Are Standard Deviation based stops, one of the best stops?

[KCOJ]

The total period tested/trained and displayed should be many, many times that lookback parameter

Yes I agree, of course … but in this instance I’m not testing/training anything.

The point of the chart was simply to illustrate that contrary to @tommcginnis comment that “once the look-back is lined up, they will behave nearly identically”, in fact ATR and SD can behave quite differently.

Suppose that …

TraderA employs SD to determine stop placement
TraderB employs ATR to determine stop placement

If both traders hold positions in the S&P 500, then based on the above chart TraderA would currently be moving his stop further away from the market (due to increasing SD), whereas TraderB would be moving his stop closer to the market (due to decreasing ATR).

The chart clearly does not provide any evidence that one is better than the other, it simply shows that they are most certainly not the same.

 

[tommcginnis]

Yes I agree, of course … but in this instance I’m not testing/training anything.

The point of the chart was simply to illustrate that contrary to @tommcginnis comment that “once the look-back is lined up, they will behave nearly identically”, in fact ATR and SD can behave quite differently.

The chart clearly does not provide any evidence that one is better than the other, it simply shows that they are most certainly not the same.

Let's call this "lined-up"....

Now, I'd say that's pretty fair evidence, wouldn't you?

 

[KCOJ]

Now, I'd say that's pretty fair evidence, wouldn't you?

Yes indeed … fair evidence that for your specific market in that particular time frame, ATR and SD do line up. And there are also many times for the S&P 500 where they will show very similar patterns.

The point of the S&P chart was simply to demonstrate that there can also be times when ATR and SD behave quite differently. A quick scroll back through some historical data will show many examples of this.

That said, I think the more interesting issue is what does that tell us? That is, at times when ATR and SD deviate away from each other, what does this mean … what is going on?

So since the US election, the SD for the S&P 500 has been steadily rising, yet the ATR has been steadily falling. If you’re a proponent of using stops, should you be moving them closer of further away from the market? Of course, any systematic trader will simply back-test both the use of SD and ATR and make their decision from there.

But I wonder, intuitively is there something we can learn about (or even better, profit from) the current market based on this divergence between SD and ATR?

 

[Simples]

Yes indeed … fair evidence that for your specific market in that particular time frame, ATR and SD do line up. And there are also many times for the S&P 500 where they will show very similar patterns.

The point of the S&P chart was simply to demonstrate that there can also be times when ATR and SD behave quite differently. A quick scroll back through some historical data will show many examples of this.

That said, I think the more interesting issue is what does that tell us? That is, at times when ATR and SD deviate away from each other, what does this mean … what is going on?

So since the US election, the SD for the S&P 500 has been steadily rising, yet the ATR has been steadily falling. If you’re a proponent of using stops, should you be moving them closer of further away from the market? Of course, any systematic trader will simply back-test both the use of SD and ATR and make their decision from there.

But I wonder, intuitively is there something we can learn about (or even better, profit from) the current market based on this divergence between SD and ATR?

ATR is biased towards daily bars and an averaging period, SD is biased towards movement away from MA for the same averaging period. They are two separate measures of volatility, depending on what you think you "need". There's no magic involved, just simple mathematical formulas. You can understand both in terms of maths, which may help in further backtests.

In practice, this means when ATR is climbing, on average your bars are getting longer.
When SD is rising, it means price is on average moving away from the MA for that period.
Beware though of artifacts from bars disappearing from the lookback, which is maybe one of the reasons many people employ EMA or WMA.

This is ABC, and not enough to profit from. You can probably profit without knowing this as well.

 

[tommcginnis]

Yes indeed … fair evidence that for your specific market in that particular time frame, ATR and SD do line up. And there are also many times for the S&P 500 where they will show very similar patterns.

The point of the S&P chart was simply to demonstrate that there can also be times when ATR and SD behave quite differently. A quick scroll back through some historical data will show many examples of this.

No.
The graph provided was *your* time period, wherein you challenged that ATR and SD would behave differently.
What was demonstrated was that they don't.
What was originally contested by me, and protested by you, is that this similarity behavior (of a tunable ATR matched against volatility) is consistent.

 

[tommcginnis]

(1) ATR is biased towards daily bars and an averaging period, SD is biased towards movement away from MA for the same averaging period. They are two separate measures of volatility, ...

(2) In practice, this means when ATR is climbing, on average your bars are getting longer.
When SD is rising, it means price is on average moving away from the MA for that period.

Not to be a stinker or anything, but

(1) There is no mathematical bias possible in either formula.

(2) What you are saying is the same thing -- truly "six of one; half a dozen of another."

 

[Simples]

Not to be a stinker or anything, but

(1) There is no mathematical bias possible in either formula.

(2) What you are saying is the same thing -- truly "six of one; half a dozen of another."

1. Mathematically this is correct. By interpretation biases creep in (even by omission etc.) Trading is mostly biased, isn't it?

2. For my uses, I try to choose the right tool for a specific job. Need to know the job first The problem becomes one of which formula out of infinity to choose? Sometimes a cheap and fast one is enough though. Been through the rabbit hole and complexity is not really helping much.

 

[tommcginnis]

1. Mathematically this is correct. By interpretation biases creep in (even by omission etc.) Trading is mostly biased, isn't it?

2. For my uses, I try to choose the right tool for a specific job. Need to know the job first The problem becomes one of which formula out of infinity to choose? Sometimes a cheap and fast one is enough though. Been through the rabbit hole and complexity is not really helping much.

Granted! But it was the ATR and σ that were being discussed, not trading biases that we bring ourselves. As I had mentioned above, I favor ATR for short-term, and σ for longer term -- but I recognize that that bias stems from my own weaknesses, and not in any objective measure of these data. I myself don't try to be "unbiased", but count it way much the victory just to be able to *recognize* 'em. ["Bastards!"]

 

[KCOJ]

This is ABC, and not enough to profit from

Well I agree that they’re both very straightforward formulae, but I don’t see why that negates any profitability. That said, I definitely need to spend more time thinking about how they may be able to be used in combination with each other.

The graph provided was *your* time period, wherein you challenged that ATR and SD would behave differently.

What was demonstrated was that they don't

Sorry Tom, it seems I’m missing something here. The chart that I posted was for the S&P 500 which clearly showed a divergence between ATR and SD.

The chart you posted was for another market (I’m not sure which as you didn’t say) which showed ATR and SD aligning.

So those two examples clearly demonstrate that for some time periods and for some markets ATR and SD will align and for others they won’t. Or are you suggesting that the S&P 500 data I posted is in fact not accurate?

 

[Buy1Sell2]

Are Standard Deviation based stops, one of the best stops?

No.