Here is an example of my database

{
    ticker : "AAPL",
    date : ISODate("2016-10-18 08:00.000"),
    value : 120,
},
{
    ticker : "MSFT",
    date : ISODate("2016-10-18 08:00.000"),
    value : 50,
},
{
    ticker : "AAPL",
    date : ISODate("2016-12-18 08:00.000"),
    value : 160,
},
{
    ticker : "MSFT",
    date : ISODate("2017-10-18 08:00.000"),
    value : 40,
}

I want to get the last N documents sorted by date for each stock and then add the values together and then divide by how many documents were actually available for each stock. So lets say we need last 10 documents but one stock only has one document available then we will simply do value/1, but for other stocks it will be sum(values)/10.

The result needs to be something like this:

[{
_id: AAPL
calculation: 1.5
},...
]

So far this is the solution I came up with:

[
     {
        '$sort': {
            'date': -1, 
            'ticker': 1
        }
    }, {
        '$limit': last_n_docs*total_number_of_stocks
    }, {
        '$group': {
            '_id': 'ticker', 
            'calculation': {
                '$sum': value
            }
        }
    }
    ]

The issue with the above approach is

1. This just grabs the last days_needed*total_different_stocks, and doesnt account for the problem that some stocks might not have enough data, so in this case we end up grabbing more days worth of data for some stocks and less days for other stocks
2. The other problem is obviously we dont divide the calculation value by the number of documents that we used.

Thanks for your help!