Sum a variable only if step is smaller than next step

I have a csv with columns like: userid, reference, step. I have imported them in python as 3 lists like:

userid = data['user_id'] 
references = data['reference'] 
newstep =data['newstep']

for every reference value, which is inside this csv, I return the first 20 similarity values from a neo4j database. I want to write "extra" that if e.g. step[2]>step[1] for that reference value add in total value, else if step[1]>step[2] not add them and keep total as it is.

for references[1] in references:
    cqlNodeQuery          = ("MATCH (p1:Item),(p2:Item) where p1.value=$params RETURN  p1.value AS from, p2.value AS to, gds.alpha.similarity.euclideanDistance((p1.embeddingNode2vec), (p2.embeddingNode2vec)) AS similarity order by similarity desc limit 20")
    # Query the graph dynamically  
    nodes = graphDB_Session.run(cqlNodeQuery, params= references[1])
    for node in nodes:
        hey=[]
        predictions = [node[1] for node in nodes]
        step = [node[1] for node in nodes]
        from collections import Counter
        c1 = Counter(predictions)
        total=0
        # if step[1]<step[]
        def rec_rank(key,dic):
            total += dic[key]/float(sum(dic.values()))
            return total

In my code, total is summed without regarding the step value. Could you please help me with this?