Wednesday, May 19, 2010

Ban Market Orders!

The SEC, in their 150 page report on the May 6th crash, has section (around page 75) on "potential regulatory responses".  I found this interesting, since I've been constantly repeating that we should ban market orders if we want to protect people from themselves:

"We are considering ways to address the risks of market orders, and their potential to contribute to sudden price moves. Areas under consideration include: (1) requiring market order “collars,” thereby effectively converting market orders into limit orders; (2) prohibiting or limiting the use of market orders; (3) requiring broker-dealers to specifically warn retail customers about the risks of market orders, particularly in volatile markets; and (4) pursuing investor education initiatives as to the risks of market orders."

I'd be shocked if they implemented #2, but it's the simplest and most effective solution, and removes any and all potential excuses from the execution side of things.  I've specifically discussed all four of the potential changes the SEC detailed in that paragraph above.    Mary Schapiro - you could at least give me a hat tip...



Anonymous said...

Funny how times change.

In, "Reminiscences of a Stock Operator", the subject (an infamous day trader) advises specifically against limit orders.

He provides the example of getting off his fishing boat offshore of Miami (or somewhere there abouts) only to return to his hotel to see the markets melting down.

Knowing he could never get in his order in a timely manner from the hotel in Florida (over the telegraph wires), he simply sends out a "market sell" order to his "trusted" brokers, who effect the order by the end of the day (presumably in a manner so as not to further pressure the market with this large sell), whilst he boards a train to head back to NY to clean up the mess.

An anachronism of a bye-gone era.


James said...

No. Ban idiot over indebted governments. kill the debt and yo kill the crashes

Anonymous said...

Ban market orders?


How about we stop the nanny-ing of traders and investors.

Grow a pair.

Kid Dynamite said...

i hear you, anon, but it's clear that America wants to protect the ignorant from themselves. Since market orders definitely do NOT mean what people ASSUME they mean, everyone would be better off if they couldn't blow themselves up with market orders.

Anonymous said...


I am confused.

On the one hand, you are mocking those who are oblivious about what a market order actual is and how you can get burned by it.

Yet, two posts earlier, you are rallying against ZIRP as a cause of some smart people and money managers making dumb investments in junk bonds.

Even in a ZIRP environment, I am unaware of any type of requirement that money managers must invest in junk bonds that are risky and have a good chance of blowing up.

Do money managers need to be protected from themselves just like Joe Public?

Kid Dynamite said...

anon - i think i've been pretty consistent all along in saying that money managers have no excuses and need to be held accountable - we need to stop holding them up as victims. instead of blaming the "big bad banks" for "selling them crappy products that lost money," we need to blame the money managers who did no work and tried to pickup free yield and got burned. THEY are the ones who actually failed in their fiduciary duties, and they are the ones who need to be removed from their positions.

I've relentlessly insisted that ACA are the real criminals in the Abacus case, regardless of GS's guilt in the matter on the technicality of non-disclosure (which had no effect on the actual value of the underlying assets or portfolio)

I was saying that the Fed is deliberately blowing a bubble through ZIRP. do you disagree? I wasn't making excuses for money managers.

as for market orders: look - i don't think there should be any crying in the markets - but society clearly disagrees. If people want to whine when the market tanks and instantly rebounds, then we might as well try to prevent it.

see, the good thing about banning market orders is that it takes away ALL excuses on behalf of anyone who feels like they got screwed - YOU enter your own limit price.

SINCE society seems to like to make everyone into a "victim," I respond by saying that we need to eliminate excuses and hold people accountable for their actions. Once you make it so that no one can play the "oh, i'm a poor hard working retail investor, i thought a stop loss order was supposed to portect me" excuse, we'll have accountability. YOU entered the limit price. If you didn't want to sell the stock at that price, you can't blame anyone but yourself...

that's what i'm getting at. it's not radical.

scharfy said...

Well whatever the plan may be, the various exchanges and regulatory agencies hopefully will get their shit together and fight to keep the American markets the most liquid, open, transparent, free - and thus the most sought after - capital markets around.

We cannot have stocks printing a penny, various exchanges busting different swathes of trades, different curbing policies, various disorganized circuit breakers - etc, etc.....

That being said - I don't think banning a particular "type" of order is necessary (though given how fragmented things are I can see the logic) . In a computerized market applying some simple, common sensical curbs is all that is needed. Christ, put em 40% wide - all you need to do is prevent "absurdity". Keep it simple and let the market work..... But for the "ridiculous" prints, the flash crash could be viewed as a successful stress test of our markets.

Kid Dynamite said...

yeah scharfy - the markets bounced back quickly, so that was robust.

my point is simply this: people who enter "market" orders absolutely positively do NOT mean that they really want to sell it at market. they THINK that's what they mean, because they THINK that they are entitled to some sort of liquidity near the current price, which they are not.

SINCE people don't really mean "sell at any price" when the enter a market sell order, I think it would be better for everyone if they entered the actual limit below which they would not want to sell the stock

and guess what - if you want to sell at any price, you put sell @ Limit = 0. or buy @ limit = infinity (credit Steve Waldman with these boundary scenario simplifications) - it doesn't take away anything from anyone - you can still sell your stock to zero if you really want to!

Anonymous said...

The SEC has still not posted the rule on their website (rather, they have just announced yesterday that they will post the rule...), but it is available from the exchanges:

I'll write a comment letter to the SEC if they do in fact hold a 10-day review period, but given that this is supposed to start rolling out June 7th, I think any flaws in the approach are already baked in, since there really is not time to have a 10 day comment period, have the SEC review the comments, decide on any changes to the approach, and still have the exchanges do the needed work to start a rollout 11 business days from today.

The gist of my gripes with the approach:

1) It is built off of the last trading price, which NYSE and Nasdaq will filter for erroneous trades and out of order trades. This means that just watching the tape, noone but the listing exchange can determine when a halt is coming.

2) The trigger point is off of trades, so to the extent that they didn't want the tape to be painted with crazy trades, this will fail. It will only take one stock to have an order book cleared with market orders for a crazy trade to go through, starting a halt.

3) It is not clear from my read after case #2 what the 10% move might be. Suppose a halt is called after Accenture next trades at a penny (which assumed that all other orders in the book which were swept by market orders were within 10% of the recent 5 minute window price). After the period where things are halted temporarily, is there no established last trading price for further comparisons, or is the 0.01 trade included, which would prevent a reopen. I'll have to re-read this part tomorrrow, but it didn't look well specified to me.

4) Computationally, this is not trivial. For each of the S&P500 components, a history of all trading prices from the consolidated tape needs to be maintained in a rolling 5 minute window, and the high and low from that window needs to constantly be evaluated to see if a trade is breaking the 10% movement barrier. That's materially more processing per symbol than I imagine the exchanges are currently processing, and I can easily see this slowing down the listing exchanges (NYSE and Nasdaq) if the implementation is not done well (which is hard to do in 11 working days!).

This really should have been based off of the NBBO, with different triggers for bid and ask movements, but it really appears that the rush to appear to do something constructive here will land us with a sub-par approach.


Kid Dynamite said...

interesting thoughts, Peter. I think the point was that they expect that ACN wouldn't trade from $40 to $38 to 1c. that there would be some trade at $36 first, which would halt it. I agree that this isn't ALWAYS the case.

i like your point 4 about the computational requirements, but I think that is solvable, even if not trivial.

Anonymous said...

KD> I think the point was that
KD> they expect that ACN wouldn't
KD> trade from $40 to $38 to 1c.
KD> that there would be some trade
KD> at $36 first, which would halt
KD> it. I agree that this isn't
KD> ALWAYS the case.

Agreed - but the events of May 6th and any firming of the erroneus trade rules will make that kind of movement much more likely in the future.

Here's the salient section from the proposal.

SEC> To attempt to ensure that
SEC> erroneous executions do not
SEC> trigger a trading pause, the
SEC> Exchange also proposes that
SEC> it can exclude a transaction SEC> price from use in calculating SEC> price movements if it
SEC> concludes that
SEC> the transaction price
SEC> resulted from an erroneous
SEC> execution.

What this means to any automated liquidity providers or stat-arb funds, is that they will withdraw liquidity as a stock trades near either the cutoff point for an erroneous execution, or the 10% cutoff for a trading halt.

Suppose that the definition of erroneous trades is 5% from the last recorded sale price (which is not a bad guess at what might be used, see here:

You could easily envision a stock trading down on bad news 9% over the last 5 minutes in more or less orderly fashion (i.e. no 5% gaps). Now with 1% headroom on a halt, all liquidity providers will pull out of the market for that security, wiping clean the bid side of the book. That will happen because noone will want to bid on an issue in free-fall with the knowledge that they'll be unable to trade out of the position for an extended period of time once a halt is called.

If however a stop loss is triggered after the liquidity providers have all pulled out (because the SEC has not taken KD's advice!!), then the stock is basically guaranteed to gap down to the stub quote bid price.

... continued

Anonymous said...

Now, according to my read of the situation, that trade would get flagged as clearly erroneous and would get reversed - in which case no halt would be called, or the trade would go through in which case the last recorded price for the security will be at 1 penny and a halt will be called.

So, either there will be air pockets formed at trading near the 10% cutoff point where the entire book is wiped clean of bids but a stock is still "trading", or some trade will go through*** that is eventually not declared erroneous and then the stock will be halted a few milliseconds later.

*** This creates a really interesting situation that will surely be gamed. If a stock is down 9% and has not been halted yet, but the drop 9% has been "orderly" with no major gaps down in trading prices, then someone can engineer a halt by placing a bid for 1 share at 4.9% off of the last trading price (movement just under the erroneous trade cutoff), and waiting for the next market order to hit. That single share would then cause a trade to go through at -13.9%, would not be erroneous by a 5% cutoff definition, and would engineer a halt. It's easy to see how someone shorting the futures or ETFs for a particular index which includes that security could profit from that, so there is an easy mechanism built in for someone to engineer a halt and a 4.9% movement in an index component with negligible financial exposure.

As I see it, they have created a new set of conditions that are going to greatly increase volatility as stocks approach certain points, almost ensuring that we have a repeat in selected issues of May 6th.

This really should have been a much shorter trading halt (say 15 seconds?) so that automated liquidity providers are still encouraged to bid without the risk of having to hold a position for what they consider to be a long time, and it should have been built on gaps in the order book rather than trade prices, so that there was no need to see a trade go by which is either determined erroneous and perpetuates the condition (waiting on someone to make a non-erroneous trade at >10% movement causing a halt), or is allowed to go through painting the tape during a halt at a lower level than a fair market would dictate.

I know most people are up in arms about May 6th, but this isn't the kind of stuff that should be rushed through. Fortunately, it is only on 500 issues, and will probably not do too much damage before a better system can be agreed to. It is discouraging however that after the SEC has done quite a reasonable job understanding what transpired on May 6th, that they couldn't extrapolate how their new rules might further exacerbate volatility, especially when they are so close to getting it right, and after literally years of debate about rules on upticks in which the SEC understood that new rules should be built off of the consolidated bid price rather than last trade price.


Kid Dynamite said...

well, peter, i definitely disagree with you on this line of thinking:

"then someone can engineer a halt by placing a bid for 1 share at 4.9% off of the last trading price"

buyers don't engineer halts - sellers do... and buyers can't insure that their 4.9% discount bid is the only bid out there... if it is, they haven't "engineered" anything... the real risk is if SELLERS try to engineer halts by triggering them

Anonymous said...

KD - totally agree. I meant that a seller would engineer the drop of the additional 4.9% by buying a small lot for nominal cost.

So, if you had a large set of PUTs or a short on a stock 9% off in the last 5 minutes and wanted to make another 4.9%, you wait for bids to dry up and put in a small bid at last trade price minus 4.9%, then you cause the price to drop more and have a halt.

So, it is the sellers who would do this, but they would effect the final drop through a small purchase targetted at the lowest price not subject to erroneous trade policies.