My own small "revelations" about volatility

Hi everyone,

I'm feeling quite generous today for some reason, so I decided to share something with all of you, especially beginner traders.

Have you ever wondered why your system works well for a pair and not so well, or really really bad on another pair? In my lurking I noticed many posts mentioning this behavior.

Only recently I encountered a similar situation when I was testing a system I developed for EURUSD on several other pairs, like GBPUSD, USDCAD and USDCHF.

To my surprise, it worked better on GBPUSD that it worked for EURUSD and worse on USDCAD and USDCHF.

I was intrigued and after giving it a few good hours of thinking, I thought about checking the volatilities of the mentioned pairs for the past month.

Guess what? GBPUSD is less volatile than EURUSD, but USDCAD and USDCHF are more volatile than EURUSD.

So there you go, especially if you're a beginner trader, it would be easier for you to trade pairs with lower volatility (as in more chances at making profits because they'd be easier to forecast), contrary to the "popular" recommendation that beginners should start with EURUSD for instance.

I checked with Alpari and they offer very good spread and max 50 lots instant execution for quite a few of the non-main forex pairs that have low volatilities.

Chances are your broker has good offers as well.

If you still want to start with one of the main pairs, it would be a slightly wiser choice to start with GBPUSD than with EURUSD, if you were to consider their volatilities for the past month.

Obviously if you're more keen on catching big moves, you're stuck with higher volatility pairs, at the risk of also loosing more.

And as always, best to due your own dilligence since there's already so much BS passed around as valid advice.

edit: EURCHF would be a nice pair for a beginner as well, besides GBPUSD

I don't really get this. Are you suggesting beginners should start with lower volatility pairs like Euro instead of pound if compared the two?

Hi, You make a good observation, but I wonder how you are measuring volatility? Also it depends on, if the system requires volatility to work (which some do).

The GBP/USD is more volatile. It tends to corrolate with the euro to a degree, but usually makes wider swings, and pushes that much further before rebounding back. It also tends to push further past support/resistance before respecting it. Like it's breaks are not working (automobile reference).

For example if the euro swings up and down from swing low to swing high in one day and makes 100 pips, the GBP will usually make 140 pips. (not exactly of course). This in many cases will make the GBP a better choice, for example in a EMA crossover system, because of the nature of such a system lagging, you get more movement with the GBP, so you get a little extra room to make some pips, before it swings against you again.

Maybe I missed your point, but that is my observation.

this is ludicrous. gbpusd is nearly twice as volatile as eurusd (both in ATR and in price movement) get your facts straight before you go misleading newbies (apparently you are one yourself)

tenatro, your calculations are wrong. Using ATR 100, which measures the average true range of the last 100 days for your pairs, I come up with the following.

USD/CAD 103 pips
USD/CHF 119 pips
EUR/USD 135 pips
GBP/USD 162 pips

Tdion...
few hours of thinking?
It must be a newbie


BTW
There are two possibilities with volatility.
1. high volatiolity caused by trends going longer
2. caused by more random moves

If it was first it's obvious everybody would trade pair with longer trend. Making money just can't be that easy - It's because trend ends just in a moment when last trendbelievers bought/sold and first of them started to take profit. It's too much to explain why higher volatility can't simply transfer in longer trends but it's easily explenable with fact, that people trade to make profit on forex not to make trend last longer.

Second case means that you can get wipsawed more often and smart gambler or any other trader will try apart of jumping into trend - jump to it on one of those stop hunting wipsaws. Those wipsaws cause your method not working on more volatile pairs. You have quite a number of fake signals. If you use same method but decide to enter one or two canles after signal [and same with exit] you'll propabily see profits.

[Merlin wrote nice method [here] of entering trade two or three candles before signals and it works pretty well but he don't share his indicator enymore ]

Hmm...
On my chart's one month takes approx. 26 days...
Which broker do you use?

Oanda. The last 26 days gives similar results. I figured that if you want to build a system based on volatility you need to go back a little further to get a better picture under various market conditions.

I wonder do you have some more observations - I ussually use last week for 5 min chart - you have any specific idea for lenght of time for different TF?

I think the idea is to go back far enough so that spikes don't skew the number. NFP release would certainly give you a misleading number if you only went back five candles.

I would be really interested in how you use ATR on the 5m. USD/JPY ATR on 5m is about 6 pips yet I see plenty of 20 and 30 pip candles.

I generally use ATR on longer TF like 3h and daily to determine possible SL and TP points.

I should've specified I was referring to volatility calculated in 22 points on daily data, for an approximate month of trading days.
In this case, GBPUSD actually is less volatile than EURUSD.

I don't really know what ATR really is and I don't care because I only use indicators I developed and find a need for.

For reference, check this:
http://www.mataf.net/en/forex/trading/volatility/forex/

Input daily and 22 points, and you'll see %0.98 for EURUSD and %0.96 for GBPUSD. I think this "relative" volatility is the one that makes the most sense when comparing pairs...take a look on Mataf on how it's computed.

In addition, you can check:

http://www.thefxmarkets.com/fxvolatility/

where you'll see EURUSD as the 10th most volatile pair and GBPUSD as the 11th most volatile pair.

Or, if you could just compute them yourself like I do for instance.

You should really stop acting like the smartest person around, even if you have a few CS degrees. I'm a CS student as well, close to graduation and I use a trading system I developed in Matlab based on DSP, AI and general econometrics techniques. I can assure you I'm not as stupid as you may think.

ps: My system trades only daily charts and if I were to recommend something to a beginner, the person to which this post was mainly addressed to, is to trade daily charts as well.It's less trouble and they can also get to have a life outside of trading.

Also, I see some people calculated historical volatilities on a relatively large window, 100 points or more. I simply wasn't interested in a time frame that large, it's possible that on such a large time frame GBPUSD is more volatile than EURUSD.I was more interested in what happened in the immediate history, max 1 month.

Sorry everyone for the misunderstanding, I should've specified the time frame I used.

-----

I also usually compute standard deviations of the daily lows for each pair I trade, to see which has stronger support. On a window of the last 22 data points, GBPUSD has stronger support than the EURUSD, with a std of 0.0050 as compared to 0.0060 for the EURUSD (smaller is better).

---

I was curious if the roles reversed for longer window sizes, so I used the tool Mataf provides to compute some volatilities really fast, the same relationship holds.

For volatility on daily data, computed on 80 points, EURUSD has %0.94 and GBPUSD has %0.84, even more obvious.Anyone still having doubts about this, try it yourself at Mataf.

I think that maybe we didn't all thought about the same thing as being volatility, and that's where many of the misunderstandings started.

Anyway, I'm done with this as I have more important matters to attend to and I don't want to start a never-ending flame war or contest about who's more stupid or smarter than who, so I'll no longer reply if someone posts something, I just don't think I have anything else to add besides what I already said.

In conclusion, it's best to pick pairs with strong support and a more "stable" so to say volatility.

Tenantro, My apologies for saying that you were wrong. Looks like you are talking about a variation of the stock market VIX index applied to forex. The ATR I referred to is just the average pip range over a defined period.

Since your volatility ranks were directed at beginning traders I would like to point out some potential pitfalls. The fxmarkets link you provided has NZD/USD ranked as #2 and GBP/JPY ranked as #12. Kiwi is known for its steady trending nature while GBP/JPY is known for explosive, violent moves. A beginner picking the Pound cross over the Kiwi thinking that the Pound was less volatile could get into trouble real fast. If your system, which you stated works best on lower volatility pair, performs better with the Pound cross than the Kiwi then you are using data inputs that would not be apparent to the average trader.

Again, my apologies for jumping to conclusions but maybe it is the uniqueness of your system that allows it to work for you. A beginner exploring their own path may find it less so.

No problem deuce.tdion's reaction is what bothered me, I wasn't bothered by anyone else, especially since I share part of the blame for not stating exactly what I was refering to.

You may be right, I didn't really know about VIX but I'll take a look at what it is now that you mentioned it.

Anyway, I'm out, I really don't want to start a flame war.

ps: I was curious about NZDUSD and computed a std deviation for the lows for the past 22 days, it's 0.0101, so the support is much weaker than in EURUSD and GBPUSD for the same period. If anyone else wants to try this and incorporate it in their analysis, take the last 22 values of the (in my case) daily lows, convert them to a return series and compute a standard deviation.

to be able to do this for other pairs and be able to compare the results and see which pair has stronger support, you absolutely have to convert to a return series, otherwise you'll be comparing apples and oranges.

On 5min it takes 24[hours in a day] x12[candles per hour] x5[days in a week] to get ATR for last week.
It's not exactly ATR what do I use. In fact I use bollingerBands(1440) on ATR(1) and on momentum(1) to spot some sudden moves on chart [third deviation on BB shows how big Mom or ATR is "too big". So those 30 pips spikes are exactly what i hunt for.
http://i36.tinypic.com/icp9xf.jpg

On a picture there is BB 120 so you can see it changing over time, but I still can use my eyes And most important of all - I know what do i look for, so indic is just for some help. I'm the boss

Ok.
I don't get it - on this page GU is more volatile than EU.
However it's very very clever to think of volatility not only as an ATR value.

How about this interpretaton:
The higher ATR/spread the better
The higher "longer TF ATR"/"shorter TF ATR" the better*

*daily is longer than hourly etc...

edit: nevermind...too tired, need to get some sleep

Kaligula, I was referring to the Volatility in % for a 100K contract size, the 4th column, I stated it in %.I was not referring to volatility expressed in pips, which is what I think everyone else was referring to.I can see very well that in pips, GBPUSD appears as having a higher volatility.

At the moment, it says %0.99 for EURUSD and %0.97 for GBPUSD in percents.

[quote=Kaligula;2194722]On 5min it takes 24[hours in a day] x12[candles per hour] x5[days in a week] to get ATR for last week.

I know the math. I didn't mean to imply that you were only going back five candles for your 5m charts. It was just an extreme example. Oanda doesn't go back 1,440 candles on their 5m charts. I generally use ATR(100) because I find it to be close enough for my purpose. http://forexfactory.com/images/icons/icon7.gif

Interesting use of ATR as a baseline for spike hunting.

if you found something that seems to work, good for you!

1) i didn't "react." i responded. my reply had nothing to do with "how smart i am."

2) if you ask 100 REAL currency traders if EURUSD is more volatile than GBPUSD, 99 would laugh and one would point to the mataf site.

3) the mataf calculation gives ATR (the pips column.) because the % column is not clearly defined, i will assume it is talking about % change of the RAW PRICE and not in terms of pips (probably lowest low divided by highest high of the past 22 days.) this might be useful from a fundamental point of view, but meaningless in terms of TA because any leverage applied will skew this number.

Using your link... I get:

EURUSD - 144.50 pips
GBPUSD - 174.00 pips

The thing about volatility (defined as absolute pip movement during the day) is that it's not really easier to make money w/ higher levels. If you're using a bar-based stop loss method your lot size (assuming a fixed risk %) will automatically scale to the volatility of the pair, meaning your end results are the same either way.

IMO the problem isn't volatility, altho dVol can be a real PITA. IMO the biggest trick is market efficiency, as in how much time does a pair spend ranging versus trending. The whole fractal efficiency thing comes to mind.

Did you use a normal distribution for the deviation mean? I've worked quite a bit lately on market stats and found that normal is completely off. Started using a log-normal distribution and have had far better results. So if you're still using a normal dist you might run into the same problems I did, ie: the classic long-tail problem.

Nod, especially recently. The cable has been wild. Great trades tho, is my favorite pair at the moment. Short the bearish trend breaks, exit partial lots, let the rest run as a position... textbook trading for now.

http://i37.tinypic.com/ruwgwl.jpg

I see here EU0.15% GU 0.21% [I set parameters for 1H chart and data calc on 22 points]
on 10 points it shows 0.95% for EU and 0.97% for GU.

I don't dig it or mozilla is cheating or higher numbers are for lower volatility, or tell me wat is it all says...