--- title: "How Many Games Before EASHL Stats Are Reliable | 431,000 Games Analyzed" description: "We measured every stat against 431,000 tracked EASHL player games to find out how many games each one needs before it describes you instead of your luck. Save percentage takes four times longer than goals." canonical: https://chelhead.com/blog/how-many-games-before-your-eashl-stats-mean-anything/ last-updated: 2026-09-07 --- # How Many Games Before Your EASHL Stats Actually Mean Anything? > We measured every stat against 431,000 tracked EASHL player games to find out how many games each one needs before it describes you instead of your luck. Save percentage takes four times longer than goals. Published 2026-09-07 by ChelHead Team ## The Question Behind Every Argument Six games into the season, someone in your club has a .930 save percentage. Someone else is shooting 4%. One of them is getting told they're carrying the team and the other is getting told to stop shooting. Both takes might be wrong, and not because of anything to do with hockey. Six games is simply not enough games for either number to be describing the player rather than the sample. The interesting question is: how many games _is_ enough? And the answer turns out to be wildly different depending on which stat you are looking at. So we measured it, across 431,600 tracked player games and 58,000 matches. ## How You Measure This The method is older than EASHL and it is refreshingly blunt. Take a player's first N games and their next N games. Work out their stat in each half. Do that for hundreds of players, then ask how well the first half predicts the second. If a stat is measuring the player, the two halves agree. If it is measuring variance, they don't. The point where the two halves are half in agreement is the conventional line for "this number is now more signal than noise", and it is the number we went looking for. The same group of players is used at every sample size. Otherwise short samples would include everyone and long ones only the most active players, and the comparison would mostly measure the gap between a Division 1 player and a Division 10 one rather than how quickly a stat settles. ## What We Found For skaters, here is roughly how many games each stat needs: - **3 games:** goals, shots, hits, giveaways, possession - **4-5 games:** points, passing %, blocks - **6-7 games:** faceoff %, shooting % - **8-9 games:** penalty minutes, assists, interceptions - **12 games:** takeaways, plus/minus At club level the same pattern holds, with save percentage the slowest of the lot at 12 games. Three games into a season, a club's save percentage tells you almost literally nothing: the correlation between one half and the other sits at 0.10, which is a rounding error away from flipping a coin. ## The Uncomfortable Part Line up that list against what people actually argue about and it is close to exactly backwards. Hits, giveaways and possession settle almost immediately, and nobody argues about them. Shooting percentage and save percentage take two to four times longer than goals do, and they are the two numbers everyone reaches for first. Plus/minus is the slowest stat we measured on a player, and it is the one that gets quoted after a single bad night. There is a reason for that, and it is the same reason plus/minus fell out of favour in real hockey analysis. Your plus/minus is mostly a record of what five other people did while you were on the ice. It takes a lot of games for your own contribution to out-shout theirs. The stats that settle fastest are the ones that measure what you _do_: how often you shoot, how often you hit, how much you have the puck. The ones that take longest are the ones that measure what _happened_: whether the shots went in, whether the saves were made, whether you were on the ice when it did. Style stabilizes fast. Outcomes stabilize slowly. ## The One We Couldn't Answer We wanted a number for goalies most of all, and we can't give you one. Roughly 77% of EASHL games are played with an AI goalie. Human starts are spread thin across a lot of different people, and across our entire collection only 83 goalies have played 20 or more complete games. That is not enough people to measure anything with, so rather than borrow the club number and dress it up as a goalie number, we are leaving it open until there is enough data. Which is its own kind of answer. If it is that hard to accumulate enough goalie games to measure a save percentage, be very careful what you conclude from twelve of them. ## What To Do With This Nothing complicated. Just check the denominator before you reach a verdict. Under 5 games, believe the volume stats and nothing else. You can tell who is shooting and who is hitting; you cannot tell who is finishing. Around 10 games, most things are readable, but shooting and save percentage still are not. Past 15, most of the page is fair game. None of which means a hot start is fake. It means a hot start is a hot start, and six games is too early to know which one you're looking at. --- ChelHead is an independent EASHL stats tracker. This web site is not endorsed by or affiliated with Electronic Arts Inc. Machine-readable index: https://chelhead.com/llms.txt · Sitemap: https://chelhead.com/sitemap.xml · Contact: support@chelhead.com