STOPPING STOPS (November 24, 2015): The following news article appeared earlier this month:Here is a continuing blog on this ban, which oddly also applies to good-till-canceled (GTC) orders starting on February 26, 2016:
Stop orders still exist on many other exchanges in the United States and elsewhere, so I think it remains important to discuss why they are so dangerous even though they seem so safe. I will also give my opinion about why good-till-canceled orders are important and why I hope that several U.S. and other exchanges will change their plans to eliminate them.
In the financial markets, no one likes risk. Therefore, all kinds of methods have been designed in an attempt to minimize it. One method which sounds good is to place a stop-loss order after you have bought anything. If you have purchased a particular security at a price of 20, and you place a stop-loss order at 19, then you believe that you won't lose more than 5% on that particular trade. Your expectation is that once the price drops to 19, you will sell it and thus avoid potential future losses.
The reason why this is appealing is that it creates the internal illusion that you are controlling your risk. Consider the fate of two gamblers, each of whom decides to play roulette continuously from 8 a.m. to noon, then to take a one-hour break for lunch, and then to resume play from 1 p.m. to 5 p.m. Gambler #1 remains at the same roulette wheel all day, since he knows all of the people there and enjoys chatting with them as he is betting. Gambler #2 has a rule: whenever she is behind by more than fifty dollars at any given roulette wheel, she immediately moves to the next roulette wheel. Throughout the day, both make bets of equal sizes, both make the same kinds of bets, and both also make the same total number of bets. Which gambler is likely to have the best result at the end of the day?
If you are reasonably knowledgeable about probability, then you know that both gamblers have the same expected average outcome. They are making the same number of equal-sized and equal-odds bets, and therefore it doesn't matter which roulette wheel they choose or how often they switch from one to another. The average outcome can be precisely computed with this kind of information. The actual results can very tremendously from this average, so both gamblers are likely to end up with very different net results--but it will have nothing to do with the strategy which they have implemented. Because the results are entirely due to chance, there will be no difference in the long run. The gambler who continued to change from one wheel to another may tell herself that she was limiting her losses as a result of her method, but there was actually no such effect.
The same would be true in theory if the financial markets behaved continuously--in other words, if prices smoothly moved up or down at any given instant in time. One person may feel emotionally more comfortable closing out his position whenever he is losing 5% on any given trade, so he always places a stop-loss order 5% below his purchase price. However, if he immediately takes the money from that sale and purchases another security of equal volatility, then the long-term result will be exactly the same as if he had done nothing. At any given time, he is just as fully invested as the person who simply buys and holds on, and therefore whether he restricts his loss to a particular percentage or not, the combined effect will be the same. If there is a long losing streak for both kinds of participants, then one person will lose 5% on each of a dozen different trades while the other investor loses 60% on a single trade. Either way, the actual expected outcome is identical in both cases.
In the real world, there are other factors to be considered. Traders who frequently buy and sell will usually have to pay higher commissions than those who trade less frequently. In countries including the United States and Australia, but not in Canada, there are favorable long-term capital gains tax rates which are about half the short-term rates and which only apply to positions which have been open for more than a year (one year and one day or more in the U.S.; one year or more in Australia). Therefore, in the long run, the difference in strategy will be unfavorable to the person who frequently uses stop-loss orders and other methods which encourage frequent trading, because the commissions will usually be higher and the total taxes owed will also be greater.
What is much more serious than these problems is the fact that the financial markets don't behave continuously. Price movements often make sudden lurches both higher and lower. These are not necessarily due to any rational behavior, but because humans will emotionally crowd into a security which has suddenly announced positive news and will panic out of the stock market when there has been a recent sharp correction and everyone decides nearly simultaneously that they want to get out. The most common time for a sudden slump tends to be on Monday morning following a downtrend, because many people will brood in an atmosphere of media gloom and doom throughout the weekend and will often place market sell orders. Those who have placed stops will end up finding that their stop-loss orders combined with market sell orders will trigger a domino effect, thereby causing lower-priced stop-loss orders to be triggered and new market sell orders to occur from panicked holders who don't understand what is happening and prefer to sell first and ask questions later.
You may be thinking that this only occurs for the most illiquid and infrequently traded securities, but shortly after the open on Monday, August 24, 2015, it happened for hundreds of popular exchange-traded and many closed-end funds. Yesterday, I was looking at a chart of various sectors and discovered this behavior for First Trust Consumer Staples AlphaDEX Fund (FXG):
Some investors in FXG must have decided that it would be a good idea to place stop-loss orders. Some likely did so in the low 40s or the high 30s, while others placed orders closer to 35 or 30. On August 24, 2015, the price opened below 40 which caused some people to simply panic and place market sell orders, while meanwhile those who had placed stop-loss orders were already having their orders processed. In the rapid plunge, even those who had placed stops in the low 30s or high 20s ended up selling in some cases below 28. As happens with all rapid collapses, eventually the number of limit buy orders--many of which were good-till-canceled--finally overwhelmed the combined selling of market orders and stop-loss orders. Once that happened, the vacuum reversed and incredibly quickly FXG surged into the high 30s and then the low 40s once again. This was wonderful for those who bought it in the high 20s, but for those who were stopped out at 29 or 28, they had to decide whether to get back in at 41 or 42 and thus pay perhaps 40% more for this fund than they had just sold it for a short while earlier.
You might think that I intentionally chose FXG because it was the most extreme example of fund behavior that day, but actually I was just reviewing the recent behavior of various market sectors and this one caught my attention. FXG was not even on the lists of the biggest deviations from net asset value which were described in the following three links:
From the above links, you can see that some funds which are much more diversified, more liquid, with greater total assets, and with much higher daily volumes than the one I cited ended up experiencing even larger percentage fluctuations shortly after the open on August 24, 2015. Funds like iShares Select Dividend ETF (DVY) and Guggenheim S&P 500 Equal Weight ETF (RSP) were especially volatile early that morning, and are considered by most analysts to be among the most predictable and consistent long-term U.S. exchange-traded fund performers. The most liquid exchange-traded funds including Power Shares QQQ Trust Series 1 (QQQ) and iShares S&P 100 ETF (OEF) experienced notably less intraday volatility than the funds listed above, but even these suffered dramatic plunges following the open on August 24, 2015.
Of the three links above, the one at the bottom is perhaps the most useful because it lists those which had the largest percentage dislocations from net asset value. It should be made clear that the actual assets which constitute these exchange-traded funds had also fluctuated much more than they would on a normal trading day, but the real issue was how steeply many funds' discounts to net asset value soared and then collapsed within minutes and sometimes within seconds. If you were fully prepared in advance to take advantage of these kinds of mispricings, if you hired others to watch for you, and all of you were doing nothing but scanning multiple monitors with one eye while placing trades with the other, then you still would have had extreme difficulty in placing orders fast enough to be executed near the most favorable prices. Only those who had placed limit orders in advance would have received the most favorable fills.
The above behavior is the primary reason why exchanges are progressively banning the use of stop-loss orders. Panics and sharp corrections will periodically happen, and those who use these orders will get burned and will complain the most to their brokers and to the exchanges. With the introduction of the internet, stop-loss orders have become far more popular than they had been in pre-internet times. If only a few traders are using any strategy, stops or otherwise, then it will have less of an impact. If many are placing stop-loss orders and they are triggered in a cascade, then paradoxically the wish to limit losses ends up ensuring even larger losses.
There are other drawbacks with using stop-loss orders. Most traders place these near similar prices, generally close to round numbers to make them easier to remember. As a result, securities will often slump just enough to knock out these stops and will thereafter rebound rapidly. Those who were stopped out then have to decide whether to buy back their positions at significantly higher prices, or to buy something else instead which might then suffer a similar fate. In many markets, floor traders and others who pay for the information can actually see others' stops, so they know exactly when they can temporarily place some sell orders just to trigger a cascade of stops, thereafter buying back their positions at lower prices. This happens routinely with some securities and more choppily with others. In addition to this behavior occurring frequently with exchange-traded and closed-end funds, it happens with individual securities on various occasions. If a stock is about to double from 22 to 44, it will sometimes briefly slide below 20 to knock out sell stops which are often placed close to such a round number. I have seen instances in which there was a rapid price drop that was reversed within seconds, so that it lasted just long enough to knock out sell stops. It also happens the other way for securities which are popular with short sellers; there will be a temporary short squeeze where the price will rise just enough to knock out buy stops, cause a rapid cascade of buying to reach a price which will force out some short sellers who are using margin, and will then collapse back to the original price shortly thereafter. Several years ago there was a particularly spectacular instance of this with Volkswagen which is a highly liquid stock and which therefore wasn't expected to behave in such a manner by most short sellers at that time.
Good-till-canceled orders are most popular with highly disciplined investors who will construct ladders of GTC orders to gradually purchase a security into weakness whenever it unexpectedly plunges in price as I often recommend near the end of any extended bear market. Because they help to balance the market on both ends and to prevent extremes from becoming even more ridiculously extreme, it is puzzling why some exchanges are considering removing them. Perhaps some traders had placed orders and forgot about them, and then when they were filled several months later these folks ended up complaining that they hadn't intended for their orders to persist for so long. Possibly there will be a compromise where instead of being truly good-till-canceled they will be valid for six months or some other period of time as many brokers already do to control such orders. The biggest advantage of GTC orders is that they permit gradual rebalancing of portfolios to take advantage of the best bargains and to sell the trendiest holdings while acting in small steps. It isn't practical or logical for traders to have to re-enter dozens or hundreds of orders every single day. Since good-till-canceled buy orders placed well below the bid are the primary source of support to prevent even worse intraday plunges, and since GTC orders placed well above the ask price are an important source of selling to keep some equilibrium during a euphoric uptrend, the exchange should be encouraging greater use of good-till-canceled orders instead of planning to get rid of them. As bad as stop-loss orders are in accelerating the most irrational behavior, GTC orders used properly are among the best kinds of market balancing which exist. I am therefore campaigning to get the NYSE, Nasdaq, BATS, and other exchanges to change their plans to remove good-till-canceled orders which the NYSE wants to eliminate starting on February 26, 2016.
A useful corollary is that there is no magic bullet to reducing risk. Being diversified helps, but sometimes numerous assets will rise or fall together even if they may seem to be unrelated. Whether it makes sense to have a combination of historically overvalued assets is also questionable, since assets which are significantly above fair value tend to eventually become roughly equally undervalued. Buying lots of small quantities of undervalued securities, especially those which historically have the least correlation with each other, is probably the best long-term approach and is used by a number of long-term fund managers including Howard Marks. In the end, the only real protection against loss is having a lot of money in safe time deposits like bank accounts which are derided when they are paying one percent interest or less but which are the only true protection against market fluctuations. The main problem with gimmicks like stop-loss orders and various kinds of technical methods is that they can create the illusion of stability, safety, and loss aversion when it doesn't exist. This can psychologically encourage some traders to take dangerous risks which aren't appreciated until we have days like August 24, 2015. If you are taking on too much risk, then you may end up dumping assets which are underperforming, or not purchasing bargains which you know to be compelling, or otherwise becoming emotionally influenced in your trading decisions. One useful test is what I call the thousand-dollar test: if you only had a thousand dollars or less in any given security instead of whatever you have now, would you trade it differently? If so, then you are taking on too much risk no matter what you think your risk tolerance is. The level which would make you uncomfortable is entirely due to psychological factors, so it will vary considerably from one person to another. I once did an experiment where I asked people before playing Monopoly to give me twenty dollars for dinner; I gave each person five dollars back and pointed out that since they had 1500 play dollars in the game then these represented fifteen actual dollars. Even at a penny per dollar, players ended up making absurdly poor decisions such as not buying enough property or avoiding trades because they kept fretting about the "real money" they would be spending. Eventually I gave everyone back their original money because they became so emotional about the progress of the game.
Disclosure: Whenever they have appeared to be especially depressed, I have been buying the shares of funds which invest either in emerging-market assets or in the shares of commodity producers, since I believe these are among the two most undervalued sectors in a world where real estate and U.S. equity indices remain dangerously overvalued. As the extremely popular U.S. dollar stuns investors by suffering a bear market instead of continued gains as almost everyone is expecting, this will lead to a major upward revision in global inflationary expectations. From my largest to my smallest position, I currently own GDXJ, KOL, SIL, XME, HDGE, COPX, GDX, EWZ, RSX, REMX, GLDX, URA, FCG, IDX, GXG, VGPMX, ECH, VNM, BGEIX, NGE, RSXJ, PLTM, EPU, TUR, SILJ, SOIL, EPHE, and THD. I expect the S&P 500 to eventually lose roughly two thirds of its May 20, 2015 peak valuation of 2134.72, with its next bear-market bottom perhaps occurring within several months of October 2017. The Russell 2000 Index and its funds including IWM had only modestly surpassed their highs from the first week of March 2014 and are currently trading below those levels; small-cap U.S. equities typically lead the entire U.S. equity market lower as they have done in past decades including 1928-1929, 1972-1973, and 2007. Those who have "forgotten" or never learned the lessons of previous bear markets are doomed to repeat their mistakes.
