Notice this trend in the volumeshare chart of HBD trading created by 
Observation
In recent months, the internal market has eaten substantially into the overall share of volume compared to UpBit and Bittrex. This is primarily due to the operation of 
Prediction and Rationale
This trend will not just continue but expand massively. My expectation is that over the coming years, not only will the internal market be the biggest market for HBD, but also for Hive.
For anyone trading Hive or HBD, the internal market has one pretty massive advantage. There are no fees for trading on the internal market at all. On a traditional exchange, participants on both of a trade each pay a fee, often somewhere between 0.1% and 0.5% of the value of the exchange. Given that the fee is paid both by maker and taker, that can amount to up to 1% of the value of the trade.
Market makers and day-trading speculators can make many such trades every day. Those fees seem small, but over many trades they add up substantially. For many speculators, accounting for fees is often the difference between actually making a profit or a loss on a day to day basis. The most important factor to consider is that the lost value in fees compounds over time. If you can make 0.5% more in one trade, the next trade has 0.5% more available to make more profit.
Now compare the cost of a traditional exchange with the internal market where you pay no fees, but you only pay RC's to set up a limit order. My own market maker account does many such orders every day, and it's able to do all it needs with a trivial 25 HP delegation. RC's recover over time, so it will not need any more unless it starts doing much higher volume. That account is among the top 10 trading accounts on the internal market by volume. So the only meaningful cost for a trader on the internal market, is either their time if they trade manually, or the cost of running/programming a bot if they trade automatically.
Over the long run, those who do trading on the internal market are going to have an easier time making money than those who trade on traditional exchanges. This will strip Hive and particularly HBD from those centralized exchanges, until we are seeing most of the volume on internal exchanges, with the traditional exchanges mostly serving just as on and off ramps.
Why it didn't happen before
You may object that all the above has been true for many years, but over the lifetime of Steem/Hive, we have never seen the internal market become a big deal before. Why would this only happen now, when it could have happened before? Thus there must be some flaw in this thinking.
The Hive/HBD internal market suffered from a chicken and egg problem. All exchanges need some initial source to start becoming the exchange of choice for traders. Traditional centralized exchanges have the advantage that you can trade actual USD there, as well as Bitcoin. In other words, they are the entry point and exit point from the system, where people go to buy in or sell out of the Hive ecosystem. As a trader, be it speculator or market maker, you want to go where the liquidity is. It is not necessarily worth it to try and speculate on an exchange with no liquidity or volume, so you use whichever exchange has those. Since the internal market cannot serve as an on/off ramp, it was not historically able to bootstrap its own trading volume, to be an exchange worth using for traders and speculators.
For HBD, the internal market may already be the best place to trade. Liquidity is not limited to the order book you can see, if you try to buy or sell Hive there, you will see bots and other traders respond with more orders. The price of HBD is also typically staying closer to $1, more reliably on the internal market than centralized exchanges. You will notice that OTC exchange accounts such as 
Conclusion
Due to the bootstrapping effect of