clock menu more-arrow no yes mobile

Filed under:

The Tampa Bay Lightning power play drought: Bad process or bad luck?

The Bolts finally started to snap out of their dry spell last night.

Tampa Bay Lightning v Vegas Golden Knights Photo by Zak Krill/NHLI via Getty Images

While the second half of the season has been great for the Tampa Bay Lightning, one area where they’ve struggled has been the power play. The team that was so dominant with the advantage last season and in the early part of this season suddenly went on a streak where they were they only converted two of 42 chances.

In a small bright spot from Thursday night’s loss to the Vegas Golden Knights, the Bolts seemed to bust that slump with two goals on the power play from Steven Stamkos. We’ll need more time before we know whether those goals will be the signal that the dry spell has finally ended but the power play did look better last night.

Even if the slump is officially busted, over a month of trudgery like that is still worth looking into. If we cherry pick the endpoint that makes the slump look most extreme, the starting day is January 9th when they failed to convert on six minutes of power play time against the Arizona Coyotes. From that game until last night in Vegas, the team only scored two goals with the advantage. The first was on January 16th against the Minnesota Wild and the second was against the Pittsburgh Penguins on February 11th. That means last night’s showing matched their power play output over the previous 18 games.

Special teams outcomes are inherently prone to wide swings in variance because we’re dealing with small amounts of time that don’t allow for stabilization of rare events like goals or even less rare events like shots. So the first thought any time a team has a huge change in special teams results should be that they’re getting either lucky or unlucky depending on the context. In most cases, that will be the primary cause of a sudden change.

But when a streak goes on as long as this one does, checking in to see if the underlying play has changed to coincide with the poor results is worth doing. The following plot uses data from Natural Stat Trick and shows the Lightning’s cumulative shot volume, shot danger, expected goals, and actual goals on the power play since November 1st. I started the plots in November to remove some of the early season noise.

Shots, expected goals, and goals, are measured per two minutes of ice time. Shot danger is the expected shooting percentage on unblocked shots. The purple dashed line marks January 9th when the slump started.

We can go through each section of the plot individually. First, the Lightning have been shooting less during the slump. The lack of shots has dropped them from one of the most frequent shooting teams on the power play to having several teams ahead of them. That’s a meaningful change and something the team has already talked about addressing. When the Lightning slump, part of the reason is usually that they’re becoming too choosy with their shot selection and that’s the case here.

Their shot danger has been relatively consistent and maybe even increased a bit during this stretch. The Lightning always grade poorly in this area because they rely on lots of shots from the circles from Stamkos and Nikita Kucherov. Expected goal models will always treat these as lower quality shots because average shooters won’t convert those types of chances often. But Stamkos and Kucherov obviously aren’t average shooters and so in reality, these shots are more dangerous than they look on paper. Regardless, in the context of this slump, the danger of the shots generated hasn’t been one of the driving factors.

The drop in expected goals during the slump mirrors the drop in shots closely. That’s to be expected as shot volume and shot quality are the two components of expected goals. If shot volume has dropped while quality has remained consistent, the expected goals should drop right in line with the volume. And that’s the case here.

So what we can surmise about the process is that the Lightning have been generating less expected goals because they’ve been shooting less on the power play during this slump.

When we move to the goals section of the chart, we see a much bigger drop than in the expected goals section and this is where we get back to the variance discussion from above. A big difference between the expected goals trend and the goals trend indicates a change in shooting percentage. And as we know, shooting percentage is highly susceptible to variance in small amounts of time.

The Lightning have consistently been one of the best shooting teams in the NHL in recent seasons. Having Stamkos, Kucherov, and Brayden Point on the roster will lead to outperforming expected goals models because they are three of the best finishers in the NHL. But that hasn’t been the case lately as the team’s goal rate has dropped closer and closer to their expected goal rate. Thus the biggest contributor to this recent slump has been a cold shooting streak.

From that perspective, the primary conclusion here should be that this is just a slump and it will pass. That passing could have already started with the two goals last night.

But acknowledging that much of this is noise doesn’t mean the drop in shots should be ignored. A team can’t break a shooting a slump by not shooting. The best thing to do when the puck isn’t going in the net is to keep firing and maybe even try to increase the frequency of shots. That doesn’t mean taking bad shots but it does mean erring on the side of shooting to counteract the natural inclination to look for the perfect opportunity when the normal opportunities aren’t going in.

In a long season, dry spells happen, especially on special teams. That could mean a lack of power play goals or it could mean a particularly bad run on the penalty kill. Two goals in 18 games on the power play is a rough stretch no doubt. But the process doesn’t appear broken. Getting back to putting the puck toward the net more frequently would help. And from the looks of the game last night, it seems the correction might have already started.