Red, the earlier post by Tommo is similar to what I do. Basically, when a spike is the result of an attack, the price is temporarily not in equilibrium and should eventually return. So, you see if you can fade it. The trick is, to learn which spikes are, and are not, predatory attacks (some markets have heavy trading on certain clock rhythms, for instance) and, importantly, if you try to fade the spike the market-making machines will "spike again" to take you out if they have nothing better to do. In a very thin market, the mere act of making a trade will cause the market to gap a bit against you, so it's about experience and feel, as Tommo suggests.
CL somewhat reminiscent of the 8 cent NG rally I guess. 31st october just caught my eye... Find it surprising that these predator moves can take 30mins+ to unfold, as was the case in NG.
Applejuice, Remember, the moves help pay for themselves by stopping guys out, including any robots who naively try to fade the moves "too soon." The NG move was in the middle of the night, when Singapore was at lunch, the deepest lull in the market. It undoubtedly annihilated many robots trading then, as none could imagine 1,000 contracts or so of size coming into the market at the deepest lull.
So if you want to see something very cool, when you look at the large spikes overnight (most often sells) of, say, 150 contracts or more, you will almost always see a "smaller" order just precede it, by 10 to 120 seconds. This order is also huge by the standards of the overnight hours, 60 lots+. I call this order the "Sentinel," or "Private Stash," as it is as-if the partners at the HFT put their own order in before they crush the market. I have a robot who does nothing but look for these presaging orders overnight, this time trading with the move, and the robot is easily ahead. You can see these clearly in the massive predatory attacks in crude Sunday, Sep 14, after the market re-opened, attacks timed to occur in the deepest lull of all in trading, namely Sunday evening, thereby doing maximum damage/ collecting maximum stops. I lost 10 contracts of Crude, somewhat luckily only $0.15 during the first attack as my robot is very fast. This is the day I gave up stops. PS Your chart with the images was awesome. PS 12:40PM ET Thanksgiving Day (Nov 27) is a recent textbook classic attack with no alternate trend explanations whatsoever.
None Business, I'm just looking at the two previous cases you mentioned Nov 20th and Today on NG that you mentioned in post 84. Would you be able to confirm the attached screen shots that we are on the same page? One thing that doesn't seem to line up is the traded volumes compared to what you mentioned in your post?
If you look at the top chart from 10:40PM (22:40) - 11:00PM this is what the market looks like normally, as this period is a deep lull in trading. Here, roughly 500 contracts in Agenda (500 bought then 500 sold, 1,000 traded in all) bought 8.5 cents of agenda push. To push the market up 8.5 cents during the New York trading day would require 5,000-10,000 contracts (massive size).
Thanks None Business, Not sure why were getting different values could be down to auto scaling etc etc. But I see what you mean comparing the traded volumes to those typical on the ladder, it would be stupid to execute orders of those sizes on via market order. Thanks for your contributions they are a great help.
What about the portion of the 10:23 volume today in NG appearing to belong to an F/G spread? Should we draw conclusions about apparent similarities in volume based on the prompt month alone?