Doomed to Repeat

Note, this was originally written on 4/21, but wasn't posted.

As I was looking through the discussion regarding the proposed Mauer/Lester/Dukes for Berkman/Holliday trade in the last post, I came across a fun little gem, addressed to me from Rup:

"NOT EVERYTHING in FANTASY. IS THREE YEAR HISTORICAL PROJECTIONS IF THAT WAS THE CASE YOU WOULD WIN THE LEAGUE EVERY YEAR. "

Now, what's funny, is that as much as I win, if you ignore the random use of punctuation/capitalization, and accept Rup's assertion that my strategy is based exclusively on 3-year historical projections, then you sort of have to accept that 3-year historical projections are are phenomenally accurate and consistent measure. This is particularly the case if you make the assertion that everyone else uses a different model for their analysis.

Let's look at order of finish in every season of the UPL, and we'll assign points like we do in Roto. N points for finishing first in a league of N teams, N-1 for finishing 2nd, ... and 1 point for finishing last.

	Max	Thugs	Jabrone	Westy	90 Reds	Phat	Cheese
2001	10.0	9.0	8.0	7.0	10.0	6.3	6.0
2002	10.0	10.0	9.0	8.0	5.0	6.3	5.0
2003	11.0	11.0	10.0	7.0	2.0	6.9	9.0
2004	14.0	14.0	10.0	12.0	7.0	8.8	8.0
2005	12.0	12.0	11.0	10.0	7.0	6.0	6.3
2006	11.0	9.0	8.0	10.0	7.0	11.0	5.0
2007	11.0	10.5	9.0	6.0	10.5	7.0	5.8
2008	12.0	11.0	12.0	4.0	8.0	5.0	3.0
Total	91.0	86.5	77.0	64.0	56.5	57.4	48.2
% Total	1.00	0.951	0.846	0.703	0.621	0.630	0.529
AVG	11.375	10.813	9.625	8.000	7.063	7.169	6.021
STDEV	1.302	1.646	1.408	2.563	2.705	1.892	1.857
Correl	1.000	0.904	0.555	0.471	-0.008	0.194	0.253

Here, we're looking at some results from some of the UPL regulars who have been there from the start (O.N. Thugs, Jabrones, Westy, '90 Reds, Cheeseheads), as well as Phatsnapper (Rup's franchise) who's a more recent addition. You'll note that for Cheeseheads and Phatsnapper, there are some strange results - in the years that they didn't play, I just used their historical average as that year's entry. Also, in 2002, I can't quite remember the order of finish, after me. I recall C-Lauff finishing 2nd, Westy in the upper half, and Greg about middle of the pack. So, those are guesses. In any case, you'll notice something. There are some high correlations between the franchises and the number of teams per season (the max score). This stat speaks to how consistently a team does. If you combine this information with the average score, then you get a nice picture of how consistent a team is, as well as how successful. Now, this was sort of bragging. I already knew how this was going to turn out, since the UPL Bill keeps track of a lot of this stuff for us.

And, one conclusion that you can make is that year-to-year in UPL Baseball, the optimization based on 3-year projections is a pretty solid strategy. Now that isn't quite what I do, but I'd guess that's a big chunk of my evaluation. But more interesting is what happens if you start to look at the correlations between the different teams.

	Max	Thugs	Jabrone	Westy	90 Reds	Phat	Cheese
Max	1	-	-	-	-	-
Thugs	0.904	1	-	-	-	-	-
Jabrone	0.555	0.613	1	-	-	-	-
Westy	0.471	0.440	-0.237	1	-	-	-
90 Reds	-0.008	-0.181	-0.199	-0.185	1	-	-
Phat	0.194	-0.054	-0.557	0.649	-0.068	1	-
Cheese	0.253	0.442	-0.132	0.482	-0.473	0.234	1

If you have a correlation close to zero, that suggests that there is a very different strategy being employed. It's no surprise that the O.N. Thugs (me) have virtually no correlation with Phatsnapper (Rup). What's also interesting is how small the correlation between the O.N. Thugs and the '90 Reds (Greg). And in turn, the '90 Reds have a virtually no correlation with Phatsnapper. So, it's like there are three distinct strategies going on (or at least, three distinctly different sets of outcomes). Interestingly, the '90 Reds and Phatsnapper have very similar track records (though Greg's is over 8 seasons, and Rup's is over 4), each with a ring (2 for Greg).

We've classified the O.N. Thugs as basing value on recent historical performance. Now, if you were to classify the other strategies, I'd label each value something like this:

Phatsnapper - value based on youth and potential upside.
'90 Reds - value based on personal perception.

In any case, it appears that the Jabrones have similar results to the Thugs. Imitation. Flattery. Yada, yada. Westy has high correlations w/ both the Thugs and Phatsnapper. So maybe some hybrid of both strategies? I'm guessing that it's a blend of historical numbers and the upside of Minnesota players doing well. Sadly, that strategy doesn't work.

The closest thing that I can interpret from about Greg's strategy is that it's the opposite of the Cheeseheads. Of course, the Cheeseheads outcomes are based on finishing in the middle of the pack every year, so we can't really learn much about his strategy here. I'm not sure what the opposite of that is - maybe something like feast or famine? But that's more of an outcome, not necessarily a strategy. Basically, my best guess is that Greg's in his own little world, and when his perceptions (think Bret Boone and Paul LoDuca in 2001, not that I'm bitter about it or anything) are all on for a given year, then he does really well. And when his perceptions are off, he finishes in the bottom half of the league.

Obviously, we don't have a lot of historical data - 8 years worth isn't a ton, though if you break this down further, it may be of interest. As I think about it, maybe it's not order of finish, but total points in a given year that's more useful. And an interesting question may be looking at what stats teams value (or at least evaluate well), historically. I can already guess that the O.N. Thugs manage to do well in OBP, SLG, HR, and RBI, a little bit less so in R, and terrible in SB. And with regard to pitching, I can see the O.N. Thugs doing well in SV, W, K, and ERA, and so-so in L and WHIP...

In any case, I think that this idea of looking at the broad, macro level of the UPL is worth revisiting.

-Chairman (aka O.N. Thugs)