Just staring at that photo leads to some thoughts on sequence ... still many details needed. And they're cleaning up already, even before NTSB gets there?

Did somebody mention concrete posts for a fence on the left side (in the direction the train was running? That could be an important detail.

Somehow the cab car derailed after hitting the truck. The "ramp" theory described elsewhere here makes as much sense as anything, and more than most. That's something we need the NTSB study for. However it happened, the cab car headed off to port. The lead truck dug in and created the trail in the photo as the back end came around, but the sudden end of the trail does suggest it hit something that triggered the rotations and overturns. If it "tripped" on something like a substantial post while still at some speed it could have rotated quickly and with enough force to tear the second car off the rest of the train.

As the back end of the cab car came around, it dragged the second car with it until the couplers broke loose, doing a sort of slingshot move. By then, the third car was hitting the damaged track too, but the couplers between #2 & #3 broke quickly and #2 pivoted out into the street on the end of the sling. It probably broke loose from #1 fairly quickly, allowing it to slide on its side for a while.

#3 and #4 may have stayed coupled until nearly the end, with #3 overturning (possibly started by a tweak from #2 as it detached) but staying in a straight line. #4 was derailing by then, too, but stayed coupled with the engine and upright. Also, by then the forward speed was probably pretty low.

Leads me to wonder how the couplers are supposed to work when twisted/rotated strongly. IOW, can the couplers keep a car upright that wants to overturn otherwise, such as going down the embankment or twisting around? This one looks like the couplers just gave up, fairly quickly perhaps, when rotated 90 deg or more, when swung well past their limits, or both.

The newsies all said 5-car train. NOT. I only see 4 cars in the pictures, with the locomotive being the only thing that could be a "fifth car." That's consisted with what I've seen of Metrolink trains - most are 4-car, with some as short as 2 cars off-peak, and only occasionally as much as 5 cars. Considering that most of their locomotives are 3000-3200 HP (F59 variants) that makes sense - not enough power for decent schedule-keeping with more than 4 cars.