Hey crypto: Learn from an idiot today, make money tomorrow
As complex as it may seem, using behavioral game theory to find signals among “idiot” traders can provide real insight to crypto builders
Midjourney modified by Blockworks
There’s an entire cottage industry of crypto gurus, prognosticators and gem hunters out there. People who’ve built a rep and a following on Twitter, YouTube, Discord, the list goes on.
‘Buy this, sell that!’ they cry, as though somehow their technical analysis, their ‘insider information,’ will actually help you to make money.
Of course, most of them are complete idiots — or at least, charlatans — whose advice will literally leave you rekt and helpless.
But these aren’t the idiots we’re talking about.
Arguing against mainstream financial economics in the 1980s, former Treasury Secretary Larry Summers famously quipped “THERE ARE IDIOTS. Look around.” Traders at the time were typically modeled as informed and rational, and the market price was considered always right.
Summers favored models which had “idiots” (called “noise traders”), who could be the source of profits for informed traders. Such models remain useful today; See, for instance, the excellent article on Automated Market Makers from Milionis et al.
And applying the lessons of these noise traders is how you can extract signal — if you’re smart.
Cryptoeconomics — already at the cutting edge of cryptography — now finds itself approaching the frontier of modern game theory.
And that frontier is thoroughly behavioral, meaning that it uses models that incorporate psychology and sociology in a clear, formal way. Crypto is only beginning to use behavioral models, but — from “fairness” in MEV, to “idiot trading” or “noise trading” in DeFi, to the crucial importance of socio-psychological “legitimacy” — it is clear that cryptoeconomics is having its behavioral turn.
When the behavioral turn happened in economics, reactions were initially mixed. A surprisingly large number didn’t believe the findings at first, and many others admitted to having suspected the results but not modeling them.
Since then, the progression toward incorporating useful behavioral factors has been steady, and it is uncommon to find a theorist who does not recognize the relevance of bounded rationality or other-regarding preferences.
But that split between “those who don’t know” and “those who know but don’t say” is a good reminder of why it’s so important to model things clearly in the first place: Failing to do silos important knowledge in individuals and leaves others in the dark.
And this is certainly the case for crypto today. The best practitioners and designers have an intuitive understanding of behavioral factors — and they use it to their advantage.
Crypto is turning behavioral for much the same reason that economics did: Traditional game theory is typically “too weak to predict anything at all.”
Because of this deficiency, game theory began to incorporate hidden information, imperfect rationality and other-regarding preferences. These new models provided greater explanatory power and predictiveness, and even helped unearth new insights — it can also provide new explanations and perception for builders in crypto.
Lessons for DeFi tokenomics: Mitigating noise trader risk
Not all “noise traders” are created equal! When designing a market, setting up an airdrop and designing a launch, it can be useful to consider which sorts of trading “types” are being drawn to the protocol, and what predictable effects they will have on market behavior.
Essentially, I’m arguing that it’s best to use behavioral game theory to find some signal in the noise traders’ noise.
Drawing on research like that of David Hirschleifer’s, it’s possible to consider a typology of traders that helps create “product-noise trader fit.” Caricaturing a bit, is it better to attract HODLers who may resist downward price movement but may also sell at the first sign of a gain? Or what about jumpy traders who are happy to watch the number go up but sell when things go badly?
Once a token is launched and a market is created, we expect that arbitrageurs will step in to keep the price within some reasonable bounds. But there are some intriguing behavioral mechanisms such as M^ZERO’s SPOG. These can further to consistently penalize certain types of risk-avoiding behavior and reward those willing to put skin in the game, thereby maintaining a certain profile.
Behavioral game theory in action
Behavioral game theory can make useful predictions, as discussed above, but it can also suggest new incentive mechanisms entirely.
To take one example, consider a proposal for “Subgame Credible Anti-MEV” that sanctions certain types of on-chain value extraction, called “MEV.” The especially troublesome kind of MEV is sometimes called “Toxic MEV,” effectively exploiting users to make a relatively riskless profit.
It so happens that behavioral game theorists have found that many humans who feel they’ve been exploited are consistently willing to pay a cost to punish the exploiter.
This deeply behavioral aspect of humans is sometimes called “altruistic punishment,” and it can be a credible way to reduce toxic MEV. The rough idea is to require the agents who could extract MEV to stake some money.
Read more from our opinion section: Decentralization is a zero-sum game
And while “staking” itself is not a novel concept, the difference here is that the specific slashing rules can not be specified in advance. This is because MEV itself can sometimes be difficult to specify in advance. To address this problem, altruistic sanctioning can be made to credibly enforce the general condition: i.e., “your block made people so mad they were willing to burn their own money to penalize you.”
The point of such a mechanism is not to have a lot of sanctioning, of course. It’s that the would-be exploiters know they’ll be penalized, which makes them behave more fairly.
It would be as if a king is passing laws, but then has to parade through town where everyone can throw tomatoes at him. The threat of the tomatoes makes him think a little more carefully about what laws he passes, and that means that fewer tomatoes will be thrown. And this mechanism can generalize to cases beyond MEV — there is a proposal in the works to use it for Sybil resistance in Gitcoin as well.
And this is just one example. Overall, incorporating behavioral models in cryptoeconomics can provide new explanations and insights for builders in crypto, leading to more efficient, effective market and token design.
Don’t miss the next big story – join our free daily newsletter.