As most of you know, I rely on statistics pretty heavily when I'm evaluating a player or team. As we've discussed a few times on here, the statistics themselves are objective, but the decisions about which ones to pay attention to and how to interpret them is often subjective.
I thought it might be worthwhile to try to explain what I personally look at, where I find it, and what I take from it. This article is part one, walking through how I evaluate forwards.
Suppose someone asked me what I think of Ryan O'Reilly. I haven't seen much of him and don't have much of an opinion, so where would I go to evaluate him?
For forwards, the three things I always look at are how the player was used, whether the team controlled the play with him on the ice, and his individual scoring contributions. Then there are some other factors I'll look at if I want to get a more detailed picture.
I usually start with usage -- I want to know what kind of situations the guy was put in so I have a sense for what to expect from him. To find that, I'll go to this page at behindthenet.ca, which I've customized to show the metrics I use the most and bookmarked so I can pull up those stats on one screen any time I want. Here's what it looks like:
The two key stats for usage are labeled Ozone% (does he start in the offensive zone more often than the defensive zone?) and Corsi Rel QoC (does he face the opponents' top players?). O'Reilly started in the offensive zone just 40.5% of the time, 11th out of the 12 forwards on the team, so he's seeing a lot of defensive zone draws. On the other hand, his quality of competition value of just 0.03 is 9th out of the 12 forwards, so he's generally not facing premium opponents.
I pay attention to both stats, but I believe zone starts impact results a little more than quality of competition does, so when they're in conflict I tend to weight the zone starts a little more heavily. In this case, O'Reilly's zone starts are also a bit more extreme than his competition, so I'm comfortable saying he was used in a defensive role. He's not a true heavy lifter facing top competition in his own end, but I still see him fighting a little bit of an uphill battle.
What can I conclude at this point? Not much, really. If he were getting the absolute toughest shifts, it would probably be safe to say he's a strong defender who his coach seeks to put in defensive spots. But his more moderate usage could mean he's a good defender, it could mean he's a mediocre defender who's even worse on offense, or it could just mean he's a mediocre defender on a team of terrible defenders. All I really know is that when I look at his results, I'll need to keep in mind that he was taking on tougher-than-average situations.
The next question is whether the team controlled the play when he was on the ice. Controlling the play is often measured by Corsi, which compares how many shot attempts each team got; since shot attempts come from puck possession in the offensive zone, this correlates closely with zone time and puck possession.
Again, I'll look at the behindthenet.ca page linked above, but this time I'm focusing on the columns that say "CORSI ON" (what was the team's Corsi when he was on the ice?), and "CORSI REL" (how much better or worse was that than when he was off the ice?). The relative measure is the one I pay the most attention to, since it helps to wash out the effect of the player's teammates, but I'll also peek at Corsi On.
Here, O'Reilly's Corsi Rel of +11.1 means that the team did much better with him on the ice than off it. That's a very strong result, and especially when you bear in mind that he was facing tougher-than-average shifts, at this point I'm convinced that O'Reilly had a good year last year.
It's worth noting that his Corsi On of +2.8 means that the team controlled play with him on the ice, getting a couple extra shots per 60 minutes. His Corsi Rel being much higher at +11.1 means that the Avalanche got creamed when he was off the ice, with a Corsi of -8.3, but were slightly positive when he was on the ice. A positive Corsi Rel on a bad team isn't quite as impressive as a positive Corsi Rel on a good team, but +11.1 is still a big number and suggests that he really helped drive the play forwards.
We know he's taking on moderately tough assignments and pushing the play forwards, but is he actually generating goals?
His goals per 60 is 7th on the team, and his points per 60 is 8th. It's probably not a great sign that he was a below-average scorer on a below-average offense, but since I still don't have a great feel for what's a good G/60 or P/60, I usually double-check the traditional numbers at hockey-reference.com. That confirms that he didn't do anything special as a scorer, finishing with 13 goals and 26 points in 74 games.
I'm almost ready to conclude that O'Reilly is a very good role player, someone who can take on defensive situations and really drive the play forwards, but isn't a big contributor at the offensive end. But before I go there, I want to check for the role of luck.
Players' shooting percentages vary quite a bit from year to year just because they don't take enough shots for randomness (bouncing puck, post, great play by goalie, etc) to even out. So let's take a quick check to see if his goal scoring was suppressed by bad luck on his part or his assist total was suppressed by bad luck for his teammates.
It's a little tricky to evaluate with a young player, because he doesn't have a long history to compare to. But hockey-reference shows that last year he had a 10.9% shooting percentage, which is pretty reasonable for a forward and substantially higher than he posted the previous year. I don't see any reason to think his goal total was anomalously low last year.
We can also check whether he was unlucky on his teammates' shots. Going back to the behindthenet page, look at the column labeled Sh%, which is his teammates' shooting percentage when he was on the ice. They shot a poor 6.6%, so it's safe to guess that his teammates' poor shooting robbed him of a handful of assists.
Still, even accounting for the slightly tougher-than-normal starting situations and his teammate's poor shooting, it doesn't look like he was making great individual offensive contributions last year. I'm comfortable saying he was an excellent role player, someone who can push the play forwards but will need more skilled teammates to finish the play.
That's where I'd stop most of the time, but there are other details that sometimes are worth looking at.
For usage, you may have noticed that while I factored in competition, I didn't even look at teammates. I'm generally not crazy about the quality of teammate measures because the player's own talents impact the results of his teammates. If I want to check into a guy's teammates, I'll go to the behindthenet page that lists what fraction of his ice time was with various specific players and qualitatively estimate what impact his most frequent linemates might have had.
For results, there are a few additional things I might look at. A double-check on whether Corsi is giving us the right picture is looking at whether his zone finishes (Fin Ozone% on behindthenet) suggest he is moving the play forwards. Unfortunately, you can't just compare this directly to the zone starts, because zone finishes tend to end up much closer to 50% than starts -- it's a lot easier for a coach to control where a player starts his shift than for the player to control where he finishes it.
So to use this, we need Bettman's Nightmare's table showing what zone finishes you should expect from a guy with a given zone starts, so we can see whether O'Reilly did better or worse than expected. That table says that players with an OZ start% of 40.5% like O'Reilly normally finish in the offensive zone about 47.25% of the time. His actual offensive zone finish was 48.3%, so he did push the play forwards, and this meshes with the positive Corsi result.
With a veteran, I'm usually comfortable judging shooting percentage luck by comparing to previous years. But for a young guy who doesn't really have an established talent level yet, it can be helpful to look at whether his shots tend to come from locations where we'd expect a high percentage. Again, in this case I think we have the right answer, as his shot map suggests that a lot of his shots are from the outside compared to a shooter like Danny Briere.
Another individual contribution that I might look for when I'm doing a closer examination is penalties taken and drawn. The behindthenet page shows that he took the fewest non-coincidental minor penalties on the team (PTAKE/60) and drew more than he took (PDRAW/60), which is another positive in his favor.
I generally don't pay much attention to Sv% (the team's save percentage when he was on the ice) or PDO (his overall luck factor -- his team's shooting plus his team's save percentage) unless people bring +/- into the conversation. A low PDO indicates that his +/- was driven down by factors largely outside his control, his teammates' save and shooting percentages.
I also don't pay a lot of attention to special teams results, especially on the penalty kill. But I will sometimes check a player's power play scoring rate; you can just change the drop-down box on behindthenet to give 5 on 4 play and see how much power play time the player got and whether he scored at a relatively high rate in that time.
And finally, another useful link to bookmark is the behindthenet player card, which lets you skim a player's career stats. The apostrophe in O'Reilly's name seems to be screwing his up, but here's a link to the card for Wayne Simmonds. Just replace his name in the URL with the name of whatever other player you might want to look up.
There's a ton of information out there, and you can spend half a day digging into a guy's numbers if you want to. But in a one-minute glance at the behindthenet page linked above and maybe hockey-reference, I can get a sense for how a forward was used (Ozone% and Corsi Rel QoC), whether he drove the play forwards (Corsi Rel and Corsi On), and his scoring ability (goals, assists, and the luck affecting them).