What Happens When You Fail to Bench a Non-starting Goalie?

A common, and seemingly benign, mistake in fantasy hockey occurs when a manager fails to bench a goalie who is not starting in a particular game. There are some instances where a fantasy hockey manager will make the deliberate choice to place a non-starting goalie in the starting slot on his/her roster. This post examines the ramifications of that mistake.

In the 2011-2012 season, at least one goalie was pulled in 14.8% of the games played. In an 82 game season, that means a goalie is pulled nearly once a night. We examined the statistics of the goalies who came into the game after the starting goalie was pulled. We’ll call these goalies who came into the game after the puck was dropped replacement goalies. These replacement goalies entered the game either because the coach pulled the starting goalie or the starting goalie suffered an injury. For the purpose of this post, we’ve ignored the reason for the replacement.

Statistical Comparison of Replacement Goalies vs. League Average
Metric Replacement Goalies League Average
Win Percentage 20.6% 45.4%
Save Percentage .909 .914
GAA 2.54 2.54
SHO 0% 7.2%

The most obvious result is that the winning percentage of these replacement goalies was very low. For the 2011-2012 season, that winning percentage was 20.6% (for reference, very good starting goalies live in the 60% neighborhood, while your Mason/Sanford types sat around 30%). The SV% of replacement goalies was .909, while the league average sat at .914. Finally, the GAA of replacement goalies was 2.54 which matches the value of the league average.

By leaving non-starting goalies on your active roster, fantasy hockey managers will cost themselves points in every standard goalie category. The quick lesson here, then, is never* leave a non-starting goalie on your active roster.

*See first comment below.

Comments (14)

  1. Mike

    While this article was written specifically for the standard Yahoo goalie categories, it has been pointed out that managers can achieve gains in certain counting categories by leaving non-starting goalies on their active rosters (e.g. Saves). Depending on your scoring settings (and whether or not you’re in a H2H league), you’ll have to weigh your decision to roster/bench a goalie on any given night. In leagues without limits on roster streaming (i.e. who can start the most players each week), a manager could benefit by keeping non-starting goalies on their active roster.

    In leagues with caps on games played by goalies, it almost never makes sense to have a non-starting goalie on your active roster.

  2. We have a significant selection bias here. “Replacement goalies” that enter the game mid-way are usually the backups, ie: not as good as the starters. That would at least partially, if not entirely, account for the worse ratios. It follows that if a team’s #1 goalie replaces the team’s #2 goalie, this trend of worse ratios should not be expected to continue. Regardless, if your league uses a GP maximum, and you are likely to reach that maximum (as you almost certainly should), you should never activate a non-starting goalie. No chance of a shutout, very little chance at a win, and even if the ratios are good they won’t help as much because the amount of saves and minutes played will be lower than if he’d started the game, and thus have less effect on season totals.

    • Mike

      The selection bias is smaller than you might think and one of the reasons for that was pointed out by you: backups are generally not as good as starters. For the 2011-2012 data set, the goalies that were pulled can be broken down as follows:

      60% were the starting goalie for their team
      40% were the backup goalie for their team

      Beyond that, consider backup goalies from 2011-2012 like Elliott, Schneider, Rask, Holtby, Giguere, Enroth, and Hedberg (to name a handful) who actually outperformed their starting counterpart in the SV% category.

      I haven’t yet broken down the data by starter vs. backup, but this will be a good project for Sunday and should help to flesh out any selection bias in the argument.

      • of course more starters were pulled than back-ups: they play far more games! its almost alarming how close those percentages are considering how few games the backups play. Wouldnt a more accurate stat be percentage of games pulled in relation to those started for both groups? (ie starters play 2000 total games in a year, starters pulled in 250 games for a Pulled % of 12.5%)

  3. In my Yahoo! Rotisserie Keeper League we have a GP (games played) limit but we also have a GS (games started) scoring category. In our league benching a non-starter is vital because if you start a “replacement goalie,” you are essentially giving away points. The reason is that GP is not the same as GS and icing a replacement goalie adds to your GP but not your GS. As soon as that happens it is impossible for you to max out your games started category. If you properly manage your goalie position you can ensure that, at a minimum, you finish tied for first in that category.

    With that all being said we know that GS is a stupid category and much debate has taken place about dropping it but in the end we decided to keep it. The reason being that this category rewards managers who are active in the league and punishes the stragglers. Call us fantasy hockey geeks, but that is our rationale.

  4. As a CBJ fan, i thoroughly enjoy that your 2 examples of losing starting goalies were both on Columbus. one day soon we’ll be contending, you’ll see ahah

  5. However, for example Brian Elliot (or Jaro Halak) is a good example of a goalie who might be worth activating even if he is not starting. Hitch tends to pull his goalie relatively quickly, and Elliot has been solid in both games when Halak has been pulled (once he even got the winn, while not letting any puck in the net and thus getting a 100% and 0,00 GAA; the second time He played solid, and if not for an absurd call against Backes which gave the Wings a 5min PP, he could have won that game as well (or at least have very solid stats from the game)

  6. I got lucky and got a win with Eliot when Halak was pulled, around the third game of the season. But, I always leave the goalie in because sometimes the other goalie stars despite the updates. Maybe this doesn’t happen anymore, are there still coaches who keep it a secret as to who starts??

  7. Wings9798

    I may be unique unless there are other Halak and Elliott owners. When i drafted one i had to get the other. If my other goalie isn’t playing I start both of them. Last week that helped me. Halak got pulled (Now we know he was hurt) and Elliott came in. That got me saves and a better Sv%.

    If I had Luongo and Schneider i would probably do the same. I know have Emery and Crawford in another league and when Emery starts i will start Crawford too odds on being Emery gives in some cheapees (except his last start, he was superb). I won’t however staqrt Emery when Crawford starts. odds are if Crawford gets pulled Emery will just be a continuation.

  8. it seems rather obvious that leaving a player in your lineup who will likely not play (and if he does play, it’ll very likely be in a loss) will kill your points production.

    • Mike

      What was obvious (and that was pointed out) was that the Win % would drop significantly. What was not known at all were the SV% and GAA numbers for the replacement goalies.

  9. What’s failed to be mentioned here is that wins and shutouts are also counting stats. While the win % drops significantly, it still remains at 20%. That’s well above the 0% chance you’d have by leaving a goalie on the bench. The likelihood of a shutout would be unchanged. 0% if you play a backup or choose to bench him.

    The GAA remains unchanged, and the relative change in SV% is less than 1% based on the above table.

    I would conclude based on the research in the table above that unless you are in a league with a games played limit you actually hurt your chances by not putting a backup goalie in your active roster. The likelihood of a win is still greater than the downside of GAA and SV%.

    • Mike

      I’d agree until your last paragraph. If you’re in a weekly head-to-head matchup and you’re winning the SV% category by a narrow margin, the numbers suggest that you could lose that category if you keep a non-starting goalie on your active roster.

      As an aside, that 20.6% Winning percentage is for replacement goalies that end up playing. If you run the math on your odds of getting a Win for just having a non-starting goalie on your active roster on any given night, the result is about 1.5%.

Leave a Comment

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>