A recent trip to Los Angeles provided a virtual laboratory for the complex mechanisms of crowds and groups and game theory with many lessons for the market researcher.

Part 1 ( of 4)

The congestion on the freeways in Southern California is a prime example of a congested complex system. When crowds of cars enter the freeway's limited space, the traffic slows and creeps forward, and sometimes almost stops. Unlike the freeways, the markets have no red lights on the on ramps to regulate new participants crowding into the order queues at new market highs. The number of orders is very high with billions of dollars on each side, so the price has no ability to move, and the price effect is a grinding creep, just like a congested freeway. Since the public and big buyers are there to buy the new highs and the great conditions reported by the news, the only way to get filled will be to take the offers. One can bid all day on a trend day and not get a fill. Micro structural theory says with 1 offer and bid, a market order will move the price up, and that is what we see in aggregation. A slow creeping up, up, up at market highs, often displaying trends up with little or no volatility and no pullbacks. No one can cut through the large orders to move the market. This condition appears at market highs, and is the opposite of the bottoms where there is little congestion and the price moves up and down, creating volatility. At tops. bears flash huge offers to try scare off the bull, but its really a slow speed stampede and nothing stops the grinding upward except for a change in sentiment which usually is quite close at these points.

Part 2 (of 4)

At the USC Trojans game this weekend one of the rules in football is that if your team makes 10 yards, they get another 4 downs, which gives your team another chance, and provides momentum. This rule is at the heart of the strategy for the game. In the markets a similar rule applies. The question is to figure out what the rule is as it changes from time to time without warning and no one tells you until after the fact how many points are needed for a first down. Say the market moves up or down big in a short play, like say the Fed move up last month or the 20 point down afternoon last week, or this afternoon's recovery for that matter. A big advance gives the players another 4 downs (4 days?) to try make some more new ground and generate some momentum. Like the Trojans this weekend, the last three years have not been very successful on capitalizing or getting their first downs, but there is some talk among the old hands that old things that haven't worked for years may see their cycle in the sun again. The line of scrimmage is often yesterday's close, but can be other yard lines as well, prior highs, prior swings, yesterday's hi/low. Today we close near the line of scrimmage. As in football as in the snap, every play in the market seems to begin with a short retreat back into your own territory. Unless one fumbles, or screws up and gets downed behind the line of scrimmage, the idea is to make forward progress. 10 yards, 10 points seems like a good rule for a first down in normal circumstances. This afternoon recovery qualifies our team for a first down.

At the USC Trojans game over 200,000 people there all were wearing Trojan Crimson colors, hats, shirts, jackets, painted bodies, painted faces. I mean all of them. It was kind of scary. I asked a friend if someone told them to wear those colors. No, there is no rule or any rule requiring this attire. It was a remarkable example of herding behavior of crowds. We saw the stampede action today in the market as well. Stampede up gap, stampede down, stampede back, close to place of start, like buffalo. fish and elephants. Many lessons can be learned about crowds, herds, and stampedes to be used in market operations.

Part 3 ( of 4)

The Getty museum was one of the most impressive art museums anywhere in the world, not necessarily because of its collection, but the presentation, the architecture, the location. The third level of the West Wing overlooking LA is the a truely memorable and impressive view not to be missed on a clear day. In the collection is a set of medieval illuminated music manuscripts. The old style notation is not the fixed note per note that we use now, but rather a notation showing the relative moves, up or down of the notes. This provided a reminder to musicians who knew all the tunes already anyway. This might provide a good market notation system to catalog the various ups and downs that make up the regular tunes the markets seems to play. Just as history does not repeat itself, it rhymes, the market does not repeat it self, but it does harmonize, engage in contrapuntal variations on well known themes. The rhythms often stay the same and are more limited than the melodies. The rhythm of markets can be as important as the melodies. Oddly European music has no true rhythm nor is it notated in standard notation even today. There is only a cryptic mention of "march", "shuffle" or "lively beat", "andante" or some unquantified mumbo. The time signature gives only a vague hint of the rhythm. A good musician knows all the rhythms and changes. A trader should also, but unlike common TA patterns, there is no language or model for market rhythms. At Disneyland riding the train, there was a Morse code being tapped out. Even Alan Greenspan learned the Morse Code as a kid for fun. That is a good example of coding various simple rhythms with specific patterns to meaning. The same type of code ought and could be used in the market classification. 

Part 4 (of 4)

An interesting lesson from LA is their peculiar measure of distance. When you ask them, "How far is the coliseum ?" They invariably answer, "About … minutes away", a measure of time not distance. At dinner, an astute market expert asked, what other x axes are possible other than linear time? A suggestion might be to use percentage or point change, rather than linear time. An obvious measure is ticks or other micro structural market time. The important measure of a trade really is the return not how long the trade lasts. 1% is a great trade in a day, but not for a year. The Y axis becomes fixed and linear and time is unstated. This is the opposite of linear time chart in which return is unstated. Interesting relationships might be uncovered. For example exponential moves in linear time become linear in ticks or return.

Related to time to destination is speed. Unlike time, speed is variable. Everyone in LA drives 80-90 mph plus. Driving at 60-70, everyone is passing you by. I am used to driving about 25. If the speed increases, say to 100 or 120 plus, you get places twice as fast, but the risk of crashing increases. Skill matters, but what about the other idiots on the road, or road and weather hazards, or the tunnel up ahead with a blind curve hiding the tractor trailer around the bend, and the three tractor trailers right behind you, defects in the roads, and other drivers that limit speed. It happened while I was there: a 15 trailer pile up in a tunnel.

That brings up the issue; is there a speed limit in the market for safe driving? Drive at high gear and make great returns, but what is the best speed, ie annual return, one can drive year after year without any accidents, and still get to the destination. The optimal speed should be able to be calculated. One experienced practitioner and theorist thought that 20% is a consistent speed to keep up year after year. That sound right to me. Much faster and accidents become more likely. How can this be quantified? Optimal f and Seattle Phil's formulas claim to measure this, but Moe's point that it is necessary to look at a full history and see the systemic risks as a measure too. The normal distribution assumes safe roads and that people obey the law. There are systemic risks and risks from other market participants tendency to herd, stampede and panic, or to drive too fast or drink,take stimulants, drive too long which lead to big pile ups. Or the Fed to raise the speed limit to 100 mph. Of course driving to slow is no good either.

The time in market is relevant as well. Sharpe measures, risk/drawdown measures, Sortino and the like measure these. Measures of speed and risk must change over time rather than be fixed. That is probably the real challenge. A fixed limit which results from looking at the entire history will either hold one back, or contribute to excessive risk depending on the market. Alex's comments Friday are an example of this problem. One will never trade if tomorrow is going to be Black Monday. Just as one changes speed for the road conditions, how does one quantify changing speed in the markets. How do you know there are hidden curves ahead, road hazards and if tomorrow is Black Monday? One suggestion is looking for anomalies like 9/18 and C-C-C start to appear. Chris suggested a few others. When the market creeps up in low and up 8-11 new highs or more in a row is a good bet that vol is about to sky rocket as everyone else is cranking up leverage with borrowed funds. I think the 1980-90's were different though. Other times leverage is forced on you when market Road Warrior is chasing you down and you have to jump to high gear in order to escape. Other times it is good to hit the pedal to the metal in high gear, then immediately back off. This is very hard to do, but a model would help. What is that model? Please help.





Speak your mind


Resources & Links