How to predict market tops and bottoms using the COT report.

"Following the smart money is the smartest thing you can do."

This article will discuss in a practical way how these weekly Commitment of Traders reports can generate proper (and intuitive) trading/hedging signals on currency pairs and commodities. A relatively simple analysis that, when added to your toolkit, can optimize your entries and warn of imminent reversals. We will take an example of crude oil price.

1. What is the COT report? What are we going to do with it?

On a weekly basis, the CFTC releases a quantitative report that details the number of futures positions held by the major market participants. The report has a pretty solid history going back to 1980's. However, a decent COT analysis will require no more than 5-8 years worth of data, which is perfect for us. Our approach is extremely basic, we seek to profit from imbalances of supply and demand by observing the number of positions held by the different participants. More details are outlined below.

2. What is it composed of?

The report shows mainly three types of market players (among others which are less significant):

• Commercial participants: These include the hedgers, which naturally indicates that the number of positions held by them has a negative correlation to the price of the said asset.
• Non-commercial participants: Clearly, these include the speculators (hedge funds, mutual funds, and other big buy-side funds). Of course, a rational guess would be to assume a positive correlation with the asset as they mainly go with the trend (our guess is correct!). They are also called the funds positions.
• Non-reportable positions: These positions tend to be mixed and are under the position limits to be considered in one of the two above categories. We are not really interested in analyzing this element.

3. What do we need for the analysis?

Let's discuss hedgers (commercial participants) in more details. A hedger is an entity who wants to protect itself from adverse price risk. So, an airline company fearing a rise in the oil prices will go long on oil derivatives so that the operational losses are offset by the trading gains (if the asset does go up eventually). Another example of a hedger, is a wheat farmer who is suspecting that prices might go lower and thus hurting his/her profits. So, the farmer decides to go short a derivative contract on wheat so as to hedge this risk with the same rationale as used in the above example.

For each asset, we will have two sides, hedgers and speculators. And for each of these two types, we will have the long side and the short side. And this brings us to an important concept in the COT analysis, Netting.

As we have seen, the airline company is a consumer of the product (oil) and thus is going long to hedge it, making it a long hedger, while the farmer, being a producer of the product (wheat) will go short to hedge it, making him/her a short hedger. We are interested in the net hedgers positions, hence, the number of long hedgers minus the number of short hedgers.

So in our case, the net hedgers would be the long oil hedgers minus the short oil hedgers. The same concept applies to funds.

The data frame above shows a spreadsheet example of the four possible COT values that correspond to their respective asset (beginning of 2007 period).

The COT report, when downloaded, will give you the details of the positions (i.e: without calculating the net), meaning, we can subtract the shorts from the longs from each of the two categories to be left with two measures, Net hedgers and Net speculators. The next step is to chart these two time series against the price of the asset and we will be able to understand a great deal.

Below is the same data frame after adding two netted columns (Net funds and Net hedgers). We are now interested in the last 3 columns (NF, NH, and OIL) and will continue our analysis by focusing solely on them (although a more thorough approach would be to actually analyze all six of the COT values but this is out of scope for this article).

4. Application on the price of Crude Oil.

Applying some basic descriptive statistics on the two COT columns, we can calculate the upper bounds and lower bounds for the net funds and the net hedgers (2-3 standard deviations away from a rolling/fixed mean). It's up to the reader to choose the settings of the statistics as well as the look-back period.

Now comes the fun part, having a statistical extreme signals an imminent reversal or at least a correction, which when combined with proper technical entry techniques can enhance trading returns or help with risk management processes. The following chart shows crude oil prices (in black) vs the net funds positions (in white) and the net hedgers positions (in yellow).

At first glance, the positive correlation between the funds and the price of the asset is clear (white and black), as is the obvious negative correlation between the hedgers and the price of the asset (yellow and black).

The two grey lines at -20 and 20 on the left axis show approximately 2.3 standard deviation away from the historical mean of the COT values (20 for the funds, -20 for the hedgers).

It's important to state here, that the two lines suffer from look-ahead bias and therefore cannot be given credit for any peak/trough prediction before September 2018. That's why we have made the grey boxes which appear when a statistical extreme has been reached (with a rolling mean so as to not include too much non-stationary data from the past). The grey boxes do not suffer from any kind of bias and were perfectly noticeable at that period.

Notice how most of the time, the COT extremes (preferably both funds and hedgers at the same time) coincided with a reversal or a severe correction in the price of oil.

If we further optimize the barriers (x * standard deviation + rolling mean(t)) in the 2nd quarter of 2018, the 2.5 standard deviation is around 23-24 and therefore, around June/July, the COT values (both funds and hedgers) indicated a strong supply/demand imbalance and an extreme bullish sentiment on oil (therefore as contrarians, we seek to be on the opposite side of this). A strong sell signal was clear. Combined with a sound fundamental and technical opinion, one could have entered at approximately the correct time (around August/September) and would have benefited from the drastic drop in oil prices these few past weeks (-30% loss of value until the time of writing this article).

On another note, It's interesting to also state that this same analysis gave a strong buy signal on Gold in september (+5% gain in value until the time of writing this article).

Detailed terminologies used in this article:

• Speculator: An entity that trades to make a profit.
• Hedger: An entity that seeks to cover its price risk by engaging in financial contracts.
• Rolling Mean: A moving average of a certain window (20 days, 10 weeks, 6 months, ...)
• Standard Deviation: A measure of historical volatility.
• Barrier: A fixed or moving line that is a function of the rolling mean ± X . Standard Deviation. The barrier is the trigger that is used to find statistical extremes.

Conclusion

Of course, perfection is the rarest word in finance and this technique can sometimes give false signals as any strategy can. And while a smart user could use these false signals to confirm the prevailing trend and ride it (after all, the trend is your friend), it's difficult to determine whether the signal has failed or not, especially that on a weekly time frame, it will take time to be properly analyzed.

Sofien KAABAR.