How Safe Is a Lead?
Your team scores first and leads after the opening period. Do you relax, or is a hockey game too long for one goal to mean much yet?
Lead safety
Across every finished regular-season game league-wide, somebody is out in front after the 1st period 75% of the time. The rest are tied. When a leader does exist that early, they go on to win the game 79.8% of the time. That's a strong hold rate for a period this early. Two periods remain, and the team in front is already the clear favorite.
So does the picture change once the 2nd intermission rolls around, with only twenty minutes left instead of forty? It changes a lot. The hold rate jumps 8.8 points, from 79.8% to 88.6%. A 1st-period lead is a lean; a 2nd-period lead is a genuine signal, because there's less time left for it to disappear.
Tie watch
If leads aren't automatic, the flip side is worth asking too: how often does neither team ever pull away at all? The model has its own opinion on how often games should end level. The question is whether that opinion matches reality.
The gap runs the other direction: the model predicts 11.5% draws but only 10.4% actually happen. It over-calls ties. Games in this league resolve one way or the other slightly more often than the model expects. So when a specific game's predicted draw chance clears 17.7%, that's not noise. It's the model flagging a genuinely stalemate-shaped matchup, which is exactly why the site calls it out.