(*) MASTER NOTES: Owning skills

In talking with Jeff Erickson on this week’s edition of the BaseballHQ Radio podcast, we discussed the appropriate use of advanced metrics in fantasy baseball, and the importance of being willing to confirm whether existing certainties are as certain as we think.

To its credit, BaseballHQ.com has always been highly willing to let its researchers and writers challenge the certainty of the status quo. And at BaseballHQ.com, no status has more quo than the mantra: “Once a player displays a skill, he owns it.”

With all of that in mind, I thought I’d take a look. And it turns out that for some skills, anyway, a batter doesn’t so much own it as have a short-term lease.

The three basic skills a player has are:

• Making contact, which BHQ measures as ABs without strikeouts as a percentage of ABs (ct%);
• Drawing walks, which BHQ measures as walks as a percentage of AB-plus-walks (bb%); and
• Power, which BHQ measures with Expected Power Index, a metric based on hard-hit-ball data.

To assess the “ownership” of these metrics, I compared the individual per-season skills metrics of the 143 batters who had at least 200 PA in each of the last three full seasons (2013-15). The batters had to be at least 23 years old and no more than 31, to ease out at least some age-related effects. In particular, I looked for the range of each metric across each player’s three separate seasons.

Contact

Making contact was the most strongly owned skill. The median range of individual players over the three seasons was just 5%, and batters with ranges of 1-4 percentage points accounted for about two-thirds of all batters. One batter, Jedd Gyorko, actually had zero variation, with a 75% ct% in each of the three years; 17 more had 1% variations, meaning they had the same ct% in two of the three years and a one-point difference in the other. Only 15 batters had ranges of eight percentage points or more, with the highest at 11 percentage points (about 16%), by Ike Davis and Aaron Hicks.

Score one for skills ownership!

Walk Rate

A few years ago, I had a study in BaseballHQ.com demonstrating that Walk Rate, contrary to what we had been saying for years, was not a leading indicator for Batting Average (although it did have a loose correlation with power). At any walk rate, the average hitter was at about .250, and we saw high-BA guys with low walk rates, low-BA guys with high walk rates, and everything in between.

So I wasn’t shocked to see that the median range in walk rates was 33 percent, with several players at 100% or higher (that is, a player’s bb% in one of the seasons could be double another year). Troy Tulowitzki had a low walk rate of 7% in one season and a high of 14% in another. Almost half of the hitters had variances of 25% or higher, although I do want to note that walk rates are often pretty low, so a wide range in percentage terms can appear pretty narrow in percentage points. A batter with a 2% low and a 4% high has a 100% variance despite only two percentage points in the variance itself.

Score one for no skills ownership!

Power

As measured by xPX, power also showed some wide variation in individual player ranges. The median individual range was 27%, with two-thirds of batters in the group falling between 10% and 44%.

Like walks, that seems like a pretty generous range for a skill that is “owned” by a batter. Is Brandon Moss a 137 xPX, as he was in 2014? Or is he a 174, as he was in 2013? Or is he perhaps somewhere between, as he was with his 157 in 2015?

Score one for questionable skills ownership.

Discussion

Why should it be that a batter does not actually “own” his key skills, or at least doesn’t own them to the level of consistency we have come to expect, and to which a lot of analysis says the batter will certainly regress?

The first question to consider in this regard is what we mean by “owning” the skill. I just mentioned the idea of consistency, but I deliberately did not put any boundaries on it. Is Moss’ xPX range of 137-174-157 consistent? If not, how much tighter would the range have to be?

Maybe we could say Moss is a consistent high-power source because anything above 130 is high-power. Maybe we could be less concerned about absolute skills and more with relative levels. We could rank all the batters year-over-year in the various skills to see if the batter is consistently in a particular percentage cohort. There is value in being able to say, “Brandon Moss is a top-10% power source” without hanging a specific numerical value on him.

The other issue is that research is now telling us that various metrics stabilize—that is, they become dependable and at least somewhat projectable—after different numbers of events (PAs, ABs, flyballs, etc.) That seems to imply that those metrics could re­stabilize over different periods, which it seems is just the same as saying they bounce around and have wide error bars.

For example, one well-known threshold list says batter strikeout rates stabilize after just 60 PA. So I took a median ct% guy, Mike Moustakas, and looked at his 2013 through 2015 seasons as well as the sequence of 60-PA cycles during those seasons. In 2013, Moustakas lodged an 82% ct% for the season, but his 60-PA cycles ranged from 75% to 90%. In 2014, his ct% was 84% for the season, in a range from 77% to 90%. In 2015, 86% for the season, in a range from 80% to 91%. (And the numbers vary somewhat more if we use rolling 60-PA cycles rather than sequential.)

If the minimum for ct% stability is those 60-PA runs, what is Moustakas’ true ct%? About all we can say is that it is in the range of 75% to 91%.

Conclusion

In that BaseballHQ Radio interview with Jeff Erickson, I asked him how fantasy owners should employ advanced metrics to manage their rosters. His advice was that we should think about what they mean and be ready and, more importantly, willing to challenge what we think and what we’re told.

Sounds like the right answer to me, too. In particular, if you hear or read an expert saying that a certain player, whether batter or pitcher, will regress to some established statistical level, be skeptical. It might be likely or even measurably probable that the player will regress, but it is not a sure thing.

More From Archives