Diversification is an effective risk reduction tool and should include trading different strategies. Almost all trading strategies consist of rules that use price as an input, and almost every trading indicator relies solely on price in its calculation. That creates an opportunity to diversify if we can find a strategy that can generate signals without using price as an input. This would offer an extra degree of risk reduction and it should behave independently of other trading strategies that you use, with an equity curve that is mostly uncorrelated with the curves of other strategies. Rather than building suspense about a non-price based indicator, I’ll admit up front that the COT data for futures contracts can be used to develop a system that does work without looking at any price data.
Every week, the Commodity Futures Trading Commission (CFTC) provides the Commitment of Traders (COT) report which shows the sizes of positions held in various commodity contracts. COT data shows the number of contracts held by commercials, large speculators, and small speculators. Commercials are market participants that use futures for bona fide hedging activities. In the grain markets, as an example, farmers and food processors like Campbell’s Soup would be hedgers. Large speculators are traders with large positions, large enough to be covered by government reporting requirements and will include many hedge funds. They are speculating on the direction of price moves rather than hedging risks in the market. Small speculators include everyone else.
COT data measures sentiment, how people feel about the markets, which is usually done with surveys. However, COT data is different from other sentiment measures because with the COT data we are seeing exactly what market participants are doing with their money rather than reading about their opinions. It is quantifying whether participants feel bullish or bearish by measuring the size of the positions they are willing to hold.
There are many ways to interpret the COT data. Among the simplest interpretations is that commercials know the physical market for a commodity the best and are likely to be the long-term winners in the futures markets. This is an important concept because in the long-term, futures are basically a zero-sum game (before commissions and trading costs) with the winning trades offset by an equal amount of losing trades. It is a logical assumption that commercials will be long-term winners; otherwise they wouldn’t participate in the markets.
Small speculators are expected to be wrong at major turning points in any market. This is the idea that small traders represent the dumb money in the market with commercials being the smart money. Large speculators are harder to generalize since some will win and some will lose and they will probably be right as a group just about as often as they are wrong.
Solely from the definitions, we’d expect commercials to be right most often and we’ll use their positions to generate trading signals. This is a simplistic interpretation that we’ll put to the test.
The data is easy to find at the CFTC web site (https://www.cftc.gov/MarketReports/CommitmentsofTraders/index.htm) and can be downloaded for free. Unfortunately the data found in the COT report is presented in a confusing manner and can be difficult to present interpret. It is a text file containing only the number of contracts held by each group and the change in holdings from the previous week. Traders have become spoiled by charts and indicators they can add with a single click, while the COT report has data that will need to be processed manually. Spreadsheets need to created and maintained. Before starting all that work, it’s useful to know if there is any useful information in the data.
To place the data into context, we can start by applying a stochastics formula to the raw COT data. That is a very useful formula for traders to learn since it places a data point into a defined range relative to the recent past. It answers the question of whether a number is high or low in an objective way and removes subjective interpreting from the trading problem.
To see where the most recent COT positions relate to the recent past, data can be maintained with a simple spreadsheet. In Excel, the stochastics formula for a six month look back period would be written as =(B28 – Min(B28:B2)) / (Max(B28:B2) – Min(B28:B2)) assuming that the data is all kept in column B of your spreadsheet. The calculated value can then be shown as a percentage. In effect, this takes the raw data used in the CFTC report and converts it into an index that will always be within a range between 0 and 100. Readings of 0 would indicate that there is a lot more selling than at any time in the past 26 weeks and a reading of 100 would show very strong buying pressure.
Just as with price-based trading systems, there are an unlimited number of possible options that we can use with this non-price index to design buy and sell rules. As usual, we will start with logic and develop a quantified set of rules.
Small speculators will be wrong at major turning points, but the problem with trying to trade that idea is that they will be right until they are wrong. Thinking about the internet bubble in the stock market, individual traders were able to make a lot of money for months and actually years leading into the top. Trading against them would have led to losses until it finally resulted in a really big win. This is the standard pattern seen with sentiment data and it makes it difficult to find a profitable and reliable rule based on what small traders are doing.
Large speculators have a very mixed record. In theory, they should be right most of the time because that is how they got to be large speculators. Traders who have a good track record find it possible to increase the amount of money they trade, while large speculators with consistent losing records should, in theory, eventually become small speculators as losses wipe out trading capital. Since some large traders will win and some will lose at any given time, there is no obvious and easy way to apply rules to the COT data in trying to find a trading strategy.
Data related to the positions held by commercials offers the most logical reason that it should be useful. Commercials are the market participants with the best understanding of the long-term fundamentals of a market and they should be the ones buying when they think the value is low. If the commercials think prices are high, they can either sell or simply stop buying – if they need the commodity but futures are overvalued they could simply buy it in the spot market at what would be a reasonable price based on their market knowledge.
Selling is a less reliable indicator of how commercials view of the markets. Just like with insider selling in the stock market, sell decisions can be based on a variety of factors that aren’t related to valuation. Selling could be caused by a decline in business which means the commercial hedger would need a smaller hedge, or it could just be driven by a need for cash. Since there is no way to know why a sell takes place, we’ll focus on a long only system as a starting point.
Many price-based trading systems work better on physical commodities than they do on stock market indexes. COT data can be applied successfully to stock market indexes. This makes it a good strategy to add to a breakout system that can be traded with a basket of commodities. Emini contracts require low margins and are generally liquid markets. Small traders can use contracts on the NASDAQ 100, Dow Jones Industrial Average, Russell 2000, or the S&P 500, which is what this test will be on. Relatively high transaction costs of $35 per round turn will be deducted from the results.
We’ll buy when the commercials are extremely bullish. Using the stochastics based index, we will buy when the value goes above 90. This would indicate that buying is very powerful compared to what it’s been over the past six months. Selling will be done when the index falls back below 90. That is just an arbitrary sell rule being used to exit trades. A time stop, selling at a predefined number of weeks after opening a position, also works well.
This strategy has a success rate of 62 percent winning trades, with an average trade size of about $150. The system trades an average of 8 times a year, and a margin of about $15,000 would be required to trade all four stock market indexes. You wouldn’t normally be in all four positions at one time.
This is an excellent strategy to consider for a small trading account. With less than $15,000 traders can use only one or two of the indexes or they can day trade with a broker that allows very low margins. This strategy is profitable over a daily time frame exiting at the end of the day on the day the buy signal is taken. That is not the ideal scenario, but it would allow a small trader starting with $1,000 or less to introduce diversification into their trading while building a larger account balance.
By Michael J. Carr, CMT