It is rare that something that seems so clear cut and obvious works multiple times. This morning was one of those times in gold where the trade just seems so obvious, you begin to wonder if you have really keyed into something that everybody is missing. The previously discussed 1742 level looked to still be in play. Following yesterday's struggle to trade or settle above 1750 (we traded above it but the action was pretty weak) we seemed sure to return to that 1742 level. It turned out to be a great buy, and it got you 8 dollars up to the 1750 area. I then looked for gold to consolidate above 1750 and build a base before trending higher. This however was not to be.
Gold began to sell off making multiple intra-day lows. It made a low of 34.4, then 31.4 and then finally bottomed out at 1726. As we slowly crept up to the 28 level, gold picked up momentum and traded quickly up to 31. The algorithms, which were likely looking to trigger stops had little luck in the 26-28 area as the volume at these levels was very low. Certain types of algorithms look to trigger stops (sell stops in this case) by selling gold down into areas where these stops are deemed to be. As a for instance: If gold is trading 1701 and 1700 is considered a big support level; algorithms will begin selling, trying to push gold down to say 1698. Traders who felt that 1700 is a level of support might put their stops in at 1699. This is a practical way to trade; put a stop in below a point of support (if you are long, above resistance if you are short) as a way to limit risk. But this is no secret. Algorithms know where the levels of support and resistance are, and will push it through levels to trigger these stops. The algorithm that sells gold down from 1701 to 1699 is rewarded if there are a large number of stops in this area, because they will trigger sell orders pushing the market down further. After those stops are triggered, the algos can buy back their shorts.
Lessons? The first is, the worst place to put a stop is a tic above or below a known point of support or resistance. Far too often, with such a tight margin for error their will be a brief violation of that support or resistance, causing you to get stopped out at the dead lows of a move (assuming your long). It is thus important to give a little breathing room below critical levels so that you don't get taken by the algos. The second takeaway that I could see today was how to use volume as an indication of whether their is a potentially profitable short term trade. As mentioned, when gold made its new low, it moved lower on very light volume. Seeing this is a tip off that there were no stops at these levels to be triggered, and as a result, the algo selling would stop. While I can't speak to how these algos all operate, it would stand to reason that they would actually have to buy back the futures they had shorted in an effort to get momentum moving to the downside. If you are quick enough to notice that this is taking place, you could get long a few futures, ride them up for a few points, and make a quick hit. Around 1728 was clearly the spot to do this today, and it was good for a quick 3-4 dollars.
I will likely not write tomorrow as I am going to a conference to try to learn a little bit about high frequency/ algo trading. Hopefully I will be able to come back with some insights. In the meantime, keep a close eye out for Friday at 12:30 when Big Ben speaks. Recently his comments seem to act like speed for the markets. Depending on the trend, it might be a good time to put on a long gold position circa 12:15 pm. We will see if the experiment works.
Best,
Ben
No comments:
Post a Comment