It's generally accepted that Trump appears somehow to have deducted $916 million of losses, over time, despite the fact that these were predominantly other people's losses, not his. The question has been how he purported to do this, but the NYT appears to have advanced the ball with this story, which is based on the discovery of a 1991 opinion letter from Willkie Farr (itself available here).
At the front, the Willkie Farr opinion lists eight uncertain tax issues on which the tax plan to make other people's losses deductible by Trump depends. One of these it describes as "more likely than not" to be resolved favorably to Trump upon a full IRS vetting. For the others, it says that there is merely "substantial authority" for Trump's position.
The Times article says "substantial authority" means that one has about a one-third chance of being legally correct. I have heard of its being used to describe as much as a 40% possibility of legal correctness.
One of the odd things about tax opinions generally is that they don't quite explain what the likelihood of correctness means. After all, it's explicitly probabilistic for "more likely than not," and implicitly so for "substantial authority" even though the term appears to address the nature of support for the position. Presumably it uses a subjectivist conception of probability, based on the writer's risk-neutral assessment of zero-value betting odds. Perhaps this reflects an underlying thought experiment involving multiple repeated independent trials (in alternative universes?) or a grouping of equally plausible (albeit distinct) positions, all of which then get tested rigorously.
Another peculiarity about tax opinions with multiple issues that are assessed probabilistically is that there is no analysis of the degree of correlation between the probabilities. E.g., suppose each of two positions "should" be correct, and that we define that as 70% likelihood correctness. Is the likelihood that both are correct 70% - the case of perfect correlation between them - 49% (i.e., not collectively even more likely than not) - the case of perfect independence - or somewhere in between?
Tax opinions generally do not address this at all. In a typical case, however, there's almost no chance that they are perfectly correlated - the technical issues are distinct. On the other hand, perfect independence is often unlikely. For example, two legal issues that are distinguishable may nonetheless both be strongly influenced by the question of whether or not a given arrangement is economically meaningful.
Returning to the Willkie Farr tax opinion, if all of the issues were perfectly independent we're in the neighborhood of a 2/10 of 1 percent chance that the overall position was valid (from assuming six 40% probabilities plus one at say 55%). With perfect correlation, it would rise to 40%. But since perfect correlation is unlikely I imagine that reasonable views about actual correlation might place the probability of overall correctness anywhere from, say, 5% to 20%. And that's based on Willkie Farr's judgment, even though (with all due respect to Willkie Farr) there may have been a bit of preliminary opinion-shopping on Trump's behalf to get the most favorable view that any reputable NYC law firm was willing to provide.
What about audit review? I suspect it was not seriously reviewed, and here's why. It involved a gigantic NOL, having in the main a deferred impact on tax liability, while also raising many complex issues. It would be natural for an IRS auditor, seeking immediate bang for the buck, to figure: Even if I get this reduced substantially, surely there will be something left, meaning that Trump would have NOLs for the next few years anyway (assuming no possibility of immediate cancellation of indebtedness income given the insolvency / bankruptcy issues). And also, while this was still early in the era of the 1990s rise of tax shelters, the IRS may by that point have had more fish to fry than auditors to mind the fry pan (so to speak).
So, a bit like Dupin's purloined letter, it may have been so blindingly visible that it didn't get looked at seriously.
Back to the presidential race. What should we make of this? Well, for me, the fact that Trump, if he were president, would probably respond to the article by instructing his people in the Justice Department to shut down the New York Times and find some excuse for indicting the article's authors weighs a bit more heavily on the scale than the tax scam itself, as does his being a self-proclaimed sexual assaulter, a credibly accused fraudster and racketeer, and apparently a stooge (at best) in support of Russian interests at the expense of our own. So it doesn't worsen my views of Trump, although at this point there is literally nothing that could do so.