While insurance can be a highly effective way to manage risks that are highly uncertain individually but average out effectively in large numbers, the problem with trying to insure against a market catastrophe is that the risks don't "average out" over time; instead, they clump together. After all, the reality is that if a large number of people buy insurance against a market decline - e.g., through a variable annuity with a living benefit rider (GMWB, GMIB, etc.) - then nobody will have a potential insurance claim while the market is going up, but virtually everybody will be "in the money" at the same time when there's a severe bear market.
Of course, insurance companies had some acknowledgement of this risk, which is why policies that "insured" against market declines did not pay out an immediately liquid benefit, but instead merely allowed the policyowners to draw out lifetime income which meant, at some point, they may eventually deplete their own assets and then draw on the insurance company's guarantee. Nonetheless, where a severe market decline occurs, regulators require the insurance company to ensure it has reserves sufficient to pay out on its potential obligations, requiring a huge allocation to be set aside; thus, while the insurance company may ultimately be able to make good on its guarantees, avoiding a knockout punch default, the impact to profits for the reserve allocation is so severe the "technical" knockout punch leads the insurer to leave the business anyway (as occurred with several annuity guarantee providers after the financial crisis).
The challenge is compounded by the fact that, given market volatility, the sufficiency of reserves to back market guarantees themselves become highly volatile, as a base of hundreds of billions of dollars of assets are backed by guarantees funded by fairly tiny (relative to the assets) rider fees. If there is an extended bull market - and the insurer has many years to collect fees before facing a market decline that results in a significant insurance exposure - the consequences can be somewhat more contained. But the fundamental problem remains that - unlike virtually all other types of insurance - it's not feasible to slowly, steadily build reserves against a slowly, steadily rising base of guarantees; instead, because all the contracts are tied to the same underlying stock market risk, virtually all the policies become a potential claim at the same time. For instance, if $300B of guaranteed annuities experience a severe 25% market decline, the insurance company is suddenly exposed to as much as $75B of claims, for which gathering a 0.5%-of-$300B - which is "only" $1.5B of fees - just doesn't cut it.
Notably, in other insurance contexts, companies are very cautious not to back risks that could result in a mass number of claims all at once. This is the reason why most insurance policies have exclusions for terrorist attacks and war, and similarly why it's so difficult to get flood insurance in many parts of the country. It's a crucial aspect of insurance that in the end, its exposure to risk is well diversified (allowing the law of large numbers to work) and not be overly concentrated.