You're both right.
Which is why the SDQL isn't some sort of automatic golden goose, and why none of the various touts that base their picks off of it running hundreds of scenarios aren't above 55% lifetime, and why most non-touts spend hours building high win-rate scenarios in the database and then see them go 0-7 or 1-10, etc, etc, very quickly...despite those scenarios being "75%+ ATS lifetime".
It's a lot like any computer-based effort: garbage in, garbage out. Which is to say that it's only as good as the person mining it for results in terms of how effective it is.
All you can do is create a basis of what's important to you in terms of what you feel truly affects the potential outcome of a game. Once you have that base, you can do research to build systems and scenarios that confirm or deny your beliefs. After confirmation from previous results it's about playing the systems for multiple seasons (fewer seasons if the amount of plays per season are larger) and then whittling down the herd to the true 'winners'.
And even then as years pass the winning systems will eventually require updates or changes as the game itself changes (NFL is a good example, few systems from prior to 2003 are worth anything because the scoring and offense/defense has changed dramatically over that time).
Test test test, and keep your bankroll conservative and in the long run the SDQL becomes a great tool to help identify EV plays in a big way.
That being said, I like none of my picks today, especially because the past few weeks I've been on a higher than normal win rate and I expect to regress to the mean at any time. 76ers, Clippers, Pelicans, none of which I would have put money on today if left to my own opinions, all brutal. But that's the game if you truly are trying to confirm whether your SDQL scenarios are 'real'...or not. Gotta play the stinkers.