=================================================================

THE FUTURES MARKET ECONOMIST

**Knowledge emerges from the interaction of individual minds**

=================================================================

The Futures Market Economist (TFME) is the electronic publication of the ongoing seminar in futures market economics. This seminar is a free voluntary association of individuals interested in the application of economic science towards the understanding of futures and futures options markets.

This seminar is presented weekly by posting on Mondays to the following USENET newsgroups:

misc.invest.futures misc.invest.technical sci.econ

This seminar is organized and directed by Vern Lyon, PhD. I can be reached at

tfme@atdot.org

BACK ISSUES: The current and back issues of TFME can be obtained at the following URL:http://www.aros.net/~vlyon/DISCLAIMER: This electronic seminar is for educational purposes only. Any use of information obtained from this seminar is not the responsibility of Vern Lyon, PhD. Use it strictly at your own risk.=================================================================Volume 1, No. 15, July 15, 1996ContentsThe Efficient Market Hypothesis, Part IV: Is there any Structure in the Sea of Noise?Delta HedgingThe TFME Toy Portfolio: Week 2, July 8 to July 12, 1996=================================================================The Efficient Market Hypothesis, Part IV: Is there any Structure in the Sea of Noise?Introduction and ReviewLast week I discussed the work of two highly regarded statisticians who on the basis of empirical studies concluded that financial markets appear to be fair games, a conjecture that was made in 1900 by Bachelier that was based on theoretical reasoning. The conclusions of Working, Kendall and many others who have followed them in doing empirical work on the EMH has been most damming to individuals who claim that all one needs to know is the history of past prices. Bernstein, among others, refers to these individuals as "chartists" or "technicians." In quoting a prominent technician, Bernstein says that "[t]here credo is ... that stock prices, which record the history of where transactions have actually taken place, are therefore 'sufficient in themselves' to reveal everything the profit-seeking investors needs to know [p. 99]."Chartists or technicians were not the only individuals that were upset with the nihilistic conclusions of the statisticians. Many economists, predictably, refused to accept the possibility that the "science" of economics was mere rhetoric. Kendall anticipated this response from the economists. The following quotation from Kendall indicates this; furthermore it summarizes what he wrote, and in the passage that I have emphasized is an important point that will we discussed in some detail in the future.The series looks like a "wandering" one, almost as if once a week the Demon of Chance drew a random number from a symmetrical population of fixed dispersion and added it to the current price to determine the next week's price. And this, we may recall, is not the behavior in some small backwater market. The data derive from the Chicago wheat market over a period of fifty years DURING WHICH AT LEAST TWO ATTEMPTS WERE MADE TO CORNER WHEAT, AND ONE MIGHT HAVE EXPECTED THE WILDEST IRREGULARITIES IN THE FIGURES. To the statistician there is some pleasure in the thought that the symmetrical distribution reared its graceful head undisturbed amid the uproar of the Chicago wheat-pit. The economist, I suspect, or at any rate the trade cyclist, will look for statistical snags before he is convinced of the absence of systematic movements. And he will be very right to do so [p. 87 in Cootner].One particular individual who took a more optimistic view was Harry Roberts of the University of Chicago. He thought "[i]t more likely that economic analysis could give predictive insight into stock-market behavior than that physical analysis could help with a real roulette." (Roberts made this claim in 1959 when economists were more highly regarded, and hence more confident in their abilities than is now the case).What is interesting is that Roberts clearly recognized that what appears completely random might appear that way because of our limited knowledge, rather than being an intrinsic property of the exchange process.This is fascinating given the fairly recent interest shown in nonlinear mathematics, specifically what is known popularly as "chaos theory." The interest in chaos theory is an example of how a development in technology, in this case, cheap computer power, can lead to further development in theoretical disciplines such as mathematics. This has also resulted in many researchers in fields not so pure as mathematics to apply chaos theory to their own particular fields. The way this have stimulated researchers can be illustrated by programming a computer (or hand calculator) to grind out a sequence of numbers generated by various simple nonlinear equations that exhibit chaotic behavior. For example the so-called "Tent Map," is defined as/ a[1 - P(t)] for P(t) >= .5 P(t+1) = \ ap(t) for P(t) < .5for certain values of a > 0 and 0 < P(0) < 1. This simple equation will generate a sequence of numbers that looks very much like a random walk. In fact many standard statistical tests for randomness conclude that a sequence of numbers generated by the tent map, and some other simple nonlinear equations, is random, yet each number is determined exactly by the number preceding it by the above formula. Furthermore if the exact initial value of P(t), P(0) is known precisely, and if our calculation of succeeding numbers is exact, (no computer round offs), then given the initial P, the entire history of the P's can be determined.Understandably this have lead economists, physicists, biologists, and all sorts of other researchers to question whether the phenomenon of particular interest to them that appears to be random is instead a deterministic process that follows, if they are lucky, a path described by a simple mathematical relationship.I mentioned last week how Doyne Farmer and the Eudaemonians, of Thomas Bass's book, Eudaemonic Pie, claimed to have found structure in the game of roulette, the outcomes of which to me anyhow appears to be purely random. This "success" has led Farmer, a physicist, and many economists to apply chaos theory in an attempt to unravel the behavior of the financial markets.However, given the complexity and "openness" of the exchange process, the question is not whether the seeming randomness is deterministic, but whether any of the variability is determined, i.e., is the movement of prices subject to both purely random forces, for example, the input of randomly arriving market relevant news, and also to systematic elements, for example, perhaps under certain conditions the market might overreact to the randomly arriving information.Micro and Macro Technical AnalysisI have recently concluded that many individuals who classify themselves as technicians are concerned with many other variables than price. Most are concerned with not only price but with volume and open interest; and others are concerned with the structure of the market in regard to large speculators, commercial hedgers, and small speculators, or various other subgroups of the markets participants.With regard to these other variables it is useful to distinguish whether they are macro or micro variables. A macro variable with respect to the exchange process is one that is at the level of the market. The significant ones of these are price, volume and open interest. Therefore individuals who use either one or all of these variables in their analysis can be considered to be employing macro-technical analysis. (This, of course, includes the traditional chartists referred to above).I have on several occasions in TFME mentioned the reason for the adoption of methodological individualism is that it is at the level of the individual where decisions are made, fear is felt, and greed in manifest. Therefore if there are any regularities they are more likely to be observed at the level of the individual. In fact the results of Working and Kendall cast doubt on the existence of regularities at the level of the market -- it's all just random noise. On the other hand there could be regularities at the level of the individual that under certain conditions can have a predictable effects on market prices.By incorporating micro variables into the analysis, where a micro variable is one that is either at the level of the individual trader, or a grouping of individuals based on their individual characteristics, we can appeal to the "forces" that operate on individuals, such as fear and greed. For example we might be concerned with the group of individuals who are vulnerable to news that is perceived to be inflationary. These would be individuals with long positions in the bond futures market and at the same time are using considerable leverage relative to their capitalization. The analyst who attempts to analyze the market by forming subgroups based on common underlying individual characteristics, while not practicing pure micro analysis, can for practical reasons be considered to be employing micro- technical analysis. I will come back to this point below.Also from the practical point of view, I consider individuals engaged in marketing research to be engaged in what can be considered micro technical analysis in that any businesses operating in a voluntary exchange environment must constantly seek out where the buyers and sellers are likely to be, and this is what individuals in the futures markets are attempting to do when they try to find support and resistance levels.However, with few exceptions that I am aware of, most technical analysis, at least formally, is done with market level data, such as prices, volume and open interest. An outstanding example of this are the simple rules that combine price action with changes in volume. Specifically these analysts claim that if a price rise in accompanied by an increase in volume then the price will continue to rise, or that conversely if a price decline is accompanied by a increase in volume then the price will continue to decline (trend extension).That this is not always the case is vividly illustrated by the action on Wall Street on Black Tuesday of October, 1987. On that day, as I recall, the Dow-Jones Industrial average decline by over 500 points, the largest one day decline on record. At the same time volume was over 600 million shares, possibly still a record. Any individual who had decided on the basis of the conjunction of a price decline and volume of such magnitude to go short, unless they were quick to get out, would have been greatly disappointed, as the market after initially going down on the next day stabilized and pretty much went up after that. Of course market technicians only claim the rules are based on the odds being favorable, i.e., if it is based on science, it is not a nomological law, only a probability law. However, when the price change and volume increase was so dramatic (a best case scenario) and it failed one has to question whether the odds were really favorable.The problem is that market level data can not distinguish between buyers and sellers because of the simple fact that for every buyer there is a seller. This means that market level analysis can not determine why individuals are buying (selling), and this makes a difference.To illustrate why this is so consider once again the situation where the long interest in bond futures is relatively vulnerable with respect to the short interest. Then suppose that an unexpected report on inflation is received by participants in the market. Let this report indicate an inflation rate of 3.4% rather than the expected inflation rate of 3.2%. Holders of bonds will then want a additional .2% yield to compensate them for holding bonds. This corresponds to a price reduction in the bond futures price. For illustration purposes assume this corresponds to a decline of 49/32 from the before news price of the bond futures. In a world where individuals were not risk averse the price of bond futures would efficiently move to a level of 49/32 below what the before news price was, say 105-17, to 104 even. However, if individuals are risk averse, and if a significant number of the risk averse individuals have relatively large exposure to risk in bond futures, i.e., our longs, then they could sell down to a point considerably below 104, perhaps as low as 103-12.It might appear that these distressed sellers are being irrational in selling at a price below 104, but the 20/32 they are willing to give up in order to reduce their risk exposure can be considered a risk premium.On the other hand if there were not a significant number of vulnerable longs, it is unlikely that the price would decline much below the 104 level. Furthermore, the case where the fleeing longs reduced the price to 103-12 will likely result in the market turning around and going back up to the 104 level. Of course as time goes by new unexpected will affect the behavior of the market participants and hence the overall market. Some of this new news could also be bearish for the bond market and so the individual who tried to profit from buying at the 103-12 level in expectation of a rise to the 104 level might be disappointed, but, on the other hand, it could be bullish news. The work of Bachelier, Working, and Kendall suggests the probability of good or bad news is equally likely. This would imply that buying in the vicinity of 103-12 would be not where the odds are favorable, but where in dollar terms the game is more than fair. I will have more to say about this in the future.The above analysis was done in terms of groups rather than with individuals. This is because, unfortunately, there are formidable problems in doing market research at the individual level. This is because the required data on individual characteristics is just not available. Consequently we must find a middle ground between analysis based on market level data and individual level data. Looking at various groups of market participants is the obvious compromise solution to the low information content of market level data.These groupings, however, cannot be arbitrary. They must be based on common characteristics. It is important to attempt to do this as only when a group has common individual characteristics, e.g., all short and highly leveraged, can appeal be made to the "lawlike" response that the individuals will make in the event of bad news, and then to reason that the groups common behavior will have an effect on the market.Since myself and other writers on futures markets consider the characteristic of being relatively vulnerable a useful category, the question arises as to who fits into this category? Small speculators, hedgers, etc., or are these traditional categories ambiguous because at times speculators hedge (spread), and hedgers speculate, let alone the differences among all participants in the futures markets with respect to their risk exposure management skills. This suggests that all members who participate in the futures markets, irregardless what they are called for tax or textbook reasons, can at times be relatively vulnerable and at other time relatively safe. This makes forming groups based on individual characteristics far easier said than done. The two groups -- vulnerable and less vulnerable, for example, are likely to contain mixes of significantly different individuals from day to day, if not from hour to hour. Therefore without privileged information that could possibly reveal such things as the distribution of risk exposure in the market, one must make do with making "informed" inferences, perhaps guided by a combination of theoretical understanding and market experience.Understanding the role of fear and greed in the markets is an example of how theoretical understanding might be useful. In the TFME I have several time suggested that fear and greed are constants of human nature and that fear and greed always have and always will have a major role in financial markets.Risk aversion and profit maximization are the economists formalization of these notions of fear and greed. Profit maximization is discussed in undergraduate economics, but risk aversion must wait until uncertainty is introduced into the study of economics. This is often not done until the student takes advanced classes in economics; however, the example of Bernoulli and the St. Petersburg game indicated a long history of interest by individuals in studying the implications of uncertainty on economic behavior.The ideas of risk preference and risk premiums, however, can be understood without advanced training in economics. It is not even necessary to use the St. Petersburg game to illustrate the idea. A simple coin flip game does just as well. Specifically go to your favorite bar and find out how many individuals will take a bet where you require them to pay $500 to play and where they have a 50% chance of winning either 0 or $1000. The mathematical expectation of this game is $500. If an individual is willing to forego $500 for the chance of winning, with equal probability, either 0 or $1000, they are risk neutral. On the other hand if an individual refused to play unless their entry fee was reduced to something significantly less than $500, say $350, then they are risk averse.Except perhaps except for the very drunk most individuals offered the bet would refuse it unless given an entry fee less than the mathematical expectation of the game. The difference between the mathematical expectation and the amount they are willing to pay is the risk premium. In the above example this would be $500, the expected gain, minus the $350 they pay to play the game, to give a risk premium of $150. I will have must more to say about risk premiums in the future, but for now let me just say that one of the reasons that wealth tends to flow from the vulnerable to the less vulnerable is because the more vulnerable, in general, pay higher risk premiums, and they pay them to the less vulnerable.Finally to conclude this weeks discussion of material relevant to the EMH the introduction of risk aversion and risk premiums can under certain conditions result in the exchange not being a fair game in the mathematical expectation sense.=================================================================Delta HedgingThe main risk exposure management techniques I am employing for the TFME Toy portfolio are adequate capitalization, diversification, and options. Additionally stop-loss orders will be used on a selective basis.The portfolio at this time consists of 11 different modules corresponding to 11 different markets. Each module consists of various combinations of futures and option positions. At this time I have all short options positions. Furthermore I like these to be well in the money options. When that is the case they act somewhat like futures positions (the more in the money they are the more like a futures they become) with the added feature that they offer the possibility of capturing some return from what is called the time decay of the options premium. There is a cost to this in that short options do not offer as much protection as do long options, but the buying of long options, which I do not rule out, requires the payment of time decay premium, that can be substantial.However, at this time I do not wish to discuss option theory. There are zillions of books out there that do a good job of explaining the basics of options theory that interested readers can refer to. In this article I only what to discuss a certain property of options, the delta, and how it relates to hedging in the TFME Toy portfolio.In standard option theory the formula for an option is a function of 5 variables: (1) the price of the underlying instrument; (2) the short-term interest rate; (3) the time to maturity; (4) the strike price; and (5) the volatility of the underlying instrument.The derivative of the formula with respect to the price of the underlying instrument is called the delta. The delta is also a function of the same 5 variables. This means that when one of these other variable changes, not only does the theoretical value of the option change but the delta also changes. This can be a problem in using options to hedge with as the change in the delta of an option can significantly affect its hedging ability. In the TFME portfolio this problem is compensated in part by using diversification and adequate capitalization. Nonetheless the delta of a option must be constantly monitored to judge the effectiveness of the hedging ability of the option.One can compute all kinds of first, second, and crossed partial derivatives of the standard Black-Scholes option formula. Some of these are important and others are mere curiosities. The second derivative of the option or the first derivative of the delta with respect to the price of the underlying instrument gives an idea as to how the value of delta changes with a change in the price of the underlying instrument. For a call option this derivative tells us that as an option gets closer to or further into the money its delta increases. Conversely as the price of the underling instrument become less in the money or further away from the money its delta decreases.Deltas for call options are either expressed as a number between 0 and 1 or sometime by a number between 0 and 100 percent. A short option position can be considered to have a delta equal to the delta of the call multiplied by -1 to indicate it is a short position. For put options just the opposite is the case, with the delta of a put being between -1 and 0, and one in a short position being between 0 and 1. Finally the delta of a future position is either +1 for a long position or -1 for a short position. Forming a simple linear combination of the deltas for the constituents of a module gives the delta of the module.The module delta indicates, roughly, the number of futures positions long, when it is positive, or short when it is negative. A module delta less than 1/3 in magnitude I consider a mildly bullish (+) or mildly bearish (-) position for the module. A module delta between 1/3 and 2/3 is similarly either moderately bullish or bearish. Furthermore one between 2/3 and 1 is either bullish or bearish. One whose magnitude is greater than 1 is excessive and needs careful watching, possibly even taking immediate action to reduce the exposure. However, these values are relative to the overall size of the portfolio. If the initial capitalization of the TFME portfolio was 1 million dollars rather than 100,000, then the acceptable module delta would also be increased by a factor of 10. Also, I might add the effectiveness of the diversification is a consideration. For example since it is usual for foreign currencies to move in parallel against the US dollar, a negative module delta for D-Marks combined with a positive module for Yen is treated differently than if they both had the same sign.===============================================================The TFME Toy Portfolio: Week 2, July 8 to July 12, 1996It was a crazy week. Added to the craziness was a mistake I made with respect to an order that turned out to be very costly. I put in an order to buy when I wanted to sell Sep Coffee. So be it, there are all kinds of mistakes that one can make in the futures markets, rather than just getting the direction wrong, but they all have the same thing in common -- you must pay for them.At the beginning of the week to portfolio was as follows:Prior to the markets opening the TFME portfolio was as follows, where the indicates the settlement prices from the previous Friday:106.13 Short 2 Sep US Bonds 1.6 Short 4 Sep Bonds 110 puts delta=-.81 module delta=1.246571 Long 2 Sep D-Mark 136 Short 2 Aug D-Mark 6450 Calls delta=.8 module delta=.49124 Short 2 Sep Yen 290 Short 2 Aug Yen 9400 Puts delta=-.87 module delta=0.26727 Short 2 (10000 bu) Nov Beans 68 Short 4 (20000 bu) Nov Beans 775 Puts delta=.65 module delta=.66890 Long 2 Oct Cattle 380 Short 4 Oct Cattle 6600 Calls delta=.71 module delta=-0.848780 Short 3 Sep Copper 1115 Short 4 Sep Copper 9400 Puts delta=.58 module delta=-0.68385.1 Long 4 Oct Gold 15.5 Short 4 Oct Gold 370 Calls delta=.89 module delta= 0.442038 Long 2 Sep Crude Oil 200 Short 2 Sep Crude Oil 1850 Calls delta= .86 module delta= 0.282794 Long 2 Sep Natural Gas 344 Short 2 Sep Nat Gas 2500 call delta=.79 module delta=0.42116.50 Long 2 Sep Coffee 6.95 Short 4 Sep Coffee 115 Calls delta=.56 module delta=-.247284 Long 3 Dec Cotton 378 Short 4 Dec Cotton 7200 Calls Delta=.55 module delta=.8The bond module looked a bit overexposed. Some of the other positions I decided to modify. The following trades were made:Monday Sold 1 Sep Natural Gas at 2.8265 close out trade 1 comm Sold 1 Dec Cotton at 7335 close out trade 1 comm Sold 1 Sep US Bond at 106-20 opening trade Gross change in equity +2405 Tuesday Bought 1 Sep Natural Gas at 2.790 opening trade Bought 1 Sep Coffee at 116 opening trade Gross change in equity +1076.25 Wednesday Sold 1 Oct Gold at 386.50 close out trade 1 comm Sold 1 Sep Crude Oil at 2080 close out trade 1 comm Sold 1 Sep US Bond at 107-14 opening trade Gross change in equity +1681.25 Thursday Sold 1 Sep D-Mark at 6590 close out trade 1 comm Sold 1 (5000 bu) Nov Beans at 770 opening trade Bought 1 Sep Copper at 8870 close out trade 1 comm Bought 1 Sep Coffee at 115.90 opening trade. Mistake, I wanted to sell. Gross change in equity -2125 Friday No trades. i tried to sell 2 Sep Coffee at 116.2, but was unable. Gross change in equity -4857.50 P/L Equity Balance at end of week 1 Gross +10422.50 +422.50 3 rt commissions at 75 each Net +10197.50 +197.50 Equity Balance at end of week 2 Gross + 98602 -1397.50 6 rt commissions at 75 each (9 total) Net + 97927 -2073.00The week started well but ended poorly. My mistake in order placement as of now cost 1481.25 plus commission. --------- The beginning of week 3 has the portfolio looking as follows:108.11 Short 4 Sep US Bonds 2.28 Short 4 Sep Bonds 110 puts delta=-.66 module delta=-1.366589 Long 1 Sep D-Mark 147 Short 2 Aug D-Mark 6450 Calls delta=.86 module delta=-0.729105 Short 2 Sep Yen 306 Short 2 Aug Yen 9400 Puts delta=-.88 module delta= -.24809 Short 3 Nov Beans 32 Short 4 Nov Beans 775 Puts delta=.65 module delta=-1.66962.5 Long 2 Oct Cattle 422.5 Short 4 Oct Cattle 6600 Calls delta=.77 module delta=-1.088840 Short 2 Sep Copper 870 Short 4 Sep Copper 9400 Puts delta=.63 module delta=.52387 Long 3 Oct Gold 17.3 Short 4 Oct Gold 370 Calls delta=.91 module delta=-0.642120 Long 1 Sep Crude Oil 276 Short 2 Sep Crude Oil 1850 Calls delta= .92 module delta=-0.842.773 Long 2 Sep Natural Gas 3.19 Short 2 Sep Nat Gas 2500 call delta=.79 module delta=.42111.95 Long 4 Sep Coffee 3.66 Short 4 Sep Coffee 115 Calls delta= .43 module delta=2.287418 Long 2 Dec Cotton 460 Short 4 Dec Cotton 7200 Calls delta=.61 module delta=-0.44The serious problem is with the coffee. This is way to much exposure. I will sell 1 on the open and another one at around 112. The beans are also a problem. Much depends on the weekend weather. The bonds require watching, but I doubt they will reach 110, the point where the puts are at the money.=================================================================Logon, learn, enjoy. Knowledge is too important to be left to the professors, or any other special interest group.