gd1

Mind you, if they lower tick sizes, that means a lot more order cancellations. A market-making algo is simply a control system, taking in a set of inputs and spitting out a fair price that it is willing buy/sell something for. The number spat out is a rational one, that is then rounded to the nearest discrete tick. Smaller ticks = more jumps in the discrete output price = more cancellations/modifications. Which is fine, except the anti-HFT brigade seem to think a trading system that cancels 10000 orders and trades 100 times is the devil. It's not, it's about as remarkable as any other control system. You may as well express outrage that the cruise control system in your car adjusts the throttle ten thousand times a second. The only downside is that the exchange needs to cope with that level of traffic.

It reduces the friction in every single financial transaction. There used to be considerably more people making markets, at first yelling at each other across the pit, then manually clicking on the screens, now automated out of existence. Machines not just doing the same job mind you, but doing it much better and faster (and without sleeping, toilet breaks, emotions). So the better consistency/efficiency means these firms can quote tighter and tighter spreads in a price war, which they do. Never has so little money been leeched from the markets by middle men (less money extracted, but concentrated in fewer hands perhaps...). An enormous benefit to society. A benefit that compounds along the supply line.

The reason you hear so much about HFT is because the traditional firms it is displacing want you to.

Difference is, most models can be validated against reality. I might model a car crash impact in LS-DYNA, then I eventually crash the real thing and compare. Unless you have a spare Earth lying around and a few centuries, then climate modellers can't do that. Instead they have maybe 80 years of reliable temperature data at best (which is just one variable), on a planet 4 billion years old.

The radiative physics can tell us a lot about basic climate sensitivity, but modelling the feedbacks inherent in such a complex system must be near impossible. Saying that climate change is "simple radiative physics" is very misleading, it is a little bit like saying that if I kick a ball it will accelerate away from me and that is "basic Newtonian physics". Well yes, but I may be standing at the bottom of a steep valley, in which case the larger system exerts a negative feedback, and the ball rolls back to my feet.

Can you explain the mechanism by which billions of people will drown? Displaced perhaps, but drown? I'm fascinated. Are they incapable of, say, walking inland a little bit?

>Are you saying that global temperatures aren't rising?

They haven't for the last 15 years, so that may be exactly what he is saying.

>Is the arctic icecap not rapidly getting smaller?

It is, but is the antarctic icecap not rapidly getting larger?

Nuclear winter, Club of Rome, ozone depletion, acid rain, chicken flu, Y2K, SARS, human sacrifice, cats and dogs living together...

There are five current street circuits. You missed Melbourne, Valencia and Montreal.

Basic biology? Are you seriously asking or just being obtuse? Do you know what the world looked like the last time we had 1000+ ppm CO2? Teeming with plant life. C3 plants were facing extinction if CO2 levels dropped any further than they had, we've done the biosphere a huge favour by digging it up and putting it back in the air where it came from. Maybe we haven't done humans a favour, but that's a different conversation.

In what possible way can freeing fossilized carbon back (where it came from) into the carbon cycle "reduce long term habitability of the planet"?. It may make it awkward for some low lying cities in the short term, but long term it guarantees a warmer, moister, more bountiful planet if anything.

Coldtea: He is saying that there are obvious lurker (or "confounding") variables. It could be that not getting a degree is caused by being poor (or black, etc.) AND that not getting a job is caused by being poor. So not having a degree and not having a job may appear to have a high correlation when in fact they are both driven by a third variable (socio-economic/race/etc.). Look up Simpson's paradox.

Because it won't be overseen by the most incompetent and corrupt federal government in Australian history?

Seems fair. Some quick maths says that 200 billion dollars (just you watch it blow out) / 30 million people = roughly $6600 per man, woman and child living in Australia. Probably $12k per taxpayer. Seems fair to make the user pay for a bit of that rather than dump the debt on the next generation. I know people don't like to let little things like numbers intrude on their socialist utopia, but this is real money I'm afraid...

"Let them eat iPad"

http://www.forbes.com/sites/afontevecchia/2011/03/11/ny-feds...

I disagree. You could argue that a musician probably thinks "I have to play a D# for one and a half beats" as well. Or they can draw a dotted quarter on the sheet. We have symbolic languages for a reason - they are, once learnt, superior. If anything code needs to move further away from spoken language, more in the direction of APL and its descendants.

A skilled musician likely doesn't engage the speech centres of their brain, they see a note on the sheet and translate it to motion. You should be able to take in the symbol for "apply a function to each item in a vector" at a glance without any clumsy English getting in the way. APL had it right, but coding has been crippled by catering to the lowest common denominator.

Black holes are capable of bending light back towards us:

http://rein.pk/gravitational-lensing-to-observe-ancient-eart...

And? How does that make it any less useful to have the existing code as a guide?

His 'second jury' were comical.

"Most were surprised by how little he had taken in relation to the whole: eight megabytes in a platform that consisted of an estimated one gigabyte of code. "

Really? One gigabyte of code? Or one gigabyte of third party binaries, precompiled libraries, documentation etc.? All you need are the source files, and those language keywords compress nicely. Exactly how much plain text do you think they have?

' "But that’s the secret sauce, if there is one,” said the juror. “If you’re going to take something, take the strats.” '

Rubbish. The strats are the easy (or at least easier) bit in the high frequency space. That's like taking a dump of the html from facebook/twitter and thinking you've stolen their site. It's the scalable, distributed infrastructure that drives large websites, and it's the fastest possible infrastructure that drives HFT. One thing Lewis got right is the Coke/Pepsi analogy, and it really is that simple. A straight up race.

" “It’s way easier to start from scratch.” "

It is, and it isn't. In even the worst codebase there will be nuggets of usefulness that will accelerate the process. Isolated parts of the system that you've perfected. Exchange message decoders for example (ITCH, FIX/FAST, etc.). This is especially true when you, the genius who will be starting from scratch, has worked on it for a few years. I assume he was doing something useful in his time at Goldman.

I'm not buying the innocent act.

Don't mistake looking complex for being complex. These languages have very simple and clean grammars (or they'd fall apart quickly), which is quite a contrast to the dog's breakfast that is modern C++. A simple grammar = simple to read once you've learnt the syntax. There's no issue in having a lot of symbolic operators if they all work in a clean and consistent way.

I find it much easier to read some Q/KDB I've written earlier than the equivalent C++ or C#/Java. For three reasons:

1) It is extremely terse. A small block of code will do what pages of boilerplate class definitions, interfaces, setters/getters, declarations, typedefs etc. will achieve. There is usually no need to navigate through a spaghetti chain of virtual function calls.

2) It is interpreted. I can quickly (without compiling) cut out chunks of the code and run them to understand what they are doing. I can create variables on the fly, fill them with test data, and run the function. If it calls another function, I execute the name of the function and it returns the function code, since functions themselves are first class variables.

3) It actually becomes easy to read because it is easy to parse, for both machine and human. K/Q is read right to left, and so breaks some traditional conventions. For example, 5*6+3 will evaluate to 45, not 33. At first this can be disconcerting, but eventually you will realise that parsing a line of code no matter how complex it looks is quite simple - you mentally start at the rightmost token and start gobbling. This is probably why the interpretor is so fast as well.

As someone mentioned below, I'm not sure why you'd consider it "too cryptic". Anything is cryptic when you first encounter it. I suppose the difference is that Q/KDB provides an (extremely valuable) solution to the financial world - the handling of big data long before it was even called that (a tick database with billions or trillions of rows) so there is a valid motivation to learn the language and it doesn't feel cryptic at all once you've immersed yourself in it.

He said it was 5% for mothers that have 14 drinks per week or more. It will be much less (if not zero) for someone who has 2-3 drinks a week, which is his point.

If you want to buy 1000 bitcoins then cross the spread, take out the best offer, and buy 1000 bitcoins. If, on the other hand, you want to stick a bid into the market in the hope that someone else will cross and take your (lower) price instead... then yes, other participants will react to that.

Even worse, not all lines of code are equal. Across languages, but also within them. C/C++ can be terse or verbose depending on style and the subset of features employed.

Nine tenths as much to be correct. From wiki: The word decimation is derived from Latin meaning "removal of a tenth". A cohort selected for punishment by decimation was divided into groups of ten; each group drew lots (Sortition), and the soldier on whom the lot fell was executed by his nine comrades.

>In exchange for the risk of getting caught glad-handed during severe events the specialist was given the privilege of making a spread the rest of the time.

No, they had old fashioned ways of withdrawing their liquidity. Not answering the phone, or pretending they can't see you across the pit.

>We allow them to effectively front-run through physical location and latency advantages (that is, they see quotes and price before anyone else and thus can react faster than anyone else)

That isn't front-running. Front-running means seeing the order flow before it reaches the matching engine and stepping in front.

>You can see the size map essentially disappear as the market's depth

When depth levels are taken out you have a great big empty space between the bid and the ask. It may have been 20 ticks wide yesterday (it was about 8 on eurostoxx, and those are chunky ticks). It's empty because someone has traded through those levels and taken all of the liquidity, so naturally it takes time for new orders to replenish the market. It isn't only because participants are withdrawing liquidity. Also, greater volatility means a greater dispersal of opinions about the correct price - so even when orders are put back in following the shock they are likely to be spread over a wider range of ticks. If you're a market maker (algorithmic or otherwise) who has just been smashed on the bid and seen the market drop 30 ticks, where do you put your new bid in?

The disinformation campaign by Wall Street against the HFT bogeyman is impressive. It isn't much different to record labels against MP3s or any other industry that is being disrupted by technology, but it is surprising to see a lot of the HN crowd falling for it.

No. It's an uncertain Bayesian world. You evaluate on the probabilities. Given a 20% chance that tweet is true, then what is the S&P worth now? Less. So you act. You don't sit around waiting to find out...

Actually it isn't dumb. If you had to assign a probability to that tweet being true when you read it, then what would you say? 20%? 40%? So yes, given that there is a real probability that the markets are about to tank, that would effect your valuation instantly.

Why on earth would you want market makers out of business?

What does this have to do with HFT? If I had someone (man or machine) managing my money and a credible source had reported two explosions at the White House, and there is some chance it could be true, I would like them to incorporate and react to that information and the likely outcomes as quickly as possible. And then if it turns out to be false, I'd like them to react to that as quickly as possible as well.

Which is exactly what happened. No different to a rumour rippling around an old school trading floor and then being discredited. Just faster and more efficient.

GDP isn't your income, tax receipts are. Tax receipts as a percentage of GDP are about ~25%, so 100% debt/GDP is 4 years worth of debt, not 1.

Exactly what kind of "austerity" do you think is happening in Britain? Government spending has gone up. Saying something doesn't make it true.