The mechanism is nearly always the same. Trades are placed in a block account without a pre-trade allocation record, and at the end of the day the winners are assigned to the adviser's own or favoured accounts while the losers go to clients.
Regulators detect it statistically. If one group of accounts shows first-day performance far better than chance across hundreds of allocations, the probability calculation alone supports a case, and examination staff run exactly that test.
The defence is procedural: written pre-trade allocation, time-stamped, with documented exceptions. That requirement sits in adviser compliance programs under the investment-advisers-act-1940 and in compliance-manual procedures at brokers.
Related: churning, fiduciary-duty, investment-advisers-act-1940, compliance-manual, hypothetical-performance-disclaimer