Rob, I would like your opinion on the value of adding stocks such as VIRT (Virtu Financial) and FLOW.AS (Flow Traders) as a portfolio hedge. These stocks represent HFT market making firms and have strongly negative betas compared to stock indices. Surely, by MPT, they should have a hefty weight in a balanced diversified portfolio?
It's true there could be some regulatory risk. However, these firms are diversified global businesses across regions and asset classes. It is unlikely there would be new regulations would impact everything from US equities to EU bonds to FX all at once. Also, I'm sure with their army of PhD's, they would be able to come up with new strategies to adapt to new regulatory environments.
I wander why it didn't work, i.e. does it mean that moving risk\positions from the "closely-related" instruments into the ones that can actually have positions is just fundamentally wrong somehow? E.g. if we have 3 bonds, one of which has the highest theoretical position and the other two < 0.5 contracts, it would be bad for performance to merge positions of the other 2 bonds into the first one? (I mean, obviously the answer is yes, that's what the experiment showed but still..)
I'm not sure eithier. It could be that we make more money than you think from idiosnyncratic things that only affect a single market. It could be that the cost penalty was too aggressive. But in the end the whole thing was just far too complicated and unintuitive - and slow; frankly I was bored with the whole idea. GAT
Although I liked reading about your research and activities, while going through the blog article I started to lose the view of the objective: "what is it that he wants to achieve?" Either I lost track of the objective, or I simply couldn't understand what was happening.
Trying to solve the problem that: - diversification across instruments is the only free lunch with huge benefits to expected SR - it consumes capital, thus making it hard if you only have a small account GAT
Yes, the objective is still very important - being able to trade more contracts than one has money for is a very promising way to improve P&L.. I mentioned earlier that I already tried backtesting it in the most straight-forward way - just trade a far bigger system on a much larger capital but limit it with the exogenous risk overlay, e.g. having my daily risk target as $10,000 but set the maximum current 'normal' risk (calculated as regular markowitz portfolio risk of the current positions) as $5,000. This way the system will take full positions in the instruments with high forecasts (according to the settings of the larger system), but only the first 'lucky' ones will be able to grab the available risk-capital and these which started to have strong forecasts later will be denied positions completely. And it actually improved performance a lot compared to just trading the smaller system (maybe it gave ~75% performance of the larger system). The problem with this approach is that there's no strict control over which asset class grabs the available risk capital first, e.g. if the world equities started trending first, and there's a lot of them in the system, the current portfolio might become 100% long equities and even if e.g. bonds also start having strong forecasts but a bit later, they will be denied positions completely because all risk capital is already allocated and they will only get a spot when some equities stop trending. I.e. the prioritization criteria in this case is very arbitrary: "whoever started having strong forecast first". and it's like why simply being first deserves a higher capital allocation? Although, in the end(most extreme case) I think it will ultimately has to come down to it anyway, because imagine if the system is very large and currently there's a lot of diverse instruments having strong forecasts, so much so that if all of them get allocations 'fairly' none of them will be able to hold a single contract anyway, so we're still forced to allocate positions to only some of them either randomly or next best(at least deterministic) option - on the first come first served basis.. Maybe that isn't such a big problem as it seems? i.e. we sort of know that similar instruments from the same asset classes trend (and do other things) together and our systems do become temporarily skewed towards a single asset class anyway.. But this will also make the system more path-dependent..
Following a comment on my blog I spent quite a bit of time this morning trying to think of a good way of doing things along these lines of a more heuristic optimisation, but really struggled to come up with something that wasn't really complicated (I had something along the lines of allocating risk budget to asset classes depending on their forecasts, and then within those taking positions based on the highest forecasts until the risk budget was filled... really messy). GAT