How I changed my mind on the decentralized casino thesis (as a go-to-market strategy)

It's not necessarily beautiful, but crypto history shows liquidity needs a bit of a kick in the beginning before you can migrate to an order book.

Credit: @unionepro on Dune (this Gains Network dashboard is a forked version of their work).

Among the “intellectual” class of crypto analysts, there’s always been a deep skepticism toward protocols that are structured as a casino.

The definition of a casino in this case is protocols where either the token holders or some other group (a pool of LPs holding e.g. SNX or DAI) take the other side of each trade.

After all, when you are trading it’s mostly a zero-sum game (net of fees). There’s a winner and a loser on each trade and LPing with any passive strategy is a dangerous game — if traders win or find an arbitrage, the entire protocol (or all its LPs) is rekt.

There have been various flavors of the decentralized casino thesis in crypto:

• The original version was Synthetix. SNX holders take the other side of trades, collateralize positions, and trading fees compensate for their risk.

• The modern variations of Synthetix are GMX and Gains Network. For GMX, this collateral pool is a bunch of assets and for Gains Network it is DAI.

• Another variation was the vAMM pioneered by Perpetual Protocol which has since changed its design due to toxic order flow.

• Drift Protocol is still on the vAMM path and attempting to fix issues by having an overlay of a maker-taker on top.

PJ @plumjooz

kinda agree with @gametheorizing here. current setup works as GMX is largely adopted by loss-making retail traders. In a hypothetical world where profitable power users adopt GMX, this chart would look vastly different imo.

Miles Deutscher @milesdeutscher

6. Oracle DEXs like $GMX are inferior to orderbooks. "The future of decentralised exchanges does not look like this." "It's sort of like a casino.. It's not an exchange. "On GMX, [price discovery] can't happen organically, so they have to get it from an oracle." https://t.co/kzN8TmvvIa

On the opposite end, you have solutions that serve how markets are ultimately “supposed” to work — namely, an order book. If you reason about mechanics long enough, you’ll always find holes unless you have a clean order book model (or something that approximates it like Uniswap v3).

Crypto markets and the most popular crypto protocols today have never succeeded with that straightforward strategy though. They’ve always needed some “not perfect” mechanism to bootstrap liquidity to a level of usefulness.

Usefulness only enters the picture after breaching a certain liquidity threshold.

EtherDelta was a fully on-chain order book and never got liquid enough to trade. 0x had off-chain limit orders and on-chain settlement but couldn’t get over the hump. Uniswap (x*y=k) finally got us there.

Now does Uniswap (v1 and v2) strictly make sense for building a trading engine?

LPs are passive so they subsidize or take the hit from profitable traders. You hope fees compensate for that, but probably not. It also helps when there are liquidity mining programs that hide LP's unprofitability.

Yet Uniswap bootstrapped liquidity above a certain threshold that allowed DeFi to take off. Uniswap v3 is the version that makes strict sense because it has its previous versions as a feature but also can approximate an order book.

Luckily, traders lose money on Gains and GMX.

“Cheat coding” liquidity plus being able to offer degen features — while protecting the protocol from losses that are intolerable — is a viable and workable path for a crypto protocol to mature.

For Uniswap, it worked to bootstrap liquidity for the long tail of tokens. The closest analogy to that in the derivatives space is Gains Network, which offers stocks, indices, commodities, cryptos, and forex trading.

Gains Network (and others) lose on efficiency in the idealized case, but win on a few other fronts:

• They kind of have to be less efficient than e.g. Binance or traders get too good of a deal.

• Risk limits have to be placed so that the losses never become too large to the protocol (or its LPs).

• If those parameters are filled, you can offer lots of trading pairs and extreme leverage (such as 1,000x on forex).

Conclusions

• Does it strictly make sense to trade on Gains or GMX over other venues?

• Is the protocol and its LPs taking on risk from the positions on the platform?

• Is this the most efficient and sustainable way to set up a market?

You are not going to find great answers to those questions.

But I think despite this, one of these derivatives platforms will become super successful by bootstrapping markets with protocol-owned/managed liquidity and overlaying an order book (traditional or Uniswap v3) on top later on.

This is probably an “easier” path than dYdX — which I’d expect to be successful too because it’s a great product already and the roadmap is sensible. On the other hand, without incentives, it’s difficult even for them to have liquidity and add a substantial number of trading pairs.

Decentralized casino tokens as a trade will be one where degen retail wins and smart investors lose because thinking about it too much leads to struggling to being able to justify an investment.

The “thread the eye of the needle” view is that GMX, Synthetix, and Gains Network bootstrap liquidity and users via a slightly (but not terribly) unsustainable mechanism — while overlaying the long-term solution on top later on.