POWERPLAY 雜誌

Well fellow PPMers,

It has been some time since I published anything of PPM-life-changing substance. I've sat on this research for awhile and debated if/when to share. Tonight I thought I'd take a break from my daily grind and return to writing something in the PPM world, so you all don't start to feel neglected ;). First, let me set the stage for you.

Season 15 was a very disappointing year for my hockey club. We sat first in Canada's II.3 with 105 points of the possible 114. We swept through the quarters and semis with 9-0, 4-1, 8-0 and 7-2 wins. Life was good. The finals were another story. I was pitted against my arch rival from Hogtown. I dropped game 1 5-3 despite outshooting him 40-19. I followed this up with a 2-0 win on the road (shots 30-28 in my favour again). The third and deciding game would be dominated by me once again, with shots 45-25, but after OT and the score notched at 1-1 I'd lose in a shootout before falling in the promotion/relegation round and remaining stuck in II.3.

So "why the anecdote?" you may ask. Well, this whole experience, which PMullings and the II.3 forum will never let me live down, made me take a good hard look at my team. His goalie was good, granted, but he is not that skilled. So, I dug deeper and upon doing so I noticed something remarkable. Thanks to my level 15 medical center my entire lineup of 20 skaters played all 38 games (okay so there were a few day-to-day games played but that's a small effect). I kept the same on-ice ratios, lineup and training ratios for the whole season.  That being said, shooting training varied considerably from player to player, which ultimately is to be responsible for this article.

I had a look at the numbers more closely. Braden Dubeau, my top player, led the team in shots with 222. This is no surprise, the developers have said from the beginning that the primary attribute (provided the secondaries are up to par) is the main influence on taking shots. Second on the list was my second-highest offensive attribute player and so on down the list.

Looking at goal generation the chart isn't so black and white. You have a mix of top shooters and top offensive players.  I dug into the data to try and see what I could find. I took all my forwards and pasted their attributes into excel including their on-ice percentages. This allowed me to determine their shots per 60 minutes played and goals per 60 minutes played, normalizing everyone to the same baseline.

It is imperative to keep in mind these results are for my situation only. They will vary depending on opponent, line structure and a whole whack of other caveats to consider. So, please consider this simply a mental thought experiment with some number crunching!

Looking at shots per 60min based on primary attribute gave a linear trend. While not completely real (negative intercept) I'll run with it for now. Keep in mind, players won't get 15 shots per game at 600 primary. That is 15 shots per 60 minutes. So if they play 15 minutes per game you're looking at around 4 shots per game given the typical defence I was up against.

A similar correlation was observed with the shooting attribute. Again, the correlation is unrealistic due to the negative intercept, but, we'll proceed nonetheless.

Now, here is where things get fun. Clearly, building a player with more primary will get more shots and with more shooting will get more goals. But, you can't train at an infinite rate. So, where is the optimum build when considering shooting on forwards?

Let's consider you have a certain amount of attribute points to give out, how would you split them up? We'll assume 2:1:1 is the most efficient ratio for primary and secondaries. How much do we give to shooting? Based on the above two relationships, we can figure it out; again, in the case of my team against the opponents I was facing.

I set a cap of 1,000 EOR. I then assigned increasing amount of shooting: 100, 115, 130 etc. all the way to over 500. However, for every shooting attribute point used, that was a point not given to the primary or secondaries, so they would decrease accordingly while remaining locked in the 2:1:1 ratio. Looking back to the above two figures you'll notice a positive trend (increasing primary increases shots taken and increasing shooting increases goal scoring). But, in this case, increasing shooting will decrease primary. So, while you'll score more of your shots taken, you'll take fewer shots. At a certain point the number of shots taken and shooting efficiency should make for the most goals scored, and that is what I wanted to uncover.

For each build simulated, I predicted the number of shots taken per 60min (from the first graph) and the scoring percentage (from the second graph).  From there I could easily figure out the number of goals scored per 60min (shots taken * scoring percentage). See the result in the plot below. Note, the y-axis is normalized to 100. You will not score 100 goals per 60min of playing time in any scenario. This is simply to illustrate the relative changes in build ratio with respect to the "optimal". Putting an absolute value on it - such as 10 goals per 60 minutes - is redundant seeing as choosing lower than 1,000 OR as my point distribution would result in less, and more than 1,000 would result in more goals per 60min, hence my choice of normalizing.

What we see for low shooting training are lots of shots taken, but few converted. A maximum is then reached followed by a high scoring percentage but low shot count resulting in a drop off. In this data set the optimum ratio was 30:40:15:15 (Off:Shot:Sec1:Sec2). However, the gain is quite slow above training shooting to 80%.

The data clearly suggests training shooting to 133% will score you the most goals from a single player. However, a test like this can be quite flawed as previously discussed. It is such a complex situation that you can't really bring all components into play; something always is left out. My main concern is, by training all players to 133% shooting, that you'll actually be hurting your offensive production due to weakening your lineup. There is clearly a sharp increase from 50% to 100% where the derivative starts to diminish noticeably. This point, the 80-90% shooting training mark, is where I'd recommend training your players. If they have the Qs train above it, sure, but I would recommend against training shooting below 80% of primary if you can avoid it. To emphasize this point, let's say the maximum you can reach is 10 goals in 60 minutes (2.5/game at 25% ice time, which is quite high). Now, if you train shooting at 50% you'll net 6.1/game. Training 60% will net 7.1 (+1.0), 70% will net 8.0 (+0.9), 80% will net 8.7 (+0.7), 90% will net 9.2 (+0.5), 100% will net 9.6 (+0.4), 110% will net 9.8 (+0.2) etc. You can see the gain tails off quickly and training in the 80% nets 92% of the maximum number of goals, which is quite sufficient without the added investment of training shooting in excess of 120%. You can, however, also set a few guys in the 100-110% range if their Qs allow it and use them as shootout specialists, I've had good luck doing this in the past prior to selling said player(s).

Take home message:
Everyone's results will be different but I think the general message will remain the same. Don't neglect training shooting. It is an easy thing to let slide but you can notice a difference if you invest the time into training it. If you don't have it already, I recommend getting a ProPack. For one, you're supporting the game we all love, but you'll also have autopractice at your disposal. You can set all your forwards to 10-8-5-5 and never have to worry about changing practices manually again.

Thanks for reading. Please, please, PLEASE, keep in mind the caveats associated with this publication. This is not meant to be an "80% is perfect!" statement, but rather "I wouldn't go below 70 or 80%" statement. Thanks for reading and I hope you enjoyed my little distraction from work I'm paid to do. Now, to watch my win totals drop due to sharing this information with my opponents.