Notes for Chat with Traders, Episode 151

Added on by C. Maoxian.

Episode 151 ... Rick Lane (66:03)

Trading Technologies CEO
Recently purchased trade surveillance software company, Neurensic
Finding bad behavior while avoiding false positives
Accurately identifying trader intent hard to do
Machine learning can be applied, fewer false positives
Neurensic score useful for compliance
MiFID II compliance ... European financial regulations -- futures markets
Every algo needs to be tested, see if it sets off red flags
AI will have applications across trading, not just compliance
MiFID II becomes live January 2, 2018 ... all automated traders aware of this
Lane formerly worked for consulting firm, modeling and simulation of war games, combat modeling, terrorist network growth
Cousin was big floor trader in interest rate futures
Cousin anticipated end of physical trading floors, knew everything would be automated
Cousin invited Lane to come work in Chicago
2005 the trading floor was still madness, 2006 you could hear a pin drop
Applying graph theory to terrorist network modeling
He's a coder, not a trader
Built visual interface that floor traders could use on their handhelds, any dummy could use it
[Lane has similar voice and pacing as Obama ... a Chicago thing?]
Technology had disrupted every industry, cousin anticipated same for futures trading
Georgia Tech undergrad, zero exposure to financial markets
Took him six months just to learn the mechanics of markets (what's a bid and offer, etc.)
Ultra-low latency trading strategies can be dangerous, need serious safeguards
Trading tools as loaded weapons ... risk of serious loss
Writing code for traders serious business, can't be cavalier
Software they wrote for internal purposes called "Algo Design Lab"
Trading Technologies, the 800-lb. gorilla, didn't have anything like their tools in 2008  
Trading Technologies acquired their software in 2010
Lane left and went to Google, but came back to TT within a year
He's a developer who is also a CEO ... an analytical thinker, still builds code to this day
Scale and latency problems guys at Google face are a cakewalk compared with FinTech problems
Shaving five microseconds off from tick to trade is a great  challenge
30 milliseconds is a lifetime (time to return Google search results)
Different exchanges, different rules, different regulatory regimes -- all challenging problems to solve
The most sophisticated hedge funds / prop firms have already reached the limits of latency race (nanoseconds ... speed of light constraints)
Trading algos embedded in FPGA, in silicon chips, in network switches
Making a fix in an embedded algo in a network switch hard to do
Hardware-based solutions for automated trading makes it fast, software too slow
Switches themselves can decode price updates, gain meaning, and change its own price; never sent to server
TT has big internship program
Definition of "trader" has fundamentally changed in recent years
Now traders are all computer science, math, engineering grads [not Neanderthal basket weavers like me]
Twitter: @r1ck_l4n3