One could also compare the rising of the base currency of the specific market with the change of the indices.
Also using ticks, count the upticks vs. downticks and their frequency, on crash the trading frequency is massively increasing.
If market aggrees, more down trades are made than uptrades.

Overall assets tick frequency, all pairs, all indices -> massive increase.
Gap frequency, all pairs, all indices -> increase
Sample pairs / indices, one for all, e.g. USD/NZD rising, oil falling, us30 falling, AUD/USD falling ... -> crash

Overall amount of price quotes massively increasing -> crash???

Since real crash data is very limited, concentrating on the forensics of the known crashs in the newer past would be a feasible option.