Comment by cma | Hacker News Reader

Comment by cma | Hacker News Reader

A good rule of thumb is if you are working in an industry where, if everyone agreed to work half as hard (or half as frequently), the end results would still be the roughly the same, you are in a largely zero-sum waste of time job.

Latency arbitrage: if the underlying price change in gold commodities takes one millisecond to be reflected in the stock price of gold-holding/gold-mining companies rather than half a millisecond, utility to society isn't going to go down anywhere near half. And since the awards are meted out solely based on ranking, not on absolute gain in speed, there is very little relationship between how much utility is provided vs. how much the provider is rewarded.

Index arbitrage: if an index fund is priced +- .001% off of its underlying basket of securities rather than +- .002%, utility isn't going to go down by half, and you certainly wouldn't want to pay some HFT guy to make it happen--he ends up charging you one cent to flip a fair dime on every trade rather than flip a fair quarter for free. Who do you think wins in that scenario?

Lawn care/car care/dry cleaning/etc. etc.: if everyone's grass got mowed/car got cleaned/clothes got ironed half as frequently, there probably wouldn't be a 3% drop in the utility provided (if your neighbors kept up the pace though, it wouldn't be the case).

Your first example of latency arbitrage is correct. But your "index arbitrage" example is confused.

If you place a sell order for your index at a price $0.02 less than the cost of the basket of securities, why don't you want an HFT to fill your order? If you didn't want your order filled, why did you place it? Would you prefer it if a human daytrader placed the order, or if a retail investor cashing out his IRA did it?

I'm also confused by your claim that the HFT is "charging you one cent to flip a fair dime". Could you explain what this means?

He's saying (I believe) that the only positive externality of HFT is a slight reduction in pricing errors. If the pricing errors were very large, then HFT would be providing a valuable service. The implication of the quarter versus dime metaphor is that the efficiency gain provided by HFT is very small.

However, my own study suggests the true situation is worse than he claims. Though HFT does appear to increase efficiency most of the time, it also creates sudden flares of inefficiency at the times when a well-functioning market is most needed.

How does HFT create "sudden flares of inefficiency"?

He's probably talking about stop orders that lots of people put in for outsize volatility because the regulators do rollbacks. That means liquidity dries up, right?

Assuming you mean stop loss orders, HFT's don't use them. Idiot retail investors do. I have no idea what the rest of your comment is talking about.

HFT provides liquidity(a good thing) except when you most need it in times of uncertainty.

Example, was the flash crash(I read the report and other sources), HFT algorithms (and people too) decided it was getting too risky to provide spreads and just decided to stop trading.

The old market maker system is gone, and we really do not have a substitute that can handle the unpredictable.

That is, HFTs are free to get paid for providing liquidity, but they are also free to pack up the shop and leave, if "things get heavy".

This does not serve institutional and retail investors.

During the flash crash, HFTs did not pull out due to market risk. They pulled out due to regulatory risk. See my explanation here.

http://news.ycombinator.com/item?id=2875927

If HFT's save retail investors and institutions billions of dollars most of the time, but fail to do so for a single 1-hour period when they run the risk of regulators screwing them over, what's the problem?

> Example, was the flash crash(I read the report and other sources), HFT algorithms (and people too) decided it was getting too risky to provide spreads and just decided to stop trading.

Note the word "people".... Are you claiming that someone (or something) has an obligation to trade? Who? (If the answer isn't you....) Why?

> The old market maker system is gone, and we really do not have a substitute that can handle the unpredictable.

We had "flash crashes" during the old market maker system too, so if you're claiminging that human market makers prevent flash crashes ....

If you work in an industry where this isn't true, but depends on the largess/spending of people who work in industries where this is true, are you still in a largely zero-sum waste of time job?

More simply, how many levels deep do you carry this analysis?

Also, can you guesstimate how much of the income generated in the US is not derived from zero-sum waste of time jobs/products when you carry the analysis at least 2 levels deep?

When does the latency arb trade add value in your mind? If delays are 100 milliseconds? 3 seconds? Half a minute? 1 hour? Etc.

You might have to amend that -

... in an industry where, if everyone agreed to work half as hard or half as frequently, the end result would be an improvement for the overall economy.

Reply on news.ycombinator.com