BaseRuns & Projections

by R.J. Anderson on Aug 31, 2008 1:31 PM EDT Comment 7 comments

For more on BaseRuns. What I decided to do was run numbers based on teams baseruns projected record (over 162 games) versus their "real" projected records based on current winning percentage. Since most teams have about 30 games left we probably won't see too big of a swing, but I'm curious if some of these teams regress (or progress) over the next month and some. In fact when the season is over I plan to review the numbers just to see which pair of numbers worked out.

Here are the numbers we need to worry about:

Team	RS	RA	Proj W	Real W	Difference
ANA	560	552	82	98	-16
MIN	603	607	81	91	-10
HOU	577	656	72	82	-10
TBA	595	528	90	99	-9
MIL	626	572	88	94	-6
PIT	579	742	63	69	-6
CIN	559	673	67	71	-4
FLO	590	602	79	83	-4
SLN	677	637	86	89	-3
PHI	630	590	86	89	-3
KCA	515	630	66	68	-2
SFN	509	587	71	72	-1
CHN	710	551	100	101	-1
TEX	720	738	79	80	-1
NYN	638	566	90	89	1
CHA	670	560	94	93	1
NYA	627	574	88	86	2
CLE	599	594	82	80	2
BAL	656	665	80	77	3
WAS	488	634	62	58	4
COL	638	647	80	76	4
DET	668	657	82	78	4
OAK	505	522	79	74	5
BOS	694	544	99	94	5
TOR	564	514	88	83	5
ARI	581	529	88	83	5
LAN	540	517	84	79	5
SEA	545	654	68	61	7
ATL	599	578	84	71	13
SDN	535	581	75	62	13

Obviously some teams have been bitten by the injury bug -- San Diego has been nearly devoured by the damn thing -- which makes their projected records a bit off if (when) they get healthy. A team like Boston for instance, without Mike Lowell and possibly J.D. Drew are probably less likely to match their production thus far this season than say the Chicago Cubs who aren't missing any key components right now. I'm not quite sure what to expect, but which of the two methods do you think will result in more accuracy?

Update: Thankfully Sky pointed out a mistake I made in the original formula, the numbers above are accurate.

Poll

Which will be more accurate?

BaseRuns

9 votes

Winning%

9 votes

18 votes | Poll has closed

0 recs | Comment 7 comments

Story-email Email Printer Print

Comments

i'm confused -- what did you do?

did you compute baseruns for offense and pitching and plug them into a pythag and pro-rate over a 162 game season?

why does it appear nearly every team is underperforming? am i reading the table wrong? is there something in your methodology that would underrate offense compared to defense?

by Sky Kalkman on Aug 28, 2008 6:05 PM EDT reply actions 0 recs

Baseruns for hitting/pitching then plugged into pythag.

The chart’s fourth column is Projected wins – real wins, so the teams with negative numbers are those who are overachieving.

by R.J. Anderson on Aug 28, 2008 6:10 PM EDT up reply actions 0 recs

That was what I was assuming

Which implies that Anaheim, for instance, is being set up for a first-round playoff loss.

Otoh, since the method appears to be consistently underrating playoff contenders (ANA, TBA, MIL, MIN), there may be a large residual that needs to be captured. (Then again, SDN has been damaged by injury; ATL has been damaged by injury and Jeff Francoeur; CIN has overachieved, since Dusty is trying to destroy more young pitchers’s careers. So the differences may be attributable to management styles.)

by klhoughton on Aug 29, 2008 12:05 AM EDT reply actions 0 recs

math issues

any projection system should generate a total of 2430 wins (81*30). The base runs estimate here totals 2302 wins, leading to systematically low win estimates for each team.

by vaclipper on Aug 29, 2008 9:00 AM EDT reply actions 0 recs

I don't see anything in the formulas used that would lead to this.

by R.J. Anderson on Aug 29, 2008 2:13 PM EDT up reply actions 0 recs

some ideas

how do you BaseRuns estimates of total runs scored and total runs allowed match with reality? are you not coming up with enough runs scored or are you coming up with too many runs allowed?

are you not including ROE or something dumb like that?

if you can’t figure out what’s making the BaseRuns estimates funky, i’d recommend modifying the B factors (just multiply B by some constant for either RS or RA) so that runs scored = runs allowed when you sum up all the teams. to be exact, you’d probably want a different B scalar for hitting and pitching, so that each of them totals the actual MLB-wide totals. then run pythag.

by Sky Kalkman on Aug 29, 2008 2:58 PM EDT up reply actions 0 recs

for those interested

the B multiplier was set to about .95 and 1.07 for batters and pitchers respectively so that baseruns predicted 18000 runs total among all teams. why 18000? it was about half way in between the estimated totals for each group.

the eight playoff teams as predicted by baseruns don’t change, but there would be some GREAT races:

Rays, Yankees, Blue Jays all within two games of each other for wild card.
Cardinals, Dodgers, and Phillies all within four games of Brewers for wild card.
Angels three games ahead of both the A’s and Rangers.
NL East and West still just as close.

by Sky Kalkman on Aug 31, 2008 3:24 PM EDT reply actions 0 recs