Recent Entries
 | May the 'Best' Team Win Posted by |
May the 'Best' Team Win
Al let the mouse into the cheese factory. There's this little information trove out there called the "MSWL United Database." It's a very nice place to play around in for a while, especially under Read-Only privileges so I have no fears that my amateur-level SQL querying and subsequent report generating might re-write our collective league history. Why should you care? Perhaps you have questions, as I do, about whether or not the rules and game formulas are actually generating match outcomes in anticipated patterns. If they don't, then what we're doing as managers doesn't really matter. It's all luck. If they do, then improving your team is both possible and highly desirable. While luck will always play a factor in a game where outcomes are determined by a random number generator, the game is more fun if it's possible to minimize the potential for luck to destroy your carefully crafted hopes and dreams for any given game or season.
Thus, these forum posts are born. Periodically, I’m going to post data-driven musings and answers to the more common questions/problems being floated around, including tackling Sweeper usage and subsequent win rate, hardness usage and win rate, hardness injury rates for self vs. others, etc. (Got a really good question for a subsequent topic? Shoot me a note through the site and I’ll see what I can do!) For this first post, I’m focusing on one question only. It's the first question that Al and I both came up with when thinking about what all the data-crunching could answer:
Does the team with more SL on the field usually win?
In finding an answer, I'm looking at total team SL regardless of SL at Gk or Sw, and also considering all Area Total sums which accounts for total GPP, Hardness, PFD, Home Advantage, and Captain's Bonus points (which started in season 14). The data available to me at this time begins from Match 1 of Season 12 (MSWL United only; U2 not included) and runs through the close of Session 3 of Season 14 (Match 11).
Here's what you're probably interested in knowing:
How often does the team with more SL win?
Here are the tallies, broken into ranges where one team has more SL on the field than the opponent:
(Note: below figures are Pre-Match SL, not impacted by cardings/injuries)
+0:
+1-9 SL: 157-75-91, 60.22% win rate, by an average of 1.86 goals
+10-19 SL: 183-52-41, 75.72% win rate by 2.24 goals
+20-29 SL: 118-32-16, 80.72% win rate by 2.95 goals
+30-39 SL: 73-11-3, 90.23% win rate by 3.19 goals
+40 or more SL: 102-3-0, 98.57% win rate by 4.48 goals
Combined: 633-173-151, 75.18% win rate by 2.57 goals
Note: Combined figures exclude the "+0" games since no team has ‘more’ SL. It is interesting to note that of the 22 matches where teams have identical combined SL, only 6 came out as draws, which is less than 1/3 of the time.
The numbers for pre-match combined Area Totals:
+0:
+1-9: 146-79-97, 57.61%, 1.83 goals
+10-19: 149-53-38, 73.13%, 2.06 goals
+20-29: 147-25-15, 85.29%, 2.63 goals
+30-39: 87-13-3, 90.78%, 3.25 goals
+40 or more: 110-6-0, 97.41%, 4.58 goals
Combined: 639-176-153, 75.10%, 2.52 goals
For those curious, here are the similar breakdowns for post-match numbers, reflecting SL and Area Totals after the impact of red cards and injuries has altered the lines. These are the numbers shown in the match reports in the Scoring Chances section of the match reports, and are the ones used to determine attack allocation, Sw stop rate, and Gk save rate.
Post-Match SL:
+0:
+1-9: 165-73-89, 61.62% win rate, by an average of 1.83 goals
+10-19: 167-61-45, 72.34%, 2.20 goals
+20-29: 118-26-12, 83.97%, 2.80 goals
+30-39: 87-11-2, 92.50%, 3.36 goals
+40 or more: 102-5-1, 96.76%, 4.43 goals
Combined: 639-176-149, 75.41%, 2.54 avg. goal differential
Post-Match Area Totals:
+0:
+1-9: 152-79-94, 58.92%, 1.88 goals
+10-19: 148-53-31, 75.22%, 2.14 goals
+20-29: 142-27-13, 85.44%, 2.53 goals
+30-39: 87-15-2, 90.87%, 3.11 goals
+40 or more: 119-3-2, 97.18%, 4.55 goals
Combined: 648-177-142, 76.16%, 2.54 goals
So what can we say for certain based on the above?
For starters (any match involving Steve Turner aside) the team with more SL, or better area totals, wins far more often than not. Which is exactly how it is supposed to be per the program design. For all those times we can all remember where we were totally robbed of a deserved win, those matches where one team was blatantly and undeniably better than the other (SL 20+ better) resulted in a loss for the 'better' team less than 20 times in more than two full seasons. For those counting at home, there's around 416 matches in one season, give or take a few depending on how many FA Cup replays happen. I'd say that things are playing out pretty well with regard to match outcomes reflecting expected results.
So based on what the game results are showing us, three out of every four times the answer to the question of “Why did my team lose that match?” is going to be: because in this one match they were better.
What happened in those rare losses and draws by teams with more than 20 SL or area total points than their competition? Um... I'm not completely sure yet. Still working on that, since it also requires going back into every match's exact figures with regard to Gk strength, Sw or not, OST, lineup configuration, hardness used, home advantage allocation, red cards, injuries, and then all those same things for the other team too. It's going to take a bit to sift through all that and see if there is any discernable pattern to the 'fluke' losses that might not make them not-so-random after all.
Can this be useful to you? Perhaps. At the very least it should remind all managers that focusing on overall team SL development is more important than winning any one match. More SL is going to lead to many more wins down the road, so ignoring the youth development kills the depth, which decreases total SL available across a session's worth of matches. Also, if you think a match is going to be very close, consider what impact GPP or Hardness could have. 10 additional area points can swing the odds of winning as much as 16% depending on where the breaks fall. Admittedly, where and how you use them probably matters more than simply using them. Still, if you’re going to be riding out what you expect to be a close match then the team that gambles more with regard to improving their area totals is going to win more often than not. If you think you're outgunned, Playing for a Draw adds 7 area points, which can reduce your chances of losing significantly. Yes, it also reduces your chances of winning due to fewer shots in return. Still, would you rather have the other team sitting at a 75% expected victory mark, or 60%?
One caveat to that: this first look at the numbers makes no account for future impact from cardings/injuries to a team. Winning one match at high cost could result in multiple losses later on. This only looks at individual stats in every match in isolation from all other matches. So playing 10 Hard and losing a key T11 player two sessions due to injury might win you one match, and lose you four more subsequently. If it’s a one-off Cup match that you are willing to risk it all to win? The numbers say go for it with GPP for all. We'll look into whether or not to play cleats high at a future time.
Have fun out there.