Well, aren't the top 10 teams in a 450 team universe also outliers?

And does it really make sense that when you are in that group, the index can cause a 7-8 point swing? OK, so having a perfect 120 points is a bit extreme. How 'bout 114? Leading every category in absolute stats is unrealistic, but scoring in the top 3-5% of 7-8 categories (or more) might not be. Yeah, I made the most extreme argument I could to make the point as clear as it could be, but the truth is that I suspect that in many years, real-world results won't be all that far off.

But you're right. There will always be a certain amount of luck in fantasy sports. Injuries, freaks, slumps... who would have expected the results we're getting from Ryan Zimmerman, Adam Dunn or, for that matter, Curtis Granderson or Ryan Vogelsong? But if there might be a way to reduce the impact of luck (rather than add to it), I think it makes sense to try to find it. After all, that's the reason we index in the first place. (If we didn't index, then I'd benefit from the extreme fortune of competing in a much less competitive league than the one you are in).

In short, I'm just saying that when you get to the top of the standings, a 7 or 8 point index swing is a little too unbalanced. It's the equivalent of 2/3 of a category in free points.

Just wanted to point out something that may be getting lost in this discussion: it's an incredible season you're having. My league, Cracking The Whip, has only a 99 index, and I've been working hard all season just to get up to 80 points (not that you should consider it a major achievement to be better than me).

So, congratulations and good luck the rest of the way.

Hey, thanks Vin...

I'm already looking forward to next season. This is the only league I've ever tried this draft strategy, and I'm not sure if I hit the lottery or if it is something that is repeatable.

In one sense, I know I was lucky: Buster Posey was the only of my draft picks to go down this year with a significant injury. If you're gonna take a hit, it may as well be at catcher where the loss is minimal and replacements are cheap.

I also had some good luck with trades. Smoak started out on fire, so shipping him off with Stubbs (in exchange for Weaver, who had just hit his only speed-bump of the season) ended up being a pretty good deal for me. And I'm not sure if anyone thought Verlander was going to be as good as he's been I gave up Valverde and Walden to get him.

But even without those trades, it just makes sense to me that by stocking up on big bats and cheap closers, you give your team a great chance of doing well in at least 8 categories (all except K's and W's). And once you've got your saves locked down, people are always looking for closers and you can use them to as trade fodder to fill any weak spots on your team. Add in the Vogelsongs and Luebke's that emerge every year, other "...co" streamers, and the vulturing you can do when impatient owners drop struggling SPs that usually return to form...

I dunno. Maybe I'm wrong. I just can't wait to try this again next year.

Thanks again, Vin. And in case nobody has told you lately, your contributions have made the league exponentially more fun to be part of. I hope you're making the big-bucks for your efforts

This is a deceptively interesting topic and I've been trying to break it down. Hope you don't mind my chiming in late.

What are some of the underlying assumptions associated with indexing? There's the concept that it should be more difficult to acheive a higher score in a more competitive league than in an less competetive league and therefore there should be a reward multiplier for the more competetive league. Then there's the matter of whether or not aggregated league stats represent a valuable measure of competition.

When thinking about the first assumption, I feel like we need to look at whether or not higher scores can be acheived in less competetive leagues. If this is not the case, then we are rewarding teams in higher indexed leagues despite the fact that their achievement of a high score in those leagues is no more difficult than getting a high score in a less competitive league. Looking at the top 40 raw (that is, unindexed) scores and the bottom 40 raw scores across all leagues you find that the average league index for the top 40 teams is 99.973 and the average league index for the bottom 40 teams is 99.806. It doesn’t really appear that league competitiveness has much influence as to the high and low score distribution.

I am pretty sure that the outliers at the top and bottom are not really influencing those numbers either because of the fact that the indexes are based on a league composite. The index doesn’t really know whether the result came from a league composed of 6 good teams and 6 bad teams, 12 equally matched teams, or (most frequently) a more mixed combination. So while the index may speak to the overall health of the league, it does not really tell you much about the relative difficulty of becoming an outlier within that league, which is particularly important if you are trying to determine an overall champion.

Which kind of brings it back to the question of what the competitive index is really measuring. I think the basic premise is that it measures whether or not the best fantasy players are consistently owned in a league throughout the season. My guess is that all the best players are owned in all leagues and were probably rostered for about the same length of time in every league. I bet that the league indexes are most influenced by the teams at the middle to bottom of the standings continuing to make incremental changes to their rosters over the course of the season, shuffling marginal players and adding marginally to the league totals, and thus the index.

The question is, how much does the increased participation in the FA pool by the lesser teams add to the difficulty of achieving a high raw score? That’s really the big question; I posit that it only has a very marginal influence. Again, I don’t think it’s as if there are lots of very good players sitting on the waiver wire in any of the leagues. If that was the case, then I think we would see more high raw scores from the low index leagues.

A couple of last thoughts. Trades can be influential and they are not tracked by the index as most of the trades consist of players who exist within the composite, so their move from one team to another affects the competitive balance but not the index. Similarly, players dropped to waivers and immediately picked up make a big difference too and again, virtually all the dropped and reacquired player stats exist within the composite but affect the team by team performance while not influencing the index. One last little bit is I wonder if being in a more competitive league doesn’t actually slightly increase the chances of producing an outlier (or champion). Perhaps the required vigilance and the competition produces teams that are more highly incented to maximize performance and squeeze every last ounce of juice out of the orange.

In summary, I really don’t think including an indexed score in determining the overall RCL champion works in it’s current form.

Interesting discussion all.

We're open to refining the index for next year but I do think there needs to be one. The top three raw scores are 114, 111.5, and 109.5. All three of these scores are insane. I think 100 is maybe the highest I've ever seen in a 12-team league I've been part of.

But to just credit the team with 114 without any context to how competitive the league is just isn't fair.

Oaktown Steve's points are solid:

The fact that the index range is rather narrow (97-104) is a sign that there isn't a huge difference between leagues in terms of FAs. My assumption (which isn't going to be universally true) is that higher-scoring leagues are indicative of active owners picking up the right FAs sooner than lower-scoring leagues. This makes it tougher for each team to succeed because the competition is tougher and it's harder to improve your team.

With Vin's help, we have a lot of data we can make available to help propose refinements to the index. One area I'm not confident on is the relevance of the bottom 1-3 teams. Do most of these people check out? If so, perhaps the points should be based on the top 9-11 teams.

If anyone can come up with a better index and prove it with the available data set (and if Vin hasn't downloaded the data, all leagues are public so one can go pull supplemental data), we'll update the index next year and credit the inventor within the league invite post + the standings page.

Thanks,
Rudy

Like I've said repeatedly, I'm inclined to agree.

But... After looking at some stats, I think I've been really, really lucky independent of league competitiveness. My draft and a coupla trades have done much of the work. FA and waiver claims, not so much. At first, I thought maybe the teams running out a bunch of "dead" roster spots may have given me a bunch of cheap points, but again, a look at the stats have changed my mind. Truth is, if everyone wins a lottery once in a lifetime, this was it for me. Check it out:

I ran some numbers this morning from the latest stat update. Of the 450+ teams in the league, I was 9th in runs and HRs, 26 in RBI, 30 in SB, and 51 in average. I'm behind the pace on starts used, but still 135th in K's, 21st in wins, 1st in saves, 3rd in ERA and 34th in WHIP.

So far (through today), my team has about 5500 AB. about 65% of them come from players I drafted. Another 20% come from players I picked up through trades, to replace a DL'd player, or before the end of May (when I think most of the teams in our league were still active).

Pitching is more difficult to quantify, but the only closer I grabbed aside from those I drafted was Hernandez (he got me maybe 4 saves before I dropped him for the extra stream slot). For the most part, my team has been anchored by Ogando, Bumgarner, Weaver, and Verlander. Weaver and Verlander I got through trades in mid-May nad early June, respectively, and Ogando and Bumgarner I grabbed from the wires in April (early May at the latest). Together, these pitchers, along with my closers, are responsible for almost exactly 67% of my IP so far. I also picked up Luebke in early June (twice) and never let him go after he gave me two gems in a row.

Streaming was always going to be a part of this team's plan, and it makes sense that streaming has probably worked out better than it would have if I was in a more competitive league, but... it's not like I've been running CC, Lee or even Cueto out there...

So... If I was in a more competitive league, the only hitting loss we'd be looking at would have come from the 15% of hitting stats that I picked up off the waiver wire. And it's not like my waiver claims have produced any extraordinary results (Altuve has been my best mid-to-late season add).

So far as pitching goes, I think my strategy of dominating the saves category, adding some "extra" K's, and holding ratios down by drafting a roster full of closers, was sound. Like I said, I've probably benefited from a deeper pool of streamers to pick over, but I don't think that whatever benefit I've gained justifies giving up 7-8 index points.

Apologies for making this all about me; that's really not my intent. It's just that I happen to be here, so I'm feeling the impact of the index and trying to figure out whether or not it's as good as it could be.

One question regarding Steve's post: Would it also make sense to look at the top scores in leagues with the highest and lowest indexes? If leagues with the lowest indexes demonstrated a pattern of higher first place scores than the first place scores of teams in more competitive leagues, would that tend toward proving the soundness of indexing?

Isn't it also possible, though, that you drafted a better team because of weaker drafters or had success with trades because of weaker leaguemates?

Look, I don't want to take anything away from a 114 point season. That's bananas. You don't have to be modest about it (re: luck). Of course you need some good breaks but still that's awesome.



It could be looked at...but the index isn't just there for first place teams. i suppose it can be expanded to look at the points for the top 2, 3, 4, 5, and 6 teams to see if there's any patterns.

I've given it a bunch of thought over the last year or so and can't think up a better index than we have today. Every other element I considered (like only including the top 10 teams or looking at transaction levels) has as much if not more flaws than advantages. It's certainly possible that we weigh down its impact but hard to justify why that would be more valid....

Nice post MStark. Without a doubt, luck is one of the preeminent factors in fantasy baseball. However, that being said, I think the quest for indexing is the quest to map those components where there is skill involved (drafting, acquiring through FA or trade, streaming, etc.) to something measurable which can then be normalized across multiple leagues. It's quite a tricky little problem.

I did take a look at tops scores in the highest index leagues and the lowest and the lower did have higher raw 1st place totals. The top index leaders average 95.6, the bottom average 97.4. However there is one really interesting outlier in the top indexed leagues and that is RCL 34 where the leading team, The BallBusters, has just 81 points. What's interesting in this league is that the bottom team has 52 points and the standard deviation across the 12 teams is 10.34. By contrast the standard deviation average for all other leagues is in the 18 range. RCL 34's teams are almost twice as tightly bunched as in other leagues. Intuition kind of tells you that this is a very competitive league. So as where Rudy notes that the dispersion of indexes currently used is not that broad, there does seem to be a more broad spectrum in terms of the deviation. Also, if you take that outlier from the top 10 most difficult league indexes, you get average first place scores very close together at both the top and bottom of the index spectrum.

So one of the things that's tricky here is being careful not to conflate cause and effect. For instance, I wanted to think about including stuff like how many teams autodrafted, and what were the transaction levels as well, but these things become quite difficult to quantify in a meaningful way. I think you have to stick with the results rather than digging into the causes.

I think if I were going to start to refine this process here's what I would do. I'd keep the current index system as a measure of, call it, total player ownership. I'd add a standard deviation factor as a measure of player distribution (which I think speaks to the league competitiveness top to bottom and factors in for the effects of weak sisters). I would use those modification factors to adjust the raw scores and get a ranking similar to what's done now but with an additional component. Then I would take the rankings and create new leagues by groups of 12 (1-12, 13-24, and so on) and I would take the actual stats for each team and score it like a regular league. In other words, score these redistricted leagues by ranking the actual unadjusted stats each of the 10 categories and arriving at a new total score.

Just a first take.

I want to rethink that last part about putting them into 'redistricted' leagues. I don't think it will work. The problem I was trying to solve with it is that I think creating a multiplier or valuation factor based on index and deviation is a little bit arbitray. How do you create a valuation factor that modifies the raw score in the correct measure? Tough one.