With the collapse of Knight Capital recently, there has been a lot of scrutiny of High Frequency Trading (HFT). Breathless reporters have been bombarding us with articles suggesting that there is danger out there.

That's all nonsense.

There are two general approaches to building a robust system with high reliability …

Scott Aaronson has a great article explaining how quantum mechanics should be taught. The basis of his idea is to describe quantum mechancis as being the mathematical structure you get when you attempt to generalize probability theory to include negative numbers. Once you do this, the L^2 norm replaces …

Ron Unz over at American Conservative has written an excellent blog post called Race, IQ and Wealth. It's well worth reading, gets into data, and is a fairly dispassionate analysis of the topic. It also (politely and subtly) calls out the supposed "experts" on the field, who ignore all sorts …

One of the more important tools in my computational toolbox is linear programming. It's a great way to solve a lot of otherwise difficult problems in a straightforward, nearly black-box manner. I've discovered that a lot of programmers aren't too familiar with it, so I'm writing this tutorial that explains …

Previously I argued that some chunk of HFT is rent seeking due to the minimal tick sizes induced by the subpenny rule. It appears now that there is a proposal to the SEC to increase the tick size in order to increase market making activity. I.e., if you raise …

In a recent post, a company selling A/B testing services made the claim that A/B testing is superior to bandit algorithms. They do make a compelling case that A/B testing is superior to one particular not very good bandit algorithm, because that particular algorithm does not take …

Machines are increasingly replacing humans across many walks of life. Huge numbers of jobs are being made obsolete. Historically this trend has been a positive force in terms of living standards. Additionally, as old jobs have been replaced, humans have found new work to do.

The claim that machines will …

In a series of blog posts, Gary Rubinstein attempts show that the Value Added Modelling scores recently released by the NYC Department of Education prove that VAM (Value Added Modelling) is not accurate.

However, whatever flaws VAM may have, Gary Rubinstein (henceforth GR) hasn't demonstrated them. All he has demonstrated …

Just thought it's worth pointing out a couple of responses to my post proposing the elimination of the subpenny rule.

Alex Tabarrok discusses this on Marginal Revolution. In particular, look at David Krych's comment criticizing my proposal - I haven't had time to fully evaluate it, but it seems insightful.

Dealbreaker …

In previous posts, I discussed the basic mechanics and social utility of high frequency trading. Of particular import is that I characterized the latency arms race as socially wasteful. Now I'll discuss a policy proposal which might mitigate the harmful effects of the race for latency, while giving better prices …