because I’m thinking about it now, some heuristics on market efficiency:
1) volume
the single best indicator for how efficient a market is is how much volume trades on it. the higher the volume, the higher the rewards are to trading against any mispricing, so the stronger your priors should be against there being easy-to-find mispricings. your uncle’s football betting pool can easily be inefficient because it’s not worth any professional’s time to trade on it, whereas anyone who can predict short-term price movements of the S&P 500 will be very heavily rewarded for doing so.
2) costs and barriers to entry
these can come in many forms:
- literal trading fees
- rules like low position limits or restrictions on short selling
- operational and technological barriers
- regulatory and other risks
- costs of capital (markets that require more collateral to trade or are harder to get capital in and out of will be less efficient)
if you are already a trading firm trading on the Nasdaq, the additional cost to trading another Nasdaq stock is very low, so even though a stock might be much lower volume than AAPL it might not be much less efficient. (at least in terms of things like incorporating information from overall market moves or sector news. it might be much less efficient in incorporating information from earnings or other fundamental news, since the costs there scale more linearly with the number of stocks you’re analyzing.)
on the other hand, a crypto exchange in Indonesia that only allows Indonesians and doesn’t have an API will probably be way less efficient than the most efficient crypto markets.
in an extreme example of regulatory risks, US stock markets (while generally fairly efficient) are very much not efficient in incorporating information about upcoming mergers, because the regulatory costs to trading on much of that information are very high.
upon looking at this I guess I’m positing a sort of meta-EMH: “markets are efficient to the extent that the rewards for correcting inefficiencies are greater than the costs to doing so”
which I think is a reasonable prior. but of course my whole job is trying to find cases where this doesn’t quite hold