Flaws in certain statistics?

[redban]

Baseball is often described as a game of statistics. There are stats for almost everything, and fans use stats to rate players.

But what are some flaws in certain baseball statistics?

For me:

All stats (HR, RBI, BA, SLUG, OPS, OBP) fail to factor consistency. To my knowledge, there is no stat for consistency.

Example:

Player A bats .300 for the entire season. All season long, his BA never dipped below .300. He also hit 30 homeruns, hitting 5 HR in all 6 months of the season. So in all, he finishes with .300 / 30 HR.

Now Player B bats .250 for the first three months of the season, with no homeruns. Then in the next 3 months, he bats .350 and hits 10 HR each month. He finishes with same line as the other guy, .300/30 HR.

Is that really correct for Player B to have the same exact line as Player A? What is more valuable, .300 through 162 games, or .250 through 81 + .350 through 81?

[dgarza]

Example:

Player A bats .300 for the entire season. All season long, his BA never dipped below .300. He also hit 30 homeruns, hitting 5 HR in all 6 months of the season. So in all, he finishes with .300 / 30 HR.

If you mean this literally, then it's impossible. All players start .000 or 1.000.

Now Player B bats .250 for the first three months of the season, with no homeruns. Then in the next 3 months, he bats .350 and hits 10 HR each month. He finishes with same line as the other guy, .300/30 HR.

Is that really correct for Player B to have the same exact line as Player A?

As long as we continue to call it Batting "Average" then we already know that it is an average number, which means it is correct.

What is more valuable, .300 through 162 games, or .250 through 81 + .350 through 81?

We need more context to answer this question.

[9RoyHobbsRF]

What is more valuable, .300 through 162 games, or .250 through 81 + .350 through 81?

which ever one helps win more games

what if he hit 30 homers while batting .250 and 5 homers while hitting .350

what if he faced all #1 starters etc while hitting .250 and all scrubs and rookies batting .350

what if he hit .250 while playing in a pitcher's park and .350 while playing in a hitters park

too many other factors

[leewileyfan]

If you have the figures [box score data, for example] for all the games, you can compare his .250/.350 performances to team W/L results. That will help clarify the extent of impact he had on team performance.

Then too, there are situational hitting stats based on in-game performance, based on the 24 base-outs situation grids. These track win probability impacts of each at bat and accumulate totals over a season.

Baseball-Reference has those.

[Bothrops Atrox]

Right - if two players have the same stat line, but you want to know who "helped win more games" with the bat...just look at Win Probability Added, which contextualizes the stats.

[abolishthedh]

This is a point which I have made on other threads, but OPS has a redundancy in its design. A player goes 1 for 4 with a double and a walk, meaning a slugging % of 0.500 and an OBP of 0.400, which would lead to an OPS of 0.900 for the game. The redundancy which I see all too clearly is that the double counted as part of the SLG % and in the OBP %. One act with the bat has counted twice in his stats. Furthermore, the denominators of both statistics are not the same, which will complicate the addition of the percentages.

Since the stat OPS has been created out of convenience, it seems that no one has called this into question. By pointing out these design flaws, the convenience would evaporate.

My point, in short, is that there isn't a single stat which is one-size-fits all and interpreted appropriately for all players. All stats will have flaws. In other words, to address the topic of the thread, there isn't a flaw in certain statistics, but a flaw in all statistics and therefore we need the full range of statistics to paint the picture for a player or team. Convenience for convenience's sake seems like a waste of time, if I want the full picture.

[leewileyfan]

Being a good citizen, I have removed my errant post, although its content was well intended.

[9RoyHobbsRF]

^ he said 1 for 4 with a walk

not 4 trips to the plate with a walk and a double

you are using the wrong stats

[Bothrops Atrox]

This is a point which I have made on other threads, but OPS has a redundancy in its design. A player goes 1 for 4 with a double and a walk, meaning a slugging % of 0.500 and an OBP of 0.400, which would lead to an OPS of 0.900 for the game. The redundancy which I see all too clearly is that the double counted as part of the SLG % and in the OBP %. One act with the bat has counted twice in his stats. Furthermore, the denominators of both statistics are not the same, which will complicate the addition of the percentages.

Since the stat OPS has been created out of convenience, it seems that no one has called this into question. By pointing out these design flaws, the convenience would evaporate.

My point, in short, is that there isn't a single stat which is one-size-fits all and interpreted appropriately for all players. All stats will have flaws. In other words, to address the topic of the thread, there isn't a flaw in certain statistics, but a flaw in all statistics and therefore we need the full range of statistics to paint the picture for a player or team. Convenience for convenience's sake seems like a waste of time, if I want the full picture.

wOBA takes the idea of OPS and fixes the issues you mentioned. It has been said that OPS is correlates closely to runs created accidentally, mathematically speaking. wOBA has an even closer correlation to runs created and is sound mathematically.