I have found the term "Straight-through processing" to be quite ambiguous in the way that it is commonly used. It gets used as a synonym for just the post-trade process where messages are send from the trading platform into a customers' RMS whereas - as you say - the term actually means more than that.

Virtually all of my current work is with that goal (logicscope.com). The group took a dirty problem - integrating dispirate banking systems - and wrote a cool abstraction that lets it work as a multiadapter. Plus support, configuration management, etc, etc. By accident, the system also turns out to be a not-entirely-elegant but much cheaper and easier-to-work-with replacement for MQ for many settings.

The idea of queueing and releasing a chain of transactions on the trading side of the transaction may be interesting but I can't see it yet. Are you referring to situations in which you do a single "trade" but with a complex series of legs? If so - this already exists in some organisations but is not widely supported by APIs or major GUI interfaces yet. Or are you thinking more of forming a position across a range of positions and then unleashing it? This could be useful, but I'm not sure who you'd target it at. The banks are likely to have internal software teams doing a mixed job of delivering this sort of fn already, and I'm not sure how you could base a startup on it. I expect there would be some hedge funds that could do this to come up with interesting situations - for example - setting up pricing queues for stocks floated on multiple exchanges and currencies and then doing automatic arbitrage based on this. Algorithmic trading isn't my speciality though, and I'm sure there are smart people doing this already :)

Hm, I said I found FX 'inspiring', but didn't say 'inspiring for a startup'. As I mentioned elsewhere my interest is more in P2P networks, and those elude classic funding strategies.

The STP I'm referring to relates to automatic roundtrips including request for quote > pricing engine > quote > order > confirmation > TMS but the real interesting part is when banks automatically back their orders with purchases from yet other banks, intra-bank. Exotic currencies add an additional dimension, as no immediate pricing for e.g. Mexican pesos in Norwegian kronors is available and the trade needs to be 'routed' through an intermediate currency, usually USD or EUR.

My view of the system didn't go beyond two banks at the time but in theory there could be a long chain. That way the banks would be forming a financial P2P network. In reality it's a pretty hierarchical network as not all banks are created equal, but it'd be interesting if you could come up with financial routing mechanisms that would find the cheapest route in a hypothetical more dense banking network and what you can learn from that for poor man's P2P networks.

I read this to be the use-case you're aiming to satisfy: (1) getting access to a range of exotics which are only available through certain providers in addition to (2) also getting the best rate available for all ccy pairs. I think this is currently provided by 'portals' like fxall and currenex (bloomberg now too I think) where majors competitively bid on quotes when they want to and the buyer buys that which best suits them.

There's a couple of problem with deliberate chaining of interbank trades in that if the market moves far you might not get your countertrade (the quote will be rejected on the trade attempt - it's more common for quotes to be non-guaranteed than for them to be guaranteed, presumably because it's easier to build and avoids the situation where hedge funds start gaming you); and - irrespective of that - prices tend to get naturally stale fairly quickly (every few seconds but it's usually inconsistent) which means that latency is a factor.

Aims (1) and (2) are correct. If there are more 'exotics' to go around and they become more easily accessbile (i.e. low transaction fees), wouldn't there be more competition on the currency market? Online retailers could quote prices in the currency of their choice and you'd get an instant quote in the currency of your choice out of that. As for the non-guaranteed quotes and the latency, I'm not sure whether someone has seriously tried to get their head wrapped around that, so there's room for innovative routing mechanisms. As for me it'd be fine to just do this as a thought experiment and then apply the results to poor man's P2P networks which are more accessible in the first place and see what comes out of that.

P.S. The portals (I worked for one of those) are a step in the right direction as their convenient UI and STP lowers the cost of a transaction in terms of human effort, but other than that add little value on top of what's offered by the banks. They are at most leaf nodes in the networks as they can't hold money even intermittently and the actual settlement still takes place through the banks. So they cannot open 'new routes' in the network.