Signal Engine

This is arithmetic on numbers I chose. None of them come from Neo, and the scores are not probabilities. Change any input and every row updates. Replace mine with Neo’s own funnel and the page becomes a real estimate.

A flag only counts if the quote behind it contains a trigger term. The code stops if it does not. Weights are my starting guesses, not fitted to Neo’s closed deals.

def score(a):
    parts = []
    for key, idx in a["flags"].items():
        q = a["quotes"][idx][0].lower()
        r = RULES[key]
        assert any(t in q for t in r["triggers"]), f'{a["id"]}: quote for {key} has no trigger term'
        parts.append((r["label"], r["points"], a["quotes"][idx][0]))
    parts.append(("Posting sits with the " + ("security or governance team" if a["team_kind"] == "security" else "platform or developer-experience team"),
                  TEAM_POINTS[a["team_kind"]], None))
    age = age_days(a["posted"])
    if age is not None and age <= FRESH_DAYS:
        parts.append((f"Posted {age} days ago", FRESH_POINTS, None))
    if a["regulated"]:
        parts.append(("Regulated industry (" + a["industry"].lower() + ")", REGULATED_POINTS, None))
    raw = sum(p[1] for p in parts)
    mult = SOURCE_MULTIPLIER[a["source"]]
    return dict(raw=raw, mult=mult, score=math.floor(raw * mult + 0.5), parts=parts, age=age)

This is how I would wire the score into Neo’s stack. Nothing here is running, and no figures on this page come from Neo.