Time to review the first larger IDO on Gnosis Protocol and to look into why the mechanisms give BOTH team and the average buyer better prices. (1/11)
I described the issues with using an AMM for an IDO already here: (2/11)
With an AMM IDO, if the initial price is lower than buying demand, all that matters is who comes first. People get very creative (spamming the network, trying to make transactions fail) to achieve that. The result is a single clever arbitrageur making $100k+ in minutes. (3/11)
This value could be given to the team that is selling and to "regular buyers" instead. Let's have a look at an IDO ($DMG) that sold $6.8m worth of tokens recently via Gnosis Protocol: (4/11)
The team placed sell orders at various price levels so that it would create a supply curve in aggregate. On GP (Gnosis Protocol), it is possible to time those orders upfront so everyone knows when they become valid. Users can also already place bids beforehand. (5/11)
At the moment when the sell orders became valid the order book looked like this: (6/11) (actually, on GP the order book is multidimensional, this is just a projection onto one dimension (DAI/DMG): i.e., buyers can buy with any token) 
The protocol now sets the "clearing price" where those curves overlap. So traders that had been willing to pay more still got the better price, while on the other hand the cheapest sell orders from the team also got a better price than they asked for as well. (7/11)
On Gnosis Protocol we call the difference between what a trader request "at least" and what they actually get "trader surplus". The whole protocol is designed to maximize the overall "trader surplus". (8/11)
Coming back to the bZx example: On GP, you cannot compete via gas prices or tricks how to get in first. Any order that makes it into a 5 minute batch window (or is scheduled for it already earlier) is treated equally and the only way to compete is price! (9/11)
In the DMG example >$290k in "trader surplus" have been generated for buyers AND the team. In another mechanisms they would likely have gone to a single arbitrageur. (10/11)
Finally, another visualization of the order book right before the sale and the first clearing auction of the sale and the corresponding Ethereum tx: (11/11) etherscan.io/tx/0x7ca90269a…
