Dear All;
I am Physical Petrochemicals Trading Analyst (Quant) in the MENA region. As you can imagine, I am quite behind real-world traders when it comes to the "actual trading", my experience lies in portfolio management which is pretty math-intensive.
My current aim is to be able to measure the risk/return for a specific trade deal between parties A and B (buy from A sell to B). I am able to get all risk and financial data I assumed relevant (pricing from IHS, costs from historical data at firm, credit risk, counterparty default risk, tax risk, NOC fees, Regulatory risk, etc) as well as a few indices I thought supercedes all transactions in the market (Market Risk Index, political risk, country operational risk, etc.)
My problem is finding literature on an actual MODEL. Or a case study, explaining how Glencore-Trafigura-Vitol actually takes this numbers and make a decision. How do these indices and other risk numbers tie into the analysis for a trade? What is known is easier, the prices are set, no forward curves or market shocks etc, is a geographical arbitrage opportunity and we are not taking a position. But how do these others risks get accounted for? And tips on where to look would be appreciated.
Glencore, Traffi, and Vitol aren’t punting around on the screen. They are using originated deals to collect edge then use their supply chain and logistics network to monetize the day 1 gain. For example, you are Mid East /East African small time oil producer. You have zero credit, political risk, questionable labor and environmental standards etc. Shell, BP, and Exxon aren’t calling you up. But Vitol shows up and looks at cost of transport then compares to relevant index and pays you below fair value but above marginal cost. The traders at Vitol sell local index to hedge purchase and buys WTI or Brent to lock in the spread. Vitol charters the boat and moves to delivery. At that point the trader lifts the hedges and the physical is liquidated in the spot. If you are working at a bank go find your marketer or originator and ask how you can get a piece of the flow book. If you are at prop shop good luck...
How much stuff is modelled out probably depends on derivative liquidity. Nat gas and power are going to have a lot more modelling than say ferrous scrap or dairy trading (big volatile markets that are still basically the wild west).
It's not true that the big guys don't trade screen, there will always be a guy running a speculative paper book. In fact some people argue that the physical book is only there to give info to the spec one.
Nonetheless, in physical, decision-making usually follows a pretty simple line of reasoning. If you are starting a book, you might be doing back-to-backs where you start from a short and find a supplier for it, or a long and find a home for it. You make sure things make sense from a freight, credit, quality, payment terms, holding period and storage, financing, hedging perspective etc.. and throw on the margin you want to make (enough that some things can go wrong for instance).
As you build your book, your shorts and long become more independent. In a large book, you need to be booking good longs and good shorts. These will meet all the criteria above and then you match longs an shorts based on what your best freights are, or more precisely compared to your average freight.
So you build good longs, good shorts, and you execute in the way that makes the most sense. It's not rocket science, but it's not as easy as it sounds either.
Was gonna say, being a quant analyst in petrochems seems like a bit of a contradiction in terms, but your explanation makes sense. Generally speaking the way the trade houses model this stuff is extremely simple (as good bread says the “fundamental equation” is basically just to glue transportation costs to quality differentials and storage plus other miscellaneous costs and then make the deal if you can get the price at a level that shows you making money, and don’t if you can’t, and maybe comparing two such modes of your goal is to look at geographical arbitrage), though figuring out the specific costs involved without a strong background in the relevant markets can be really tough (if your company is asking they’re quants to do it instead of the guys with experience moving this stuff then this might already be unideal, but as a start looking into how similar quality commodities have priced relative to indices, and getting a handle on freight costs possibly by talking to brokers will get you a lot of what you need). I’d go a little bit further and say that even with liquid forward curves and derivatives, lots of modeling is still not required as long as the physical deals youre valuing don’t have a lot of inherent optionality (Often the case for power stuff with heat rate optionality, and also for natural gas stuff where storage has to be treated like a time spread option and not just an operational cost, but not for oil even with the large array of relatively liquid derivatives and forwards curves), though especially with the forward curves it’s worth being careful even in the vanilla models as a lot of the value can come from the fact that the trip is long enough to buy in one month and sell in the next (though this is tricky for petrochemicals I’ve heard stuff about proxying a forward curve with a related oil curve/curves).
I mean if you really wanted to you could throw all of that stuff into some sort of credit model/cva type scheme and have it spit out a slightly different number for how you treat financing/credit costs and reserves (comparisons to other geographical routes could help you derive expected recovery rates). For instance, you could run a montecarlo on the whole book/company where you modeled survival/default probability curves using some of these factors so that you can get company wide risk that would say something about advantages of stuff like geographic diversity of customers and the advantages of right-way vs wrong-way risk and maybe charge traders based somewhat on how the deals impact the aggregate (though in this scenario you should really be modeling forward curves too as they will effect all your numbers and also possibly have more direct effects on the fair values of specific deals). People have definitely done this sort of thing, though especially for a company your size with such illiquid commodities and also maybe for physical commodities more generally it’s not clear to me that this should actually ever effect any commercial decision since the main problem for making a business more profitable is usually just how to find more longs/shorts (I.e. building relationships) and even if there were liquid enough cds/bond prices for you to base your model on and have some hope of hedging there are so many modeling assumptions and the things aren’t actually liquid enough that in most cases you’d just be adding noise to your pnl more than anything else (in the best case your models would generally agree with what the traders were already thinking and they’ll do what they were gonna do anyway, and in the worst case you’d just get laughed at and ignored if they were too different). Grains of salt maybe since I have a pretty jaded view on the value of these sorts of quant jobs/analytics in physical commodities, but generally the reasons why people do stuff like that are mainly political (e.g. you have to do something to justify your paycheck/make your boss happy) or compliance/credit related (e.g. you have to demonstrate that you have “processes” that take these things into account so you can get a better credit rating or to get auditors off your back). If none of these apply in your case then I’d scrap the idea entirely and focus on something else (for instance it sounds like just getting the simple model down in a form where you can easily apply it to evaluate new opportunities would probably add some value), and if they do, then the game ends up being to find a way to model this stuff so that you check all the boxes for reasonable enough assumptions and methodology while also showing little to no effect on how you should continue going about your business, possibly slightly in the direction that whoever you need to make happy wants it to go which will depend on your political situation (this is roughly my sense of what people mean when they praise certain quants in the field as being “more commercial” vs “too theoretical”).
Firstly, you are vastly overestimating the sophistication of the commodities industry (unfortunately).
It's as simple as buying something at the reference market price, costing it up to take it to destination, selling it at destination with a margin.
GoodBread covered all the costs to account for that I can think of, some of the risks you list don't come into the calculation (political risk, tax risk) others are embedded into the costs listed (financing costs - varies per country).
The thing that strikes me here (you are working for a producer) is the very reason trade houses exist is to sit between the producer and the end user and take on the many risks of both parties, allowing the producer and end user to focus on what they do best. The role of the trading house allows them to buy multiple origins/qualities from a wide range of regions and then arbitrage them to a variety of destinations while making a small margin.