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)
Lately I've been trying to refine how my risk/reward. Instead of a using arbitrary numbers or my own intuition, I want to have both risk and reward defined by market conditions. ATR is the obvious way to do this, but if you use the same ATR for both risk and reward, you'll be left with 1:1.That isn't necessarily bad, but we always strive for better, right?
Don't beat me up too much yet because I haven't backtested this at all.. here's my theory:
Take a fast ATR and a slow ATR, and make your risk on a given trade equal to the difference between the long ATR and the short ATR. The profit target is taken directly from the long ATR. If this results in better than 1:1 RR, then the trade is permissible.
Your target is now based on how far price is likely to go if it goes in your direction, and your stop is based on how far price has deviated from 'normal' volatility. This deviation is key. In a range, the fast ATR will be close to or below the slow ATR, thus reducing risk without reducing reward or keeping you out of the trade entirely.
To illustrate:
In the above example you wouldn't take the trade because it would result in less than 1:1. Given different conditions you'll end up with different ratios, but they will always be tied to volatility.
And here's what happens when R:R moves in your favor:
I realize that using the difference between long and short ATRs for risk is arbitrary, but it's the best I can come up with within this context.
This might not (and most likely wont) be any better than existing risk management methods, but I will be backward and forward testing this next week to find out if it shows any promise. In the mean time I'm very open to any suggestions.
a problem that pops up in my mind: if the fast ATR is lower than the longerterm average, it essentially means that you are in a phase of low volatility.
Almost every strategy, apart from breakout trading maybe, is likely to get grilled by low volatility market conditions.
Because if there is no volatility, there is no movement. No movement, nothing to be earned.
Maybe an alternative possibility would be to use a given ATR for your stop and the same period ATR of a higher timeframe for target setting - would obviously require htf confirmation though.
anyway, I like ideas that are different - so go on ahead and proof me wrong!
My thinking is that once the fast ATR crosses under the slow ATR, you are in a no-trade zone because volatility is no longer any better than the average, and thus your opportunity for profit is less than favorable. I also want to look at using standard deviations, which would give a statistical edge if they could be implemented properly.
AHH - got you wrong at first. Didn´t realize that the difference between fast and slow shall be your stop. Sry for the confusion..
My guess would be that the stops gonna be way too tight - essentially you go in a phase of trading where the average candle is bigger than normal but utilize a stoploss which is smaller than even the average candle. In fact even your target would be expected to be intra-candle so to speak, if reoresented by the slow ATR.
Yeah I thought of that. Unfortunately the only thing I can think of doing is using an ATR multiple to widen the stop. The problem is that you run into the old issue of "which number do I use?". I'll look at tying the multiple to the something like volume or momentum.
Ok so I'm going to use VWAP standard deviation bands as a way to approximate the ATR multiple. It's pretty simple:
if your entry will be between VWAP and the first standard deviation, then use a multiple of 2. Use a multiple of 3 for the 2nd and 3rd. This is because statistically 68.1% of prices will be within the fist SD band, 95% within the first 2, and 99.7% within all three.. The math is as follows:
3 [standard deviations] * .681 = 2.043 or roughly 2
3 [standard deviations] * .95 = 2.85 or roughly 3
3 [standard deviations] * .997 = 2.991 or again roughly 3
simple as hell and it adds probability (bell curve statistics), volume (the volume weighting), and again volatility (the SDs) into the mix.
In this case the I'm using a 3 minute chart. The fast ATR is 14 period and the slow is 2400 period (5 days of 3 minute bars)
The calculated stop is 100.09 and the target is 99.75. The risk to reward is slightly less than 1:1 (.71 would be exactly 1:1), so this would be a no-trade situation. On top of that the open is at 99.75 which could mean a hard bounce, and professional delta is to the upside, but really I'm just looking at R:R here so that isn't what I'm concerned with.
The fact that I have not only unfavorable R:R but also two others of my own criteria working against me tells me I need to either look at trading in the other direction or wait for a better setup.
Just so everyone understands what was done:
The slow was subtracted from the fast,
.111 - .0493 = 0.0617
This was multiplied by the multiple, which in this case would be 3 (because the entry would be between the 2nd standard deviation and the VWAP proper)
0.0617 * 3 = 0.1851
So 19 ticks is the maximum.. 99.90 + .1851 = 100.0851 (rounded up to 100.09)
And the target is simply the slow ATR, so 99.90 - (.0493 * 3) = 99.7521 (rounded down to 99.75)
I have found the VWAP sd bands to be very useful in many ways; principally in terms of reversion to the mean type of situations. Something interesting is likely to happen if price gets up/down into the 3rd band, and even just on the borders of the bands, especially if the VWAP itself is sort of flat on the day.
I'm having trouble understanding its relationship to ATR; probably I'll have to read the previous posts in the thread to understand better what you're driving at.
By the way, I seem to remember that @Fat Tails made the distinction somewhere between the "directional" volatility of the standard deviation (price has to actually move a certain distance in one direction for the sd bands to move) vs. the "non-directional" volatility of ATR (price can just whip around with big bars and go nowhere on a net basis and the average range will still be big.)
Apologies to @Fat Tails if I didn't get his point right, but I do see a potential difference in the volatility being measured by the two methods. Large bars generally are going to follow one another up/down in a trend, but not necessarily -- they may just chop; and lots of trends proceed with a sequence of little bars, and so have a lowish ATR.
Not sure what that means for what you're doing, but something else to put into the conceptual mix.
I hadn't thought of it that way, thank you! I'll have to contemplate whether VWAP SDs are the best choice. I may try a SD of the fast ATR, but for now I'm going to test out the VWAP idea.