“Don’t chase wins.” This is perhaps the most frequently dispensed piece of advice by fantasy baseball analysts. Everyone from Mike Siano (MLB) to Scott White (CBS) to Eric Mack (SI) says it. Heck, I’m sure I’ve said it at one point or another. Ron Shandler probably puts the notion most succinctly:
Bad strategy. DON'T CHASE WINS. You never succeed. Just make sure you have highskilled pitchers and hope the wins follow. That is the best, and ONLY thing you can control.
It’s easy to see why this is such a popular piece of advice. The table below shows the yeartoyear correlation for a number of pitching stats (20022010, 100 IP min, no relief appearances):
Stat 

0.84 

K/9 
0.80 
BB/9 
0.68 
0.45 

0.34 

W/GS 
0.22 
Clearly, there is a lot more variability in a pitcher’s wins than there is in his other stats, especially his “skills”: strikeouts, walks, and groundballs.
“Don’t chase wins” has become perhaps the most accepted strategy in all of fantasy sports. So why do I bring up the topic today? Because it’s overly simplified and not the optimal course of action. It’s also topical, as there is an ongoing discussion at the website for the CardRunners Fantasy Baseball Experts League (of which I’m a part) and because during my live chat the other day; the first question I received came from reader Andrew from Las Vegas:
Is there any real way to try to improve in the wins category?
The answer is “yes.” There are two ways, actually. I’ll defer the first way to CardRunners leaguemate Chris Hill–a poker player who has teamed up with Hollywood director/writer/actor Nick Cassavetes for the league–who put it eloquently:
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.
A leagueaverage pitcher will take home a ‘W’ in 35 percent of his starts, so the easiest way to accumulate wins is to do just as Chris says: send as many donkeys out there as you can. This is the only surefire way to accumulate a lot of wins. Of course, most of us don’t have the luxury of putting anybody with a pulse into our fantasy rotations due to league settings and/or the need to protect our ERAs and WHIPs.
This brings us to option two: picking pitchers that have the best chance of winning games. While many analysts will preach that the best we can do is to pick pitchers with good skills, this is not really true. A pitcher’s skills play a part in winning games, but not the largest part. The average pitcher only throws 5.9 innings per game. His bullpen takes the remaining innings in the field, and his offense takes the other nine halfinnings. This means that the offense is involved in 50 percent of a game, the bullpen 17 percent, and the pitcher himself 33 percent (ignoring the fact that luck, defense, etc. come into play while the pitcher is on the mound).
Because of this, if all we cared about was wins, we’d actually be better off picking our pitchers based upon their supporting offense than picking them based upon their own skills. Of course, it’s not an either/or scenario, so we can do even better by combining all of these factors to come up with a more accurate expectation of a pitcher’s win total. To do this, Bill James’s Pythagorean Theorem serves as a good tool. The formula is:
This next part will look familiar if you read my piece on Tim Stauffer a couple weeks ago. Today I’ll use Dan Haren—who I happen to own in CardRunners—as my example.
First, we should first come up with how many innings we expect a pitcher to throw per game and what his expected run average (RA) is. Haren averages roughly 6.7 innings per start, and we’ll use 3.35 as his expected run average. Haren’s bullpen’s run average on the year is 3.81 (ideally we’d go through projections for each member of the bullpen individually and put them together, but this should be a good enough ballpark estimate).
Assuming Haren and his bullpen pitch a total of 8.91 innings per game (league average, since the bottom of the ninth doesn’t always happen), their combined expected RA is 3.46. The Angels’ offense has scored 3.67 runs per game this year (again, we could go through individually, but we’ll use this for simplicity’s sake).
When we combine these numbers, we see that in games Haren pitches, the Angels would be expected to win 52.8 percent of the time. From here, we just need to figure out how many times Haren should be expected to get the win in these Angels victories. Here is a table that shows how often the pitcher gets a ‘W’ in games in which 1) his team wins and 2) he allows just two or three runs (Haren’s expectation is 2.5 per 6.7 innings) based on the number of innings he throws:
W% 

5 
65% 
5.1 
62% 
5.2 
68% 
6 
70% 
6.1 
71% 
6.2 
75% 
7 
75% 
7.1 
80% 
7.2 
82% 
8 
85% 
8.1 
89% 
8.2 
91% 
9 
94% 
Based on this chart, we could expect Haren to get the win in roughly 75 percent of his starts that result in an Angels win, for a final win percentage of 39.7 percent of all games started. Over a 32start season, that would give Haren a bit under 13 wins. He’s currently getting a win in just 33.3 percent of his starts, so we can say that Haren has been a bit unlucky in the wins department to this point.
To help you with your own analyses, you can see a complete chart of win percentages based upon the number of innings thrown and the number of runs allowed below.
W% 
SAMPLE 
W% 
SAMPLE 
W% 
SAMPLE 

5 
0 
.809 
236 
6.1 
0 
.857 
84 
7.2 
0 
.925 
106 

5 
1 
.787 
474 
6.1 
1 
.834 
223 
7.2 
1 
.909 
176 

5 
2 
.670 
572 
6.1 
2 
.732 
287 
7.2 
2 
.846 
162 

5 
3 
.626 
604 
6.1 
3 
.690 
242 
7.2 
3 
.787 
108 

5 
4 
.557 
467 
6.1 
4 
.594 
160 
7.2 
4 
.727 
44 

5 
5 
.470 
287 
6.1 
5 
.551 
78 
7.2 
5 
.786 
14 

5 
6 
.444 
117 
6.1 
6 
.320 
25 
7.2 
6 
1.000 
1 

5 
7 
.339 
56 
6.1 
7 
.333 
12 
8 
0 
.909 
628 

5 
8 
.000 
12 
6.1 
8 
.500 
2 
8 
1 
.887 
856 

5 
9 
.500 
4 
6.2 
0 
.928 
111 
8 
2 
.859 
544 

5.1 
0 
.854 
41 
6.2 
1 
.817 
268 
8 
3 
.840 
250 

5.1 
1 
.785 
130 
6.2 
2 
.776 
357 
8 
4 
.774 
84 

5.1 
2 
.654 
188 
6.2 
3 
.722 
299 
8 
5 
.760 
25 

5.1 
3 
.596 
225 
6.2 
4 
.623 
183 
8 
6 
.667 
9 

5.1 
4 
.531 
147 
6.2 
5 
.609 
69 
8.1 
0 
1.000 
26 

5.1 
5 
.475 
122 
6.2 
6 
.524 
21 
8.1 
1 
.930 
57 

5.1 
6 
.447 
47 
6.2 
7 
.250 
4 
8.1 
2 
.868 
38 

5.1 
7 
.263 
19 
6.2 
8 
.500 
2 
8.1 
3 
.917 
36 

5.1 
8 
.200 
5 
6.2 
9 
.000 
1 
8.1 
4 
.818 
11 

5.2 
0 
.776 
58 
6.2 
1 
.500 
2 
8.1 
5 
1.000 
4 

5.2 
1 
.784 
153 
7 
0 
.907 
1226 
8.1 
7 
1.000 
1 

5.2 
2 
.702 
215 
7 
1 
.849 
1934 
8.2 
0 
.875 
16 

5.2 
3 
.657 
248 
7 
2 
.785 
1575 
8.2 
1 
.897 
39 

5.2 
4 
.517 
180 
7 
3 
.703 
990 
8.2 
2 
.906 
32 

5.2 
5 
.440 
125 
7 
4 
.693 
401 
8.2 
3 
.917 
12 

5.2 
6 
.419 
43 
7 
5 
.625 
136 
8.2 
4 
.750 
4 

5.2 
7 
.357 
14 
7 
6 
.794 
34 
8.2 
5 
1.000 
2 

5.2 
8 
.000 
3 
7 
7 
.571 
7 
8.2 
6 
1.000 
1 

5.2 
9 
1.000 
1 
7 
8 
.000 
2 
9 
0 
.985 
716 

6 
0 
.877 
626 
7.1 
0 
.894 
94 
9 
1 
.960 
446 

6 
1 
.789 
1216 
7.1 
1 
.867 
188 
9 
2 
.944 
196 

6 
2 
.722 
1482 
7.1 
2 
.838 
198 
9 
3 
.935 
77 

6 
3 
.663 
1223 
7.1 
3 
.736 
129 
9 
4 
.880 
25 

6 
4 
.554 
724 
7.1 
4 
.702 
67 
9 
5 
.800 
5 

6 
5 
.506 
340 
7.1 
5 
.773 
22 
9 
6 
1.000 
1 

6 
6 
.514 
109 
7.1 
6 
.500 
6 
10 
0 
1.000 
2 

6 
7 
.565 
23 
7.1 
8 
.000 
1 
10 
1 
.500 
2 

6 
8 
.000 
4 
I’m not saying that you should forget everything you’ve learned to this point and gun solely for wins; I’m just saying that wins aren’t a complete crapshoot as some suggest. Combine this type of analysis with your expectations for the other categories and you’ll be better off in the long run. Or I’m sure PECOTA does something similar to arrive at a pitcher’s projected win total, so when looking at a pitcher’s projection, don’t just look at the ERA and call it a day. The wins in the projection line are relevant. As such, you can and should set your fantasy team up to generate more wins. At an absolute minimum, make sure to take the pitcher’s supporting offense into consideration. Sure, there will be variability with wins—more even than with ERA—but if variability is something you can handle, your net expectation is going to be better for it.