efinancialnews and fixfortechies says that Participants expect growth of automated execution in equities and foreign exchange to tip over into other asset classes, and also points out that while the use of algorithms to automatically execute trades is thriving in equity and derivatives markets, fixed income trading has failed to follow suit.
It’s no surprise the article predicts that customers will utilise algorithms for government bonds before credit (my latest 2c on e- Credit in this waratah post), and that the fixed income algos move will only occur as a “next step” after the buyside conquer using their algos for foreign exchange.
But while the buyside use is clearly still some way off, the efn article quoted Apama to show that sellside are already using algos in their pricing and quoting .. “receiving requests for quotes, analysing all the streams of information and then calculating pricing on the fly is becoming more established within our customer base“
The mention of RFQs leapt out at me. Sticking to the existing RFQ model (i.e. “whats your price for this size in this specific security?”) surely means buyside never get to fully utilise algorithms for fixed income. If the buyside algo machine has to make its stock selection before doing price inquiry then surely it’s going to spend a fair amount of time spinning its wheels reqeusting and re-requesting executable prices, based on the indicatives it has used in working out that it is interested enough to request a firm price. It of course then also needs to send RFQ for every security that meets the criteria it is seeking (cpn, mty, drn, issuer, sector, …. and many more). So are the platforms and dealers really going to be able to deal with the number of “open” RFQs increasing exponentially just because an algo box has been plugged in?
To really add full value the customer-facing algos actually need an executable stream of pricing of all of the inventory that is in scope for their trading decisions – as is commonly available for FX and also the markets where algos are already used – so the machine can do its magic then request execution with a good degree of certainty of getting it. The thing with the fixed income space is that given the breadth of inventory in FI space, multiplied by the number of dealers to contribute, I can’t help thinking this will only really work if customer indicates the attributes of the inventory they are interested in (ah, so something similar to that proposed in my post about executable bid lists then) rather than having to send hundreds of RFQs for specific securities just to get a truly executable price comparison… Then the dealers can respond with anything in their inventory that matches the attributes and which they can price in a way that comes close to meeting customer expectations in terms of the aggressive bid/ask spread achievable today using RFQ.
So again I’ll conclude that new transaction types are required in order to take the market forward. But this isnt one of those flags in the sand for two years time. Surely the existing platforms could offer sellside customers an attribute search for inventory that comes back with executable pricing (so “just click to select which trade you wish to do”) is really not a huge step? They offer search functionality anyway, so it’s just down to sellside supporting the new order type behind the scenes – in order to [algorithmically] offer their true inventory out to customers.