Forex Dealing Break Down (1) Click on the link to you to open the “pdf” graphic of trade execution flow exchange comparisons between retail traders and High Frequency Trading algorithmic “machine-to-machine” trade executions.
How Slow is the NBBO?
A Comparison with Direct Exchange Feeds
“the oracle machine”
Thirty percent of Wall Street’s intraday liquidity is transformed through price formation acceleration, explicitly influenced by high frequency trading thriving in millions of dollars of profits by exploiting the exchange’s trade execution “latency” electronically sent over the Internet ot capture the “best price” trade executions for the consumer trader.
To be efficiently profitable as a predator on incoming bid/ask trades, the HFT servers are set up is co-located server near an exchange/electronic communications network premise and plugged into a ultra-fast bandwidth. Through network stacking, messaging protocols and raw market data processing.
Though the financial media claims HFT brings greater “liquidity” to the market, in reality it is causing price dislocation for the consumer day trader. For example, the NP-Hard can be used as source code applied to the arbitrage decision problem presented by the variance of the bid and ask spread among the exchanges data flows. Here is where the “investors” and “brokers” can virtually integrate the exchanges through their computational technology without “transparency” to the consumer trader.
“clock-rate – faster to market”
Likening the linear exchanges to a convex polyhedral geometric, the functional aspect of the HFT algorithm parameter compilers are encoded into proprietary computational complexity that solves the “problem” through polynomial time within the concrete semantics of an oracle machine or “black box”.
The HFT algorithms must be able to reconfigure frequently in a compute-intensive application to efficiently respond to public data order flows, thus HFT platforms have utilized custom-optimized “field programmable gate array” (FPGA) integrated circuits.
Contemporary (FPGA) optimize I/O speed and allow for programmable logic blocks to be wired together. This is key to minimizing data flow through the bandwidth to the exchange. Though time consuming to program these “logic blocks” can be configured to perform complex combination functions.
Moreover, as mentioned above, soft processor cores implemented with FPGA logic are equally robust. Achieving the ability to be re-programmable at “run time” is leading the FinTech pack with these HFT systems; including non-FPGA architectures (
ref: Redline Trading Sources).
Today, FPGAs are being replaced with an “off-the-shelf” processor (Stretch S5000).
Software-configurations based on “clock rate”, such as the Stretch S5000 are an adaptive hybrid of the FPGA. Having software that is configurable within the processor associated with the general-purpose processor (GPPs) and DSPs and application specific processors (ASPs), serving parallelism and flexibility with FPGAs, programmable logic that is completely embedded inside the processor architecture.
Moreover, the Stretch S5000’s patented Instruction Set Extension Fabric (ISEF) is a game changer in the world of being able to program a processor for compute-intensive applications. The sky is the limit now for HFT to overcome and exploit exchange trade executions clock time.
The “ping” comes from active sonar terminology so named by Mike Muuss in 1983. It is the means of sending a pulse to a target host across the Internet Protocol (IP) network. In consideration of market exchanges, the “ping” represents a “price prob” activated by an application programming interface (API) plugged into a specific market exchange.
Consequently, by using the “ping” proprietary data feeds (expensive subscription access) have a tremendous advantage over “public” (consumer) consolidated data feeds in consideration of trade execution latency. Even though The National Best Bid and Offer (NBBO) is meant to “halt” price dislocations (latency), but it has shown otherwise.
In light of the highly advanced computational algorithm trading systems such as value weighted average price (VWAP) and weighted average price (TWAP), HFT algorithms maintain an informational advantage by remaining constantly plugged into the major exchanges (listed below) to use latency issues to their advantage.
It is no longer the case that the price shown upon trade execution will be the fill price. Maintaining NBBO “best price” is undermined even more are inter-market sweep orders (ISO) and non-transparent dark pools that send a trade execution to multiple exchanges for instantaneous execution, disregarding the “best price” regulation. This is allowed by the SEC.
“13 exchanges and nowhere to go”
Wall Street has two trading systems. Registered exchanges and alternative trading systems. The exchanges are regulated to provide the “best price” through the consolidated quotation system (CQS), and must file any rule changes with the Security and Exchange Commission(SEC).
The electronic communication networks (ECNs) and dark pools, do not provide CQS, but are mandated to match NBBO price quotations. In 2007, the Securities and Exchange Commission established Regulation National Market System (Reg NMS) to protect consumer traders from improprieties of the “best price” execution.
Reg NMS requires the exchanges to provide the quotes to the primary exchanges, such as NYSE. Data that is collected under the Security Information Processors (SIPs) for the NYSE and NASDAQ publish the National Best Bid and Offer (NBBO). Consequently, brokerage houses are required by Reg NMS to give consumer traders the best price at execution.
Out of the 13 exchanges accessed for market trading, the NASDAQ, NYSE, NYSE ARCA, BATS BZX, Direct Edge EDGX and EDGA have approximately eighty-eight percent of the lion’s share in total volumes. Dark pool trades, that are non-transparent account for more than 12% of the trading volume.
“the larger the latency, the greater the uncertainty”
For consumer traders, this short duration of dislocation of price, becomes costly in commissions while bolstering optimal profits for HFTs. Empirical data comparison from examination of publicly traded market data and data sold directly from exchanges (tapped by High Frequency Trading algorithms) proves the fact that the “latency” of the consumer’s executed trade is picked off by HFTs, monitoring the data flow with direct access to the exchanges.
In one control study between the public NBBO and a synthetic NBBO (Redline Trading Source using a software programmed processor similar to the Stretch S5000) turned up 54,734 price dislocations; tabulated within 6.5 hours of the trading session with the equity Apple (AAPL).
It is estimated that price dislocations happened every 2.34 seconds with the latency lasting as long as 1.5 milliseconds. Consumer trades went to the wrong market exchange 0.175% of the time. The average price dislocation was $0.034.
MEASURE YOUR COMPUTER’S TRADE EXECUTION LATENCY
The “packet-switched” network measured either “one-way” or “round trip” is being exploited as a “fixed game”.