Take, for example, an S&P out of the money vertical credit spread... for example, say it has a $1000 limited upside with 90% probability and a $10,000 limited downside with 10% probability... Taleb's point (Black Swan etc) is that this is a losing strategy, from an expectancy viewpoint, even though the trader may be on a seemingly endless winning streak... that winning streak will soon enough be rudely interrupted by one or two trades that will eliminate all gains to date...
Larry McMillan is in the same camp -- that camp being, if you are gonna spread, pay for a cheap debit spread and do not simply book small income from a credit spread with an unfavorable R:R -- McMillan, for this reason, is a proponent of long straddles i.e. many losers, but sooner or later you get it all back and some...