Elastic - Filter after selecting top 5 hits - elasticsearch

Question

I'm using the alerting feature in Kibana and I want to check if the last 5 consecutive values of a field exceed a threshold x but if I use a filter in my elastic query, it gets applied before the top N aggregation.
Is there a way in which I can apply the filter after or check if the last consecutive values exceed a threshold using some other selector or method? I don't want to check this in the trigger condition in painless because that will return all the documents in the ctx and not just the ones which exceeded the threshold which I want to display in my alert message.
I've been stuck with this for a while and I have only seen blog posts saying sub aggregation is not possible on top N so any help or work around would be much appreciated.
This is my query :
{
"size": 500,
"query": {
"bool": {
"filter": [
{
"match_all": {
"boost": 1
}
},
{
"match_phrase": {
"client.id": {
"query": "42",
"slop": 0,
"zero_terms_query": "NONE",
"boost": 1
}
}
},
{
"range": {
"#timestamp": {
"from": "{{period_end}}||-10m",
"to": "{{period_end}}",
"include_lower": true,
"include_upper": true,
"format": "epoch_millis",
"boost": 1
}
}
}
],
"adjust_pure_negative": true,
"boost": 1
}
},
"aggs": {
"2": {
"terms": {
"field": "component.name",
"order": {
"_key": "desc"
},
"size": 50
},
"aggs": {
"3": {
"terms": {
"field": "client.name.keyword",
"order": {
"_key": "desc"
},
"size": 5
},
"aggs": {
"1": {
"top_hits": {
"docvalue_fields": [
{
"field": "gc.oldgen.used",
"format": "use_field_mapping"
}
],
"_source": "gc.oldgen.used",
"size": 5,
"sort": [
{
"#timestamp": {
"order": "desc"
}
}
]
}
}
}
}
}
}
}
}
}

Answer 1

Did you try to use a sub filter aggregation:
https://www.elastic.co/guide/en/elasticsearch/reference/current/search-aggregations-bucket-filter-aggregation.html
Or you can use a pipeline aggregation to manipulate your aggregations results
https://www.elastic.co/guide/en/elasticsearch/reference/current/search-aggregations-pipeline.html
by the way, a term query on the client id looks more appropriate.