Kind of a thought experiment for you folks:

Imagine a quant-trading firm in a developing country with a very small market. Small = low liquidity, little participation by institutions and retail, no options. It's more of an 'investment' kind of market than a 'trading' kind of market. This firm has no real competitors, and literally no one else using automated strategies (beyond stop loss etc.)

I imagine the firm would start off by capturing all the 'no-brainer' arbitrage opportunities, and then attempt to deploy more complex models. It wouldn't have to be ultra-fast or smart, it would just need to beat other human traders.

In my opinion the firm would be very limited by the actual size of the market - it doesn't matter how fast or clever it is if the opportunities are just not that frequent. The capacity / scalability of it's strategies would be too low to make financial sense (unless the firm was very lean). Further, I imagine it would very quickly start trading volumes which could move the market, at least for the lower turnover stocks.

I've always wondered if it was practical for me to just move to a market where I just wouldn't have to be spend millions on trading infra or quant talent.

What do you think? Are there examples you know of this being done?

[My first post here btw]

If you’re describing something realistic, where would you get the data from such a market to feed your models? Like in Nowhere, where I live now, its probably controlled by insiders.

If you’re describing something realistic, where would you get the data from such a market to feed your models? Like in Nowhere, where I live now, its probably controlled by insiders.

I actually do have access to some historical data and a real time tick feed for UAE markets (ADX, DFM), some South Asian markets, Jakarta (JKSE) and Egypt (EGX) - that's why I'm so tempted. You are absolutely right about the insiders though - I imagine we can trade against them somehow - although I wouldn't know where to start.

In the past I have traded some 'ready-future' (cash & carry) arbitrage in Pakistan (I doubt there's many places in the world where an individual can do that). It was decent money for me back in the day but I don't think I could sell that strat as a product today, even though the volumes/vol have picked up recently.

You could try with NN models to detect insiders and then use whatever microstructure models advise to do with that (avoid interacting? Increase spread?) Depends on the data I guess. I’m quite experienced in modelling all sorts of timeseries also with DNNs, but not specifically in trading, just for the record

An interesting technique to find manipulators^
Would be hard to find / tag data though, particularly from the outside. Would probably need some local traders with a really sharp memory to help tag the data.

And what’s your role in all that?

Well I'm still doing my research, and this thread is part of that journey.
1) So far we've registered a company and we have ingested a lot of historical data.
2) We're building the real-time connectivity to the PSX and the UAE exchanges. 
3) Building a backtesting framework, almost from scratch. I am still looking for some turn-key solutions which could suit our needs.

Another issue I should have mentioned earlier is things move VERY SLOWLY in these markets, simple email responses can take 10 days, etc. e.g. The administrative process of just obtaining the live feed to the PSX took 3 months! And we've been waiting for a PCM to give us access to their testing environment for 6 months!

Once we're in a position to start development of strategies I absolutely will need people with boots-on-the-ground experience, and some senior quant devs/traders who have worked in more mature markets.

So my role right now is to actually try and see for myself what opportunities there are, and I have invested a big chunk of my savings and time to this end. Alongside keeping my day job which is in real-estate development!

One thing I really don't understand, why make all this so public? 

"This is Wall Street Dr Burry. If you offer us free money we are gonna take it."

Haha well the way I see it, the priority is figuring out what we're doing rather than secrecy right now. And it ain't exactly free money. That said, it is entirely possible that someone comes along tomorrow and eats my lunch just for kicks - at least there's a degree of validation in that - but I don't think it'll be because of this forum post...

Anyway, ya got any ideas to share?

It's certainly not free money, there are lots of costs. 

Any ideas? Not really, I am a fan of reinforcement learning and hidden Markov models, I'll leave it at that. 

Pleasure to meet you Jason

Pleasure to meet you Jason

And you too. Hopefully EmptyPockets will be changed to FullPockets in time.