Would love to see this article, if you can recall exactly where it was etc; Did it mention "who: they thought was dumb enough to let algo go like this ? The big money in gas trades the curve (convexity) years out in the cal. There are a lot of small time people playig with themselves in the front. Dont get me wrong, the front is good, although comparativley he curve is where the big money is won and lost, and it is just that a huge gamble. Dirty hedges at best, guesstimates about weather, storage relationships. Someone always blows up every 3-4 years.
Here is a link to an article which touches on this: http://ftalphaville.ft.com/blog/2011/06/09/589621/when-algorithms-fight-natgas-edition/ It's not the one I read, but shows the growing influence of algos in futures. re: trading the curve. Yes, I heard that energy is a spreader's market. Isn't one of the NG cal spreads called "the widowmaker?"
I recall the sell off that evening., Heard it was a fat finger to sell 100 instead sold 1k in a light markrt at market Not so sure algo's at work that time a night..... seems like more misinformation from a "source"
This is an interesting thread I stumbled accross while looking for an ACD indicator or something similiar that can plot breakout levels based on % of ATR. http://www.bigmiketrading.com/traders-hideout/6742-acd-trading-mark-fisher-4.html The poster fat tails has an interesting concept he uses that he prefers to A levels. Instead of A levels he uses what he calls noise levels. In his words: "That sounds too complicated. My problem with the Fisher method is that the breakout points seem to be set somewhat arbitrarily. Sometimes they are very close to the opening range and sometimes they are far away. So I have settled for a different concept, based on failed breakouts. A failed breakout is the smaller move made by price as counted from the open: Failure move = Mininum of (High - Open) and (Open - Low) The smaller of these moves can be considered as noise produced by the market while it was in its typcial bipolar state of mind. From this I am calculating noise bands that define the levels beyond which a price move becomes meaning full. I have produced the noise bands indicator," You can set this to plot at 100% of the calculated value or a smaller % which would make it a closer approximation of A levels if that is what you are shooting for. The main difference is that it is calculated off the opening price as opposed to an opening range. Either way it does a good job of establishing a framework and bias to stay on the right side of the market and keep from overtrading. He also created an indicator that plots a target range based on the average daily range of a set time period. It changes throughout the day as the range increases. The thread has lots of charts if you want to see what it looks like. I like his idea because it allows you to quickly get a visual on the risk/reward of a trade based on how far the market should move on average.
I think the target levels are really cool because they help you to set a target that is really realistic. I know one of my struggles when I was daytrading was always shooting for the homerun unrealistic target. This helps you to see what "should" happen most of the time and trade accordingly.
I have this target range set on my TOS charts. It looks like it's the same as his. It dynamically updates after each 5 minute bar when a new high or low is made to pull the shaded range up or down. You need to be careful with this though. Like anything else with trading, there is nuance right. You have to know when to use it and when not to use it. I find it works best for either stock traders or someone who is trading a lot of different products and isn't that aware of the usual trading range for that product.
I am not sure if my idea is similar, but I have been toying with the idea of average opening range bands to give an indication of historical noise at the open. Whatever your time frame, just take an average of X days back, and have it plotted. I am thinking this might give an indication of out of the ordinary activity.
Just out of curiosity, but why are some of you trying to re-invent the wheel? ACD operates on the simple concept of keep it simply stupid. Using bands and standard deviations and this and that, don't you think that is only going to corrupt your ability to make fast decisions?
Thats true, indicators dont work well when tested through and through. The ACD is a risk mangement tool= good. Although cant really be tested as you say, and nobody can quantify much about it, other than some use it and those partcular folks dont have any public performance reports...
I know for me I'm looking at the volatility bands as a way to create profit targets and estimate R:R. It seems like it fits well with the ACD levels since it is all based on volatility.