Are you sure you'd want reading recommendations from an idiot who struggles with the singular form of "phenomenon"? To be honest, I don't know of any sources that covers it specifically. I've just had an interest in chaos and complexity in general for some time and I've done my own analysis. It's definitely an interesting topic...
You have no references to provide, yet you called the post "ramblings". Why is that? I'm trying to understand. Is the post bad or you are a negative person?
Synchronization (flocking) happens all over the place in nature (even for innanimate objects...) See this Ted talk for detailed examples: http://www.youtube.com/watch?v=aSNrKS-sCE0 It is only logical that HFT would follow such universal behaviours... The Flash crash was a good example of this phenomenon. this and other illustrations of the phenomenon can be easily observed in the research conducted by nanex http://www.nanex.net/FlashCrash/OngoingResearch.html
I didn't mean to come off as harsh as I probably did. My criticism was simply that he doesn't provide any evidence for his theory. Where are his references? He took a topic that's been "everywhere" in popular science the past few years and applied it (conceptually) to HFT. He then states that trading higher time frames might avoid the impact of HFT. These are hardly novel ideas, and I assume every decent trader has thought about both a long time ago. I think I've even seen a few posts about it on this very message board... The blog post is basically just a few loose thoughts without anything to back them up; not even a theory of how one would exploit the potential occurrence of such a pattern. Thus, I don't see how the post holds much value to anyone but extremely naive chartists.
Flocking or chaos theory is not relevant at all to what I said. HFT is akin to legalized front running. The bots see the order flow before our eyes can blink. They are able to trade off the order flow in a rapid manner. The flash crash was not flocking it was a response of algos all triggering sell orders at the same time.
All the algos were triggering sell orders and cancelling their bids as a response to a large sell order... no different from fish moving out of the way when a dolphin is comming through... once one of the algos starts cancelling the others do the same... just like an stampede...
Whatever. Still no one has yet to disprove my thesis that HFT is about trading on order flow information not available to those with slower computers or whose gateways are troubled with latency.
Some hfts do behave as you describe. Some of them don't. Because of scale economies they can't all do the same thing while theyre 70% of the volume. They have to do different things. Some hfts just care about getting a good place in the queue when the price changes to get a better chance at scoring credits. Some work of complex quant models to try to optimize for an aspect of the markets behavior others aren't looking at. Behind every hft there's a guy trying to figure out how to take more $ from the market and everyone of them cannot be the fastest ( only one of them actually is) and only a few of them can actually profit from getting info and acting on it before others do. The others have to come up with different ways to make money.
The coined term "HFT" is a joke. There are a few key factors within automated trading: Holding Time Markets traded (or products traded) Daily Volume Cancel to Fill Ratios Latency is also a factor but latency depends more on the above than on it's own.