Comment by jackgavigan | Hacker News Reader

Comment by jackgavigan | Hacker News Reader

With the electronification of markets, I think retail investors are always going to be at a disadvantage in volatile markets like Monday's. The speeds at which the markets operate today is far beyond anyone's ability to keep up with, even if they weren't at the end of a retail internet connection, with at least one layer of abstraction (i.e. their broker) between them and the market.

Fast human reaction times are on the order of 200ms, to which you need to add network latency. The HFT firms operate at single-digit millisecond timescales. By the time a retail investor sees a price on the screen, it's probably already gone.

However, there are steps the exchanges could take to reduce volatility, especially when liquidity is thin, which would help prevent the triggering of stop-loss orders. For example, they could switch from continuous CLOB trading to periodic auctions until the market stabilises. I reckon that would go a long way towards reducing instances of wild swings like we saw on Monday.

Would you rather be a retail investor in a fully automated market, with no opportunity to compete on reaction time and uniformly terrible prospects for day trading, or would you rather be a retail investor in the 1980s, paying extremely high spreads set by colluding market makers?

I'd rather the SEC mandated volatility interruptions with auctions to stabilise volatile/illiquid markets in order to reduce the likelihood of big price dips (or spikes, for that matter) that cause retail investors to get stop-loss'd out of their positions.

You realize things keep happening while your volatility interrupt is in force right? Everything else in the world keeps trading. There is no guarantee when it starts trading again that it won't open 20% down. If you don't want a stop loss to trigger, don't place one.

Imagine you have a stop loss, which doesn't trigger, then the market vol interrupts, then the price opens 10% below your stop loss.

Oops, you'd be moaning about that instead.

> There is no guarantee when it starts trading again that it won't open 20% down.

That's why I used the words "reduce the likelihood" instead of "eliminate".

As to the efficacy and benefits of volatility interruptions, research has been carried out on this - 'Price Discovery in European Volatility Interruptions': http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2365772

Monday saw the first ever limit down halts on major equity index futures (YM/ES/NQ) on CME Globex. It was an unmitigated disaster. It happened at a critical time (heading into the cash open), and caused chaos in the ETFs that have these futures as the underlying. Market makers price and hedge ETFs like SPY (the Spyder S&P 500 tracker) off the future. So being unable to do so since the underlying was frozen, they did the only thing they could naturally do and blew out their spreads and backed away. Even then, some unfortunate and panicky people crossed those spreads and some SPY changed hands at -14%. There is no legislating against stupidity.

Then when ES and the other futures did unfreeze, they promptly plunged to a second limit down in extremely thin trade. Why so thin? Because when you stop trading, everyone empties out of the orderbook, and they don't necessarily dive straight back in once it unfreezes. Liquidity gradually comes back as participants assess that the price is stable and they can start quoting again.

Monday was not a good day for your argument.

As it currently stands I don't think there's any guarantee that your stop-loss will sell at the price you've set anyway, rather than 10% or 20% or even more below it. The only difference is that right now it could cause all your shares to be sold for less than you expected because of a temporary market blip that lasted seconds.

Clearly, both outcomes are suboptimal and inefficient - for retail traders. Both are actually instances of phantom liquidity, where the tape doesn't match the actual price that is being executed.

One also has to keep in mind that retail traders are often matched with other retail traders in the same brokerage; their orders probably never make it as far as Getco, Citadel, etc

NYSE's concept of DMMs is not a bad compromise between the Wild West of full automation and the corrupt 80s fatso that takes a huge spread. Unfortunately, this solution doesn't scale well.

By definition, a 'wild swing' is a very short timeperiod event which typically doesn't affect prices too much on average. If I hold stock and it's down 50% on one day and recovers on the next, I don't really care because I'm not trying to make money off market events. What I actually care about is the stock's performance over months/years.

For most retail investors, market blips, and thus their access to low latency are irrelevent because they don't really alter the month/year picture.

> If I hold stock and it's down 50% on one day and recovers on the next, I don't really care...

You do if you get stop-loss'd out of the position at an unrealistically low price.

Which happened to me Monday morning, although not down 50%, and I had just two days before put that stop in place because earlier in the year I had lost a ton of $ on two different stocks by not having one. I think people like me are easy pickings for HFT algorithms.

So you put in a stop loss order and didn't want it to trigger when the share went down? Perhaps it's not the correct hedging tactic for your trade? Maybe you should have been using options?

If you don't understand proper hedging for your trade, you aren't easy pickings for HFT algorithms, you are easy pickings for everyone in the market.

My point is that stop loss orders, which exist because they are a prudent way to limit your losses, are highly susceptible to flash crashes, and depending on their actual implementation, possibly subject to exploitation. In retrospect, in this case I'd undoubtedly have been better served by options, but options are (or can be) complex, it is a whole new way to lose money.

History has shown me that I shouldn't be trading individual securities!

Based on your statements, you shouldn't be trading anything. If you don't know how to use the tools of the trade, you are all but guaranteed to lose. Not every trader is an options trader, but every trader should understand the fundamentals of options. Trading is a specialized business all its own, no less so than dentistry, chemical engineering, or lobstering. If you haven't attained the required expertise, the results are going to be awful. Non-traders would all be much wealthier if their brokers took away their Sell buttons.

I think that's unnecessarily harsh. If a retail investor had set a stop-loss on a stock like the RSP ETF [0] that got triggered at -50% because the price diverged hugely from the NAV, I think they have cause to feel hard done by.

Simple solution: don't set up a stop loss. Accept that the best you personally can do is track the market, and if the market crashes so will your portfolio.

Stops aren't for investors, so "tracking the market" isn't relevant here. Investors get their returns primarily from dividends and possibly (based on luck) from buying low and selling high, on the rare occasions they sell. Traders get their returns primarily from buying low and selling high at a later time as well as other strategies such as various forms of arbitrage. Trading is itself a business, and as such a trader's returns do not track an index any more than a single arbitrary company's profits do.

A trader will usually have many positions open, and successful traders have a plan for every position. Stops are one tool that a trader can use to execute that plan, usually to cut their losses on one of the many trades that they know will go against them. A loss on a particular position is not a disaster and every trade's plan accounts for the possibility of having to exit with a loss. This is much more important to traders than investors because traders make or lose money based on the behavior of the market; investors make or lose money based on the success or failure of the underlying businesses they own, so buying or selling an asset based on market behavior is not appropriate.

Traders are well aware of the risks of using stops: information leakage (stop-hunters), exposure to bad data, and bouts of illiquidity. Some of those risks can be mitigated by using stop-limit orders instead of the normal stop-market type, but not all, and stop-limit orders carry the risk of accumulating additional losses without exiting. Everyone knows all of these things; there are no surprises. Weep not for the unfortunate trader. He knows it's just part of the business.

That was pretty much my point. If you don't have a particular advantage over other traders (i.e. at least that you're doing it full time, and not just after hours), stop trading and start investing.

Why does trading have to happen at a millisecond timescale? Wouldn't it be much easier on everybody involved if trading followed a clock tick of one second or one minute or something?

But then how would companies like Goldman siphon off everybody's money?

HFT is just another name for front running at the millisecond scale.

https://en.wikipedia.org/wiki/Front_running

There's a difference between knowing someone placed an order and using that to front-run, and guessing that someone will place an order and using that to HFT.

It is a distributed systems problem.

First in an order book you have many different orders for many different prices and quantities on both the buy and sell side. At any given price, at any given time, in an order book you can have more buyers at a price than sellers (or vice versa). How do you decide tie breakers? One common way is to do FIFO matching, this means that the participant that had an order in the longest gets filled before someone who had one in a shorter amount of time. The higher your time resolution the less viable this gets.

Second, there is also a desire to get OUT of the market. So as you process new information you realize that your current orders are riskier than you anticipated, so you want to cancel them. The longer it takes you to cancel them the more risk you are exposed to. In the case of market makers this risk will get passed on to their customers in the form of higher spreads and less liquidity.

Third, there is no 1 place where trading happens. Trading happens in hundreds of venues around the world in just as many jurisdictions. They would all need to do distributed clock/queue sync for higher timescales to feasibly change anything. Global distributed clock/queue sync is a Hard Problem TM.

In short, it is much harder to do non-continuous time trading than the alternatives without obvious benefit and with lots of potential unintended consequences.

Reply on news.ycombinator.com