsend though.
For the average deviation, just use a simple average of the deviations. An EMA is not applicable in this case.
I think I may be a little premature in my explanation, because I use a different version of this gauge, long story there (it's simple but it's not as easy to code), so I really need to finish coding this, and then I'll re-post more explanation. Sorry about that! So I really need to finish the code and then I'll re-post in the future with definitive info.
But for now, basically, I think what I should have said is to calculate the deviation based on a simple average (of 2 wavelengths lookback), so for the high band, each bar's deviation is the absolute value of that bar's high minus the simple average. Then, after you have the average deviation calculated, plot 2 avg dev's from the EMA.
The reason for 2 avg dev's is because that's what results in the best approximation of the swing (peak or trough). Imagine a perfectly symmetrical sine wave, of one wavelength, the average of all the data points (bars) will be half way from the bottom of the wave to the top of the wave, right in the middle. Draw a horizontal line right at that middle value, now, starting from this middle of the wave, the average deviation will be half way to the top (or, vice versa, half way to the bottom). That's your 1 avg deviation, but since we want the value to be at the swing, IOW the peak of the wave (or the trough), we have to double it so that's 2 avg deviations.
You could use maximum deviation instead of average deviation, and then you'd use 1 deviation from the EMA instead of 2. This is probably easier, and just as good.
Also, you could only use deviations on one side, so for the high band you only use deviations above the average, and ignore all deviations below it. But, I'm not sure if that really matters much or not yet, it's just an optimization idea.
ALSO, I need to clarify the EMA lookback. When I say 2 wavelengths lookback for the EMA, I mean the EMA factor is 1 divided by 2 wavelengths. That's what you write in the code. But the standard EMA formula is calculated as 1 divided by (HALF of the lookback). So sorry about that!!! So, if you're using a standard EMA formula then you want the lookback to be 4 wavelengths. I really need to finish coding this before I start explaning it LOL. But I just cleared some things up, so hopefully what I said in this post is now correct. So, for calculating the average deviation, use 2 wavelenghts lookback. And for the EMA calculation, use 4 wavelengths lookback. EMA's are confusing in that way.
Average deviation is simply a standard math thing -- just an average of all the individual bars' deviation.
Yes, I recommend only using the highs and lows, don't use the Close or (High + Low)/2. So for the high band you use the highs of the bars, that's it, nothing else. You want that edge, you don't want the Close near that edge. And, obviously, the high band and the low band are basically two separate calculations, they're like two completely separate indicators. Although the middle is the average of both bands, (high band + low band) divided by 2.
Yes, I use this gauge on multiple time frames. Basically I just use my main scale (12 to 14 tick range chart for Emini500), I wait for the market to reach one of the bands, then start looking for a reversal pattern & smaller scale signal pattern. But, I also use larger scales to further filter the trade, the larger scales show the net composite trend of all market participants. So, I wait for the momentum to run out on larger scales before I'll take a main scale trade. How I pick the larger scales is another topic.
Thanks! I hope my ideas help. I was basically clueless some years ago, but I figured out some basic concepts like this gauge that helps alot... I started with the approach that the market is a mysterious thing and I was just trying to find some kind of "magic wand" or "handle" on it, but then I realized that just basic mathematically sound data analysis was what helped me the most. IOW just being able to establish a logical high & low value gauge, and also finding the trend on larger scales (not discussed here in this post). I figure what all market participants do is basically random, since I don't know what they're going to do or when, but what we traders can do is price data analysis to see where the value levels are, and the market always tries to buy low & sell high at those value levels, otherwise they wouldn't make as much money or they'd lose money. The data (mainly swings) tells you what the market is doing.
The gauge adapts to volatility, if there's alot of amplitude, the 'channel widdth' of the gauge is wider, and vice versa. It's always tracking what the market is doing. So, anyways, I like the concept because it gives me a framework to know "where" the market is on any given scale.
Again, I plan to write up some code (and post it online), but haven't done so yet. I have code of my own, but I have alot of other code mixed in, so I have to clean it up first, and haven't had time yet.
Oh, BTW, there's some details about what constitutes an out-range reversal pattern, and an in-range reversal pattern. For example, out of range is defined as 1 full barrange past the band (using a range chart). 1 barrange on a range chart is the minimum significant amount of price movement for that scale, it's the minimum reversal amount to make a swing. 1/2 a barrange is a "leeway" (near enough) amount, so if the market reached within a 1/2 barrange of a band then that's considered near enough to say that it reached the band. The out-range reversal pattern is: the market goes out of range, then has a retracement, then it Closes inside the band, but you don't always use the most recent band value, you use the band value that it went out of range of... so the pattern is a sequence, you use the most recent pattern, sometimes a pattern starts but before it can complete it gets replaced by a new pattern. Also, the main scale doesn't have to have a full in-range reversal to trade it, it can just reach the band (but not go out range) and then I only need a smaller scale signal. Anyways, I'll probably explain more when I package up some code.
Trading is my passion and my experience is that you learn better if you try to do it yourself rather than having it done for you.
Of course, there is/will be an exception when I feel its more suitable to have a hired hand, especially when it comes down to coding (platform setting is done for free in my country by brokers as "demo") as coding your indicator yourself can be bit of a hassle if you are not well versed with both math and programing language behind it. And learning it just for one particular indicator/job doesn't make any sense.
Having said that, again from my experience, its highly likely that it wont be just "one" instance and you will end up playing with 100's of indicators and in such a case its better to learn and do it yourself than to pay.
Finding your edge with just one attempt would be very very difficult. At least from my experience, I've been at it for couple of years and I dare not claim to be profitable consistently, but I also dare not say that I don't code/try something every week to find that edge in trading.
considering there's many courses and systems out there charging thousands i think one on one advise from a reputable source would be well received and I think people would definitely pay a premium for that? I don't even want to think of all the courses and books I've purchased that have been a let down. i think a good hourly rate for good advise would be welcomed.
Finding multiple edges that are resilient and time-tested is the best, I think. There so very many things that just fall apart after working for a few days, weeks, months. Then we find some stupid pattern that has been around since Babylonian days and it still works.