Quote from: picobit on May 08, 2013, 01:06:07 PM

2) When the price thus moves suddenly, margin calls may happen. But since the relevant side of the order book is empty, forces sell/buy cannot happen, and the "worst case scenario" will be invoked. If we could pre-fill the relevant part of the order book before the clearing, we could act as counterparts to the forced liquidations. The person being margin called would get a better price, and less people would see forced liquidations. This is where it becomes important that there is still a limitation on the orders we can place, for example of another 10%. That prevents me from placing buy orders at absurd prices like 0.1$ and hoping to profit unreasonably from someone to getting margin called.

I think this is the major cause of the problem. It seems that the exchange has basically handcuffed itself with the very rule that is supposed to protect them in the first place.

Since there is no recourse against individuals that go into technical default, and no background checks to determine if a trader is financially sound to use margin/derivatives, it comes down to how to self-regulate ourselves better than the CFTC could. We have to throw out some of the rules that normal clearinghouses use and use more creativity to protect both traders and the exchange.

There are really four variables that can be modified here:

Leverage ratio - the quickest most direct fix. The downside here is that I think most of us would actually like more gearing, rather than less. Going any lower than 2.5x would probably divert a lot of volume away. It can and should go higher, once the exchange is more stable, but now's probably not the time for it. However, once it's higher, it opens up the possibility of having a dynamically adjusted level based on market conditions.
Trading range - uncertain impact. With a lower range, the market locks quickly and could remain locked across multiple trading sessions, thus further stifling trading. With a higher range, the greater the chance positions get wiped away when someone can't get to their computer and refill their account.
Margin call condition - effective and somewhat unobtrusive. Increasing the percentage decreases the probability of liquidations, but at the expense of traders having to refill more frequently.
Frequency of clearing - in practice, quite impactful, but too obtrusive. With a higher frequency (say 3x day), you can run into situations where two trading sessions pass while someone's asleep. And higher than that you brush up against the natural delay in confirming BTC transactions.

The way I see it is, the best option is changing the point at which maintenance margin has to be refilled. Something like: issue a call for maintenance replenishment at 50%, giving the trader one trading session (12 hrs) to top back up to 100%. If the call was not obliged, then force liquidate. Within the one trading session, if the market continues to move quickly, margin call at < 25%. So you basically wind up with an added buffer where traders can top up margin without giving them the implicit of option of running away, and without having to offset their trades in a locked market.

In addition, I would like to see the trading range move to something that's algorithmically adjusted. Rather than a fixed number, it should be something that adjusts relative to market conditions and the current risk present in the exchange's book. For example, if the Average True Range over the past n hours in the spot market is > x BTC, or if y% of trades in the book have < z% maintenance margin, decrease the range by a function of these amounts. And vice versa. The trick here is that you want to be protective, but also not drying up trading completely. Which leads me to...

To picobit's point about pre-filling, I'm having a hard time determining why this would be a bad idea. Obviously you can't have market orders being crossed outside the range, but at least there is the possibility to get a more realistic quote during liquidation. It seems that "pre-filling" already occurs indirectly by not clearing the book after range adjustments, and having old out-of-range trades sitting there. Just looking at the current market depth for BUM3, you have ~75% of the asks outside the range vs. ~5% of the bids outside the range. I think you either have to allow limit orders outside the range at all times, or none of the time.

There is also a fifth variable - the length of the settlement period. But, it would probably take a long sampling period to determine what works best.

Just my two cents. I've only been trading for 2 weeks and only once has this scenario come into play while I've been on the site. I'm sure it's happened many times before, but unfortunately I missed the discussion in the past. And I'm sure these topics have been debated plenty of times as well, so my apologies in advanced.