391 Auctions

Jason Zweig of the Wall Street Journal has an interesting piece up called Could Computers Protect the Market From Computers?  I appreciate Jason, he writes a lot of intelligent stuff, and had the guts to revise one of my favorite books, “The Intelligent Investor.”

We are talking about positive feedback loops, where computers amplify the actions of humans demanding action now.  Computers, for all of their strengths, are rules based, and we the humans feed them the rules, or the information that allows them to react to data as they emerge.  The rules may be very complex, but they are rules, and do not allow for humans to modify the computer’s reaction to the market on-the-fly.

I’m skeptical that we can stop unusual things from happening resulting from computers trading rapidly by having other computers monitor it.  First, stocks are volatile, and news can break that leads to significant rallies/declines.  Second, part of the difficulty from the “flash crash” was computers getting out of sync with one another.  We can’t guarantee that the regulatory computers might not fall behind the trading computers, and what might happen if the “right” action to slow trading emerges slowly.

Third, one has to recognize that you should only have regulations that are understood easily by participants, and accepted, or else the rules will face a lot of lobbying pushback.

I think that there is little to no gain to the market as a whole from sub-second trading speeds.  I think we could slow down the market, and force a more rational market than what we currently have, by limiting the ability to cancel orders — all orders must be good for at least one second.

Markets need good rules and structure to work well.  Rather than having shadowy computer overlords, which only academics could like, craft a rule that says, “One auction per second.”  Or create a central order book and eliminate alternative venues for execution.  The cost listed in the article is cheap.  I’m agnostic on what the best solution is, but to me, the best solution involves slowing things down, so that information does not cause cascades off of short-term signals.

Even simple rules like, “Stop trading for any company that has dropped/risen by more than 5% on the day for 30 minutes,” would be preferable to any guidance from computers that is less clear.

Rather than using computers and complex reasoning, we need simple rules to slow things down, or…who cares, let errors happen.  I made money on the day of the “flash crash” by buying shares of a company that was solid but temporarily depressed.  Teach people not to use market orders or they could get harmed.

This is the market, after all, and if you are “bellying up to the bar,” you should be ready for the fact that you are outgunned.  You are likely not smarter than all of the resources being deployed against you by hedge funds, high frequency traders, etc.  Secondary markets in equities exist to provide flexibility to holders of the equities, most of whom hold their stocks every day, with only a small fraction trading.  Trading is a sideshow to value creation, which happens in the companies, not the exchanges.

Which makes me take step back and mention that Buffett wouldn’t care if the exchanges were closed for a year, because he buys solid companies.  Suppose for a moment, I had written an article called 391 Auctions, where I would suggest that the markets have one auction each minute, and that all orders must last until the end of the minute, with no cancellations.  (After I wrote, this I changed the article title, so I did do it.)

With 391 auctions per day, who couldn’t think that we were providing enough opportunity for price discovery each day?  Slow things down, and ignore those arguing for technical efficiency versus those arguing for rational markets that allow people to make reasonable decisions in real time.  One auction per minute?  Could work well — watch the bids and asks line up, once per minute.

Markets need structure to work well.  This could be one way of doing it; I am open to other ideas, but letting the computers attempt to do it opaquely seems like a loser to me.  Slowing things down seems like a winner, because secondary trading is a sideshow to the real value creation that happens inside the companies.

4 Comments

  • Lloyd says:

    Speed will still matter.

    In the case of one auction per second it will be the speed of being able to place one’s order last. For most listings that means most retail orders can still be ‘picked off’ by HFT much of the day, if an edge is seen. The fight will be still be won by the swiftest.

    It could potentially change the dynamics of the most liquid stocks and those with temporary high liquidity. There will be more instances of crossed orders as one example. It could be like having 391 opening auctions every day. In some respects that would encourage use of the market order for those liquid listings!

    I still favor a single opening auction at the start of the day plus reopenings after rare halts. The one second execution window for all orders should be powerful without creating additional complications. HFT will still pick off retail trades but price discovery will be limited to real activity. No doubt it could still be gamed, but it seems a natural first step.

    • Not really, because all orders are secret until they are revealed when the second or minute is up. There is no advantage. This is a sealed-bid auction; no advantage to going last, because you don’t know any more than the next guy.

  • Lloyd says:

    Correction. I misread the article originally. One auction per minute is different than one auction per second.

    That could be interesting.

    The preauction order pile up might be largely carryover limit orders and fresh retail. Market making would be more similar to its origins.

  • Shrivathsa says:

    This would go against the natural evolution of Trading.

    My two cents for what it is worth would be for computers to call for flash halts in the event of anomalies detected.

    If a stock normally trades up or down $20, like say Apple, that is is its range, while if it does not budge more than $1 like AT&T, then that is its range.

    Any shift beyond the range will call for a halt for 30 minutes.

    After that, let the merry-go-trade continue.

    Instead of doing this through humans / exchanges, better to do it through computers.

2 Trackbacks