Spread Trading Performance Drivers
What are the main performance drivers of a crypto spread trading strategy?
Spread trading on crypto has a few differences compared to traditional financial markets. The large number of venues and pairs can considerably complicate things, especially when it comes to accurately calculating your roundtrip costs (think ETF vs spot on margin) and backtesting.
Here are some of the main performance drivers of a spread trading strategy, with a particular focus on crypto spreads:
Your Cost Basis
If you don't have the cost basis or economies of scale (think capital), things are going to be very tough. If you don't have the right cost basis, you'll end up picking the tail volume in a trade that's already about tail volume. Your capital utilisation will be very low, as will your returns. You still have to keep the capital tradeable, so your exposure is effectively the same without the return to compensate you.
Pair and Venue Selection
There's a lot of different combinations you could trade, but the reality is that your optimisation process will probably tell you that you want to trade somewhere around the 80th pctile+ of the spread (if you have the cost basis), otherwise you'll be looking to quote or snipe at 95th pctile and above. You should be getting compensated for providing liquidity, so you should be seeing higher potential returns on smaller or lower volume exchanges.
The amount of volume that actually trades on the pair and venue of your quote leg at these extreme levels will probably be very low. It may also not be inline with general market share of that venue, but is there anything you can do about that given a small sample and that you want to have a decent level of capital utilisation? Probably not.
This is in some ways similar to US equity spreads (think alike ETFs or ADR spreads) where you have a similar complication in terms of the level of internalisation that occurs. Each product has it's own subtlties and rules.
Capital utilisation and concentration
Because you're trading at extremes, most of the time (unless you have the cost basis) you're not going to have a position or be increasing one. Though you don't have a position, you may still have to be exposed to counterparty risk. There are several ways to increase your capital utilisation, but you'd probably be better off by concentrating your capital.
Your optimisation process
The next biggest driver (by a narrow margin) is going to be your optimisation strategy - this is basically how do you dynamically set your levels. This is a relatively straight forward optimisation of a sided bolly band strategy over different lookback windows. You can speed this up if you have the hardware, I personally got a close to linear speedup on the same hardware by parallelising at the most granular level that I could. Realtime optimisation is also very applicable and relatively easy to do.
What you'll find is that because everyone runs a similar optimisation process, those with the same cost basis end up with similar levels. This is why you'll see crowding at the same levels. As market volatility decreases, everyone tightens up their levels and concentrates their capital to max returns and utilisation. So when you then get a big volume driven move in those kinds of markets, the spread can fly out further than you'd typically expect because the expected profitability of quoting that far out was very low or zero.
Your legging strategy and other ingredients
If you get legged, the longer you leave it open the more it can go against you, but it can only go in your favour by a known and fixed amount. Almost the definition of a bidirectional bet. You typically are not getting compensated for leaving your hedge hanging out there. Your other ingredients can help (when to cross, when to retreat etc).
Best of luck and Stay Tuned!