מגזין פאוורפליי

Last season, in preparation for my inevitable promotion to I.1, I took it upon myself to acquire a I.1-ready goalie. Given the composition of my team (young, high-Q, low-Exp) I thought it would be best to have a strong goaltender to keep me in games since my forwards likely won't be a threat for another season or two. I spent the 9-figures on Mateus Pernica and started playing him right away.

My goal was to get his chemistry up to 100 prior to the start of the season in I.1. This happened sooner than I expected, taking only 29 days. Now, keep in mind, the rate of chemistry gain has since been changed and in order to get the same 29-day rate you'd need to play normal importance for all games, including friendlies. The plan appears to have worked, as I sit atop the standings for much of the first quarter of the season and only recently began a bit of a slide while I rest my players and send my rookies out for some added experience in games I don't think I will win (an article on that coming later). I would say the goalie plan was a success. I have one of the lowest scoring rates in the league (as I expected) while (excluding the games I strategically rested my top-27 players) my goals against is at a league-low. But, I didn't start this article to boast about my new acquisition and my standings. No, I started writing this article to share with you some findings about long-contemplated and often-ignored values when setting up your lines.

You see, during this 29-day climb from 0 to 100 chemistry, I kept track of all Pernica's vital stats: energy, experience, build and most importantly, chemistry, in addition to the rating he received each game. I've sat on the data for some time since I wasn't able to get it to converge. Maybe it is the air at 38,000' over the San Rafael Desert as I type this on my way home from Los Angeles, but something clicked in my head when I looked at my numbers. And wouldn't you know, I got, for the most part, convergence.

For those of you who do not want to read on further, let me summarize my findings now. Chemistry adds a 25% bonus and experience adds a 20% bonus. These bonuses are added to the energy-modified player build. For those of you who want to see the numbers and equations on how this works to test it on your own, read on.

I took the raw data collected and assumed the following equation:

Score = min(Goa,2*Pas,2*Tec)*(Energy/100)*(1+C_bonus+E_bonus).

Note, the chemistry and experience bonus is added together on the energy-adjusted player build. These bonuses were tested as being on top of each other (i.e. chem. bonus takes into account energy-and-experience-modified rating) without convergence. The bonus factors were calculated by:

C_bonus = C_factor*Chemistry/100
E_bonus = E_factor*Experience/100.

I then varied the C_factor term and E_factor term at will and looked at the Score/Goa_TS. (Goa_TS is the goaltending team strength displayed and is controlled strictly by the starting goalie provided he plays the full game. It is calculated using the energy and attributes he enters the game with.) By looking at Score/Goa_TS (hereafter SGTS), if by varying C_factor and E_factor I could get the SGTS to converge, that will give a strong indication that the two factor terms are accurate since we would have a means to consistently convert things we know about the goalie into a game-ready strength.

I first tried 10% for each, this gave a SGTS of 3.2-3.6, hardly a convergence. 20% each was better giving a range of 3.9-4.1 but still not perfect. I then went to try 25% each and when setting chemistry to 25% and having not changed experience from 20% to 25% I noticed something: the SGTS now only varied from 4.05-4.10. In an ideal world there would be no variance in this number. However, there will be rounding error associated with the team strength observed being of strictly integer values and this 0.05 range is around 1% of the SGTS value - basically negligible.

I then took the average SGTS factor (4.06) from the 29 data points and used it to reverse engineer the Goa_TS we would expect to see given the goalie's build, energy, experience and chemistry. Rounding the outcome values to the nearest whole digit (as that's what we'd see in the game summary etc.) the result was remarkable: 21 of 29 matched exactly. Of the remaining 8, all but one were off by less than half a percent with the lone anomaly being the initial 0 chemistry data point - likely the result of an error on my part. You can see the full dataset below for your perusal in addition to a plot of Projected/Actual against chemistry. Please note the scale of the y-axis as we are looking at sub-percent differences between observed and projected.

As you can see the numbers are in fantastic agreement.  The 0-chemistry point is by far the worst and even then it is only off by 1%.

I encourage you all to check your own goalies. Perhaps these numbers only work for mine and the impact of chemistry and/or experience may not be linear as I have assumed. In either case, this at least gives an order-of-magnitude guideline on how to value players of your own.  To test for yourself, use the following equation:

Goa_TS = 0.246*min(Goa,2*Pas,2*Tec)*(Ene/100)*(1 + 0.25*Che/100 + 0.2*Exp/100).

I cannot emphasise how important chemistry and experience are. I urge you not to undervalue it on your own teams. I have noticed experience even more important for big games, but I have no data to back that up. I hope you found this article helpful and thanks for reading. I now need to figure out how to kill the remaining 2.5hrs of my flight with a screaming baby behind me. I am sitting by the emergency exit. Perhaps I can overcome the massive pressure gradient, open it, and toss the sanity-destroying, O₂-to-CO₂ converter out the door. Or maybe not, I suspect that is slightly frowned upon. Well, in any case, Happy Easter everyone. May you not confuse chocolate rabbit eggs with not-chocolate rabbit droppings. And if you do, just play it cool and ask yourself how rabbits, a mammal, are able to lay eggs - you got me on that one.

-Scott

Edit: Now that I have WiFi I can check my other goalies. It appears it isn't so close for my youngest goalie. I'll combine the data discussed here with data NOW from my other goalie(s) and see what works across the whole spectrum. I will publish an updated article in the future. In the meantime, take the above with a grain of salt.