Market Efficiency is a convenient argument of faith which has a more nuanced rational basis, but absolutely in the general case is untrue.
In the long-term, markets are certainly efficient. However, the period of inefficiency is determined by the absolute quantity of profit dollars available in the market which in itself determines the competition. That is, the quantity of ticks (time) required for a dynamic optimization w.r.t. profit extraction is the measure of competition in some market, and thereby the degree of efficiency.
It is then true, that in the most competitive markets, the ones with the most capital, have a very small window of inefficiency, but they do exist. It requires a lot of capital to dynamically extract profit from those inefficiencies. An example of this would be the near hegemony of Citadel Securities in making markets. It is impossible for anyone today to compete in the market-making (an advanced arbitrage) industry without tons of capital.
However, we are time-bounded beings so there is always present an inefficient market. The challenge of the proprietor is then to determine that market which can provide a return to his constrained capital.
For instance, a restaurant can give a 15% annual ROI, which outperforms the S&P. The Citadel is not going to be interested in this inefficiency because that ROI is unavailable to the amount of capital they're trying to invest. So what we see is that the most competitive markets beget the most capital, because they have the most profit available to extract absolutely; then the most amount of resources, where capital is the analogue, are invested into those markets--so any inefficiency is quickly eroded.
This means that just like the restaurant, there are markets that remain untouched for long periods of time due to their inability to return large amounts of capital in absolute terms. These markets then have a great ROI in terms of percent. It is these markets that are available to the proprietor with low capital.
This is why the startup advice of finding a niche first is so salient, they are always necessarily underserved non-competitive markets. There's the other cycle from crossing-the-chasm which removes enterprises from this market who have the same idea as they scale, that's a separate explication, but this cycle allows the period of inefficiency to be maintained in those markets through that self-regulation.