At long last, I've completed a comparison of key statistical results when manually controlling a team v autoplaying all teams. The results show a marked difference in results when manually controlling a team. And oddly, the results also significantly differ depending on the computer system utilized to play the games.
Here's what I did. First, I created a league and established target values for individual teams and the league as a whole, for various statistics (SOG / G, G/GP, A/G,Pim/G etc.). I ran about 150 automotic full season replays, tinkering with settings and ratings and other manipulable features to bring the teams, players and overall league within acceptable variance of the established targets. When I was satisfied with what I was seeing, I ran 10 final replays and recorded the results and averaged them. All tracked statistics fell within the actual low and high results seen in the NHL during the period used to create my league (1967/68 through 1977/78). All that was done on a PC running Windows 10.
I then copied my league file and began playing the league on my laptop, which was running a prior version of Windows. I took one team, Cleveland, and manually coached it with the offensive play calling option activated. I manually coached all 60 minutes of every Cleveland game for more than a half season, as a result of which every other computer coached team in the league played at least 40 games. I then recorded the statistical results.
Next, I autoplayed the same number of games for the same league on my laptop and recorded the results. I also completed the autoplayed season on the laptop and recorded those results too.
And lastly, for comparison sake, I took the league from my laptop, transferred it back to my PC, and autoplayed both a half season and then the completed full season, tracking both results.
My findings follow. Note that when manually coaching games, I wasn't overly aggressive in seeking shots. I shot from outside mainly when there was a screen in front of the net. I normally shot from inside or medium range when a player had *** next to him. When given an option to penetrate, I usually took it and then shot if it succeeded. Otherwise, I typically kept passing until I lost possession.
As shown in the tables below, among the notable variances are these. CLE's target SOG/G is 33.4. In both 10 season autoplay average and in single season PC full autoplay, CLE had 32.1 SOG/G. Manually coaching them on laptop for a half season saw 39.9 SOG/G. And quite oddly, laptop full season autoplay saw 34.8 SOG/G. Why laptop v PC should so vary has me scratching my head.
Overall league results also are mystifying when comparing laptop autoplay to PC autoplay. SAG on laptop for whole league were 33.2, while PC was 31.2. PiM/G were 30.7 laptop and 28.6 or 28.3 PC. Goal scoring and related stats also much higher on laptop than PC. Why? Beats me.
I plan to pass this directly on to Dave but I thought others might find my tests interesting.
|Team||Target SOG/G||CLE 1/2 Manual||Laptop 1/2 Auto||PC 1/2 Auto||Laptop full auto||PC full auto||10 yr low||10 yr high||10 yr avg|
|Team||Target SAG/G||CLE 1/2 Manual||Laptop 1/2 Auto||PC 1/2 Auto||Laptop full auto||PC full auto||10 yr low||10 yr high||10 yr avg|
|Lg Avg.||Low||High||Target||CLE 1/2 Manual||Laptop 1/2 Auto||PC 1/2 Auto||Laptop full auto||PC full auto||10 yr low||10 yr high||10 yr avg|