I mostly thought he was just... wrong. In retrospect it's interesting that his intuition was completely different than mine (and a lot of ours here).
Now consider yourself versus the average bike rider you see. Are they wearing a helmet? Following traffic rules? Using lights? Taking the entire lane? Signalling? Making left turns from the perpendicular street instead of merging into the turn lane for cars? Chances are they might only be doing one or two of those things, but probably none of those things on average based on my own anecdata of what I see cyclists looking like in my city. If you consider riders who do all of these things, whats the fatality rate then? I'd bet its effectively zero a year.
Crime works accordingly. We hear people say how they think the murder rate is a good proxy for crime you'd experience in a city, since you can't really hide a body and fudge the stats like other crimes that might go unreported. However, the murder rate is also a case of a small number of occurances happening over a massive population, and probably subject to a lot of variation. Likewise, it probably disproportionately affects people in domestic violence altercations who already own a gun that they keep in the home and know their victim, then secondly people involved with organized crime. Avoiding these factors that put you at outsized risk might make the murder rate effectively zero for someone like you.