Of Wins and Whining
June 16th, 2011 by Eric Kesselman in UncategorizedAt the time of this writng the season is 40.7% done. You don't do that kind of math when things are going well for your fantasy team, and this time is no exception. I spend about as much time trying to decide if my position is better characterized as 'dire' or 'embarassing' , as I do trying to figure out how to improve.
I'm not sure where things went so far astray. I certainly ran into the Brian Roberts Sandtrap ™, and I've had far more guys underperforming than overperforming (or even performing), and had my share of injuries. Mostly I feel I've been unlucky in Wins and Avg. So, I started playing around with some data on CBS. This isn't solely to vindicate my belief that I've been unlucky (although thats always a comforting part of the process) but its also to find a weaknesses in my approach. Like in poker, you can't dismiss substantial results as running bad automatically, you've got to do some poking around and see if something more substantial might be wrong with your game. I'm hoping either I'll convince myself I've been unlucky, or learn something in the process.
Here's what I started with:
| NAME | GAMES STARTED | QUALITY STARTS | % of starts which are quality | wins | wins/starts | wins/qualitystarts | ||
| Paul/Jason | 105 | 61 | 58% | 39 | 37% | 64% | ||
| Larry | 93 | 53 | 57% | 40 | 43% | 75% | ||
| Wiggins/Hastings | 74 | 51 | 69% | 33 | 45% | 65% | ||
| Derek | 78 | 50 | 64% | 26 | 33% | 52% | ||
| Scott | 74 | 43 | 58% | 36 | 49% | 84% | ||
| Chris | 79 | 41 | 52% | 29 | 37% | 71% | ||
| Gimps | 70 | 39 | 56% | 24 | 34% | 62% | ||
| Peter | 69 | 38 | 55% | 32 | 46% | 84% | ||
| Dalton | 67 | 35 | 52% | 35 | 52% | 100% | ||
| Trevor | 68 | 33 | 49% | 31 | 46% | 94% | ||
| Clark | 46 | 26 | 57% | 34 | 74% | 131% | ||
| Shawn | 55 | 23 | 42% | 24 | 44% | 104% |
In total the league has started 878 games, of which 493 (56%) have been quality. There have been 383 wins which makes on average for .44 wins per start, or .78 wins per quality start. I didn't do a ton of poking around, but those appear to be similar numbers using 2010 data too.
My Gimps have gotten only .34 wins per start and .62 wins per quality start. I'm pretty squarely in the middle of the pack with 70 total starts, and only Larry and Paul/Jason have really stood out in that department. I'm right at league average with 56% of my starts being quality. Maybe reliever quality matters more for wins than I suspect it does, but I've had pretty good relievers this year too in Rzepzinski and D. Bard. I have a hard time believing Balfour and Downs' wins vs Bard's are anything other than variance.
So, It looks like I've been pretty substantially unlucky here, while some teams have been really fortunate. Dalton, Clark, and Shawn are all averaging over 1 win per quality start. Scott has 4 more starts than me, and 4 more quality starts than me, and has twelve more wins. Dalton actually has fewer starts, fewer quality starts, and eleven more wins! Clark has 34 wins off a mere 46 starts and 26 quality starts (almost all of them from two starters Jered Weaver and Scherzer). Sickening! Also, does this mean that Shawn's last place pitching (13.5 points!) has on some level been overperforming?
I must also pause over a disturbing find in the data. Derek has been even more unlucky than I have! He's getting only .33 wins per start and .52 wins per Quality start. He has eight more starts than me and only two more wins, despite 64% of his starts being quality.
I'm only 10 wins behind a massively clustered wins pack. But with my team bleeding everywhere, is this the category to chase? And if so how do I buy more starting pitching when my offense has SO many holes in it, and I'm already punting saves?
Return to me Jake Peavy! You are sorely needed ………
Loading ...
There is clearly a negative correlation between games started and wins/quality start. All of the high percentages have few games started. I am the most extreme case, with 9 less starts than the next fewest. Scherzer has been lucky in the wins department for me, but I also have gotten several from my middle relievers.
Wow, so I'm clearly getting hosed. Second best in the league in QS and worst in W/GS. Pratically tied for third in GS but third to last in wins. Yikes.
The assumption that wins and quality starts are correlated is unproven. In general, starting pitching receives a win in a mlb game about 45% of the time, with a teams starters generally accumulating between 60 and 70 wins per team (although Harrah's in Vegas offerred an even money bet that the Phillies "Phab Phour" would total 80, a bet I promptly booked against for everything I had on me at the time). One assumes pitchers with more quality starts have more wins than pitchers with fewer quality starts, but then Ivan Nova stumbles into the conversation and Jeff Francis starts to laugh.
Our data segments niceley into 4 teams with a per start win rate below the league average, scoring about 35% (meaning we four have generally either unlucky or sucky pitching, or in Shawn's case both), 6 teams with win rates at about league average of 45%, and two teams with significantly higher than average win rates (52 and 74, with Clark at 74 either being the luckiest or smartest pitching picker on the planet). Once again, our wins distribution actually describes a pretty nice bell curve with a slight weighting towards non-wins as we squeeze maximum starts out of the Penny-wise pitching foolish starts of sub-marginal guys who flame into Nova's or end up burning like dry Chatwood.
As a league, our wins distribution is essentially exactly what it should be. Who is where on the bell curve of winning percentage per start can be attributed to (a) starting pitching variance, (b) who your pitcher pitches for (Nova wins as often as Bedard, go figure), and (c) middle reliever variance.
Chasing wins is like chasing the moon over Miami, the best part is giving up and heading to the bar. Wins are essentially a function of total starts. You send as many donkeys out to the mound as you can every week and a hope a few come back with baskets full of gold instead of hits and losses. The vast, essentially overwhelming majority of pitchers will win between 40% and 50% of their starts over the course of a season (are you listening Jered Weaver? May you be the American Leagues' Ubaldo Jiminez this year. Or at least mortal, for pity sake).
Their are simply not enough pitchers in a 12 team league to stream or cherry pick. So wins will be determined mostly by total starts. Clark has a better chance of spontaneously combusting than maintaining a 74% wins per start rate, and I offer the fearless prediction that as the sample size increases everyone will move closer to league average in win per start percentage and that the guys like Larry and Jason/Paul with higher start totals will likely dominate the category.
I also fearlessly predict I will finish third in total starts and hope for at least that score in wins. I have a theory that Yankees pitchers win more than Kansas City Pitchers (I call it "The Obvious Theorem of Wins"). I will test it this year, Phil "I blew a Fa" Hughes willing.
"The assumption that wins and quality starts are correlated is unproven."
I just checked, Chris, and since 1993 among pitchers with at least 15 starts in a year and no more than 3 relief appearances, there is an 0.54 correlation between QS/GS and W/GS.
I agree that a team who leads total GS by a large margin will also likely lead W and that the pitcher's supporting offense is more important, but the quality of those starts is also extremely important.
Derek, well, duh, Dude. Guys that give up three runs or less receive wins more than guys that give up four or more. In MLB, 14 pitchers have 11 plus QS season to date, a total of 163 QS between them (or slightly less than 12 each, or roughly 88% QS out of their total starts). They have won 101 games. This is a winning percentage of 61% of QS and well in excess of 50% of total starts. So yeah, guys with quality starts win more than guys without them.
On the other hand, there are only 15 pitchers who have 8 plus wins in MLB baseball. Only 7 of those 15 have 11 or more QS. (but all of them have at least 8 quality starts, or roughly a QS rate of 75%). Only three of them (Corriea, Arrieta and Tomlin) play on teams that have little chance to win their division.
Clearly if you are smart like Clark, you want guys with high QS percentages. Or you can be smart and lucky like Larry and spike a Vargas, a Fister, a Blackburn and an Ogando and so on and run the pitching table. But in terms of the wins category, you can certainly make the argument that from a management standpoint you are better off looking for wins from semi-decent pitchers on division contenders than excellent ratio/qs pitchers on sucky teams (Vargas and Fister notwithstanding).
Obviously, target pitching is always high qs pitchers on teams that win a lot. But once you chew through that first tier of Sabathias, Verandlers and Lackeys, er Lesters, if you want above league average in wins per start, you are better off chasing the 80% of top tier winning pitchers that pitch for contenders, as opposed to the 20% of top tier tier winners who are better or luckier than their contemporaries.
So let me revise my statement: if you filter out elite pitching, you are more likely to add wins to your total with adequate pitchers on contending teams than you are adding pitchers who make a higher percentage of quality starts for non-contending teams.
Now the picture is clear to me, I need more starting pitching. Maybe the underperforming Gimps starting pitching would blossom on my team.
I have no doubt that it would, however you refuse to ever buy any!
Shawn's fear of trading comes from a deep seated belief that only hitting actually counts for year end standings.
my team is now 24-45 with a 3.70 ERA