I think you've nailed down the main points. I will only add two thoughts based on my experience with OTC trading:
- Volatility
- The flip side of inefficiency
- Slippage versus impact
The volatility on smaller, lesser known exchanges could be more intense due to greater inefficiencies. By volatility, I don't mean just day-to-day noise, but long-term dislocations and extreme swings in both price and fundamental drivers of price. Anyway, I presume inefficiencies are what you're after in the first place.
...which leads to bullet two: "markets can remain irrational for longer than you can remain solvent". Presuming you're entering an inefficient market because the long term expectancy from identifying inefficiencies is greater, there is also a darker side of inefficiency which may be hidden from plain view. The flip-side of inefficiency implies that price and expected value of price can diverge for quite some time resulting in large losses even if you are "right". But then again again, the market is always right.
Because implementing said strategy in an illiquid market could fundamentally change the expectation, I believe that is is important to think about the difference between slippage and impact. First, slippage does not equal impact. I define slippage as the realized transaction price in relation to the published transaction price. For example, if my market on close order was filled at \$100.1, when the the published closing price was \$100, my order experiences slippage of 10 bps (.1%).
On the other hand, market impact is an estimate for how an order changes both the short and long-term equilibrium price. As a result, impact cannot be directly observed. For example, if I had not submitted my market on close order, how would that have affected the market price? In all likelihood, the published closing price would have been less than \$100. But we cannot directly observe this, so we must infer it based on the observed instantaneous and residual relative (i.e., excess) drift in prices after such an order is placed and filled.
The net result of slippage and impact are the same: an implicit cost not included in direct transaction costs. But I think that untangling this conflation is important when dealing with smaller exchanges where smaller orders can drive a huge wedge between your expectancy in some backtest versus reality.
In practical-speak, I would think it extremely wise to implement said strategy with close regard to impact. In reality, we cannot change the fact that impact happens, but we can improve our price realizations by never attempting to remove liquidity from an already illiquid market.
