Well if you re-read the original post, you'll see that I did not run the simulation with the elastic supply enabled, but with the emission algorithm as defined by Satoshi
So in fact this test was a very close simulation of attempting Bitcoin stability as the supply is not modified in any way other than diverting some of it to fund the buffer, and the overall emission follows Bitcoin exactly.
well if you are not creating the supply when emulating the bitcoin historic data. then your results would fail because the historic data would be void once you start playing with orders (butterfly effect)
you cant change orders and think that the future demand/supply and orders would still exist. so im still wondering why you would think that bitcoin would be stable based on your use of irrelevant historic data.
anything after the first day of daytrading in 2011 becomes a new dimension and no longer part of historic data.
people who cant get their orders processed because your system cant find a counter order to match. would cause people to withdraw funds(affecting your buffer) or they would change their order price(affecting history and demand). which as i say makes history change. and no longer relevant to use any further data from the next days data, as it would be invalid and not the same.
or more concisely put
You can easily create back-dated models for trading that work perfectly on history but will fail after they work live so it isn't a valid test.
(anyone who does automated trading knows this)
EG.
if you changed the winter 2013 spike of $1000 so that it stayed around the $100 level. lots of people would withdraw from your system(buffer supply dries up) to sell coins privately on other exchanges. which would mean that your exchange is then lacking supply and stuck at a false price of $100 with no resistance or no demand to sell.. what would then happen is lots of dollar holders throwing in more dollars to buy your limite supply of cheap coins that are left(by those stupid enough to not withdraw) thus they would lose out selling too cheap and the buyers would get cheap coins to then arbitrage until all supply buffer is gone.
leaving your system in a totally different dimension of value than other exchanges.
the only way your system would work is if you were the only exchange and no one had any other means of exchanging away from your system.
and so i think your system would be more suited to the fiat controlled and centralised economy. and not so much to manipulate bitcoin at the expense of people who want a true freemarket
I didn't alter the sentiment of the trades though, the sentiment of every single one of those 58M trades is exactly as it was.
If that trade was posted because the trader wanted to cash out because the price was $1000, then that same sentiment is present in our test.
Therefore if the sentiment isn't changed, then there is no butterfly effect.
In fact I would argue that due to this, it is HARDER to stabilize the price, as we are applying rational sentiment against prices that now make it irrational, and irrational trades are MUCH harder to stabilize.
For example User A has a Bitcoin that was worth $1000, as of the pump at the start of 2014, a few days later it was at $800, a lot of people cashed out because the sentiment was panic.
In our stabilized model we have that same trade, but it is applied to a stabilized price with the same sentiment, panic. Even though in our stabilized test, there was no reason to panic because the price hadn't dropped 20% in a few days, perhaps only 0.5%. In the stabilized environment that user A probably wouldn't have sold, but in our test he still process the sell with the same sentiment, even though its now irrational.