Project advice on commodities trading

Physical trading is all about spreads. Whether that is physical location spread between two commodities, price spreads through time, or price spreads between different grades/specs of the same commodity.

All trades are put together using at least one of these concepts and a lot of the time putting together more than one at a time. e.g ultra low sulfur diesel is trading at $2 /gallon in the gulf in the spot market, trading at $2.10 in for delivery in two weeks in Chicago, and trading at $2.20 /gallon in Chicago for delivery 1.5 months from now. A trader can buy spot, ship it up the pipe to Chicago (capturing the physical arb) and sell it in Chicago for the $2.10 (product takes roughly 2 weeks to get up there, give or take). Let's say shipping cost is $0.08 so profit is $0.02 /gallon. The other option though would be to buy in spot, ship to Chicago, and then store it for a certain amount and sell it at the $2.20 (capturing the storage arb as well as the location, this obviously depends on storage costs which one would be more profitable).

So having said that, a good project could be pulling in historical data for various commodities and building a model that can easily look at various spreads time series. Oil in the Permian basin vs. oil on the US Gulf coast, gasoline in New York Harbor vs. gasoline in Houston, front month corn vs. various deferred months, 13.0% protein hard red winter wheat vs. 12.0% protein hard red winter wheat, Kansas City hard red winter wheat vs. Chicago Soft Wheat, etc. Build a model that can look at all these spreads easily and efficiently, chart it, and then analyze what you see. Build some technical indicators into the chart such as RSI, MACD, Bollinger Bands, etc and see if you can predict a reversal of certain spreads. Are the spreads seasonal? What months do the spreads act a certain way? When the spread reaches a certain level how often does it snap back quickly? How long does it take? For the physical location arbs what is freight between the two locations? Was the arb open? What I find interesting in the physical space is that these spreads typically have tangible, fundamental factors that drive the spreads you will see. For example, gasoline in the Gulf vs. gasoline in NY Harbor typically trades close to a certain level due to the cost of shipping from Houston to NY. Corn and wheat hover around a certain level to one another due to the protein levels of each crop and their usefulness for feed.

Using that info you can then go do some research on the fundamentals at play during the times the spreads behaved in ways that caught your eye. By doing this, you might end up finding some fundamental factors that drive certain behaviors in the spreads and be able to predict certain moves.

Disclaimer: I don't know what your access to market data looks like so some of this might be hard to come by. The data here isn't as important as understanding why spreads are important, what drives them, and the model itself that analyzes them. The other data can always be added as it becomes available. The data I do know is available though is WTI and Brent crude, KC hard red winter, Chicago soft wheat, Minneapolis hard spring wheat, RBOB vs. other exchange traded gasoline, etc.