Don't be misled by the idea that trading more frequently is better. These 100-lot ES trades that get dumped out 1 tick later are certainly making the broker rich, but not necessarily the trader. The only "super-fast, super-frequent" trading that I would not be suspicious of is arbitrage.
The article was published in the Traders magazine by a guy called Erich Florek. The problem is that it is written in German and probably most people here can not read it. In the article the author describes the work of Scalpers and how they do sometimes 1000 Transactions per day. Some guys trade more than 100.000 contracts in the Bund per day !!! The author then describes the various styles they implement. Many of the scalpers use sound machines where an electronic voice reads out the bid and offers and thus creates a "pit environment". The faster the market is, the louder and more hectic the voice becomes. There are "Placers", "Pushers", "Surfers" and "Analytics". All have different style in trading super fast. They place huge fake orders and even push markets in some direction. They trade such huge sizes that they try to mislead everybody. They place huge bids even though they try to sell -- then they takes out the bids and sell the market down etc etc. But the ones that fascinate me the most are the Analytics. They work with neural networks and are very rich niche players who employ a whole team of IT specialists, mathematicians and Traders. And i think that is the future. People are programming chess computers and all sorts of artificial intelligence stuff -- so I would not be surprised if there are already some "Matrix - like" machines at work. Nice to know who your counterpart is in trading I personally have heard of a dutch guy in Switzerland (Kanton Zug) who does work like this very successfully.
i looked into this for a system on common stocks, and i got an estimate 500k, 200k for the software to get started, which includes a compiler of sorts that supposably analyzes data faster than any other software on the market (it comes from a group that worked in the intelligence community, who took their software private), and another 300k for servers and what he estimated to be about 6 months of programming. the problem is this is already being done in mass on both NYSE and OTC stocks. . look anytime there is news on an nyse stock, a flood of ecn's appears and disappears all at once.
Tape reading this way is still predicting. And the shorter the time frame the easier it is to manipulate. Especially orderbooks with hidden and cancelled orders. If someone tells you about their edge the odds are 90% they don´t tell the truth.
I have done statistical analysis on the depth of the markets on ES and NQ. I have never found anything significant. Worse still is the fact that phantom bids and offers are thrown in all the time to try to influence price action by people that belive this stuff. Doesn't mean there isn't something there, but my "tools" have not picked it up. nitro
Very interesting . . I was thinking about looking at that. Have you done any analysis of the tape (i.e. number of trades going off at the bid vs. ask), or intermarket analysis (i.e. ES vs. DAX)? [Richard]
... The fact that this "phantom" data can be recognized on a statistical level through a set of rules is itself information. That is you can characterize these data points and use them to gauge the attempted plays being made by market participants - not perfect signals but additional statistical information.
so true. you almost have to trade commission free to make this stuff work. for the last few months the slower you trade the better of you have been. its hard to outperform a melt up market trading.