Type of ML Model to represent features with a time link

Question

I am trying to build a model to predict the price of a contract. A contract has certain events and states on certain days.

For example:

2025/01/10 - volatility: 1.2, barrier:1.5,  coupon 1.1
2025/01/14 - volatility: 0.9, barrier:1.55, coupon 1.15

Brief explanation is that the coupon is paid out when the barrier is met.

I have 15 features for each event date including the state of the market at that date and financial params like volatility and forward price.

I want to be able to put in a variable length sequence (diff contracts may have diff number of barrier dates/coupon dates) into a network and get a price.

Firstly I normalized everything, including point in time normalised to the length of the contract

At first I thought of using an LSTM, but LSTM's are used to predict sequential data. This is not sequential data as the previous time step is unrelated to the next time step. Although there is a time dependent dimension to each of the features as each feature is linked to a point in time

What type of technique should I use in this case?

Example feature matrix
Time barrier coupon volatility forward
0.1    1.1     1.4     1.5       0.98
0.3    0      0        1.3       0.97
0.9    1.4    1.6      0.3        0.95
0.95.   1.0.   1.8.    2.4.       0.97

This feature matrix represents the definition of the contract. Basically at time 0.1, there is a barrier of 1.1 and pays a coupon of 1.4. At time 0.3, there is no barrier or coupon, but we have some volatility and forward information that does affect the price of contract. Changing the time values can significantly affect the price.

So it's not time series data as it's not a sequence over time. It's just representing events and state of the environment the contract lives in. A monte Carlo method with some of these parameters will calculate the price, but I need to use ML to do it.