What's more important? (MLB)

[Spence]

What carries more weight early on in a season. The season stats or career stats?

Do you expect numbers to shift towards the career's?

Or do you look at the performance of last week and play off that, or better yet, what % would you give towards the importance of each?

[KUJustin]

Last week? 0%. But some people know more about this stuff than me so maybe there is some correlation that I'm not aware of.

I am actually interested to see how people adjust expectations over the course of a season. I mean if a guy is in a slump for 80 games but there's no injury or other external factor to point to do you assume he'll go back to doing what he's always done or do you assume that he's regressed and adjust projections accordingly?

[Moneyline]

[ QUOTE ]

What carries more weight early on in a season. The season stats or career stats?

[/ QUOTE ]

Neither.

Handicapping using in-season stats is a recipe for bankroll suicide. Not only are you subjecting yourself to being tricked by sample size, but since much of the betting public handicaps this way to some extent, the lines will often be inflated against you. As an example, lets say a 33 year old starting pitcher who has always been marginal to crappy is 4-0 in 4 starts with a shutout in his most recent game. Not only is this scrub pitcher unlikely to keep up this string of good performances for very long, but people will bet him thinking he is "hot" or has suddenly become a drastically better pitcher at an age when nobody becomes a drastically better pitcher, thus making the price on this pitcher worse than it should be.

Using career stats can produce problems if you don't account for a player's age. A player entering his prime will likely outperform his career stats. A player entering his mid 30s will likely perform worse than he did in his prime.

I think the best way to approach this problem is to look at advanced (meaning sabermetric) career statistics in concert with a player projection system like PECOTA from Baseball Prospectus (although there are other good ones as well). As the season progresses you can use in season stats combined with a knowledge of aging patterns to chart the player's deviation from his projections, and thereby get a good guess at how much this deviation is for real and how much of it is due to luck.

When this question was posed last year at this time of the season one "respected" regular of this forum (who I don't think is a winning capper) was vehement in disagreeing with this. That said, I could be wrong and he could be right, or we both could be wrong. Also, there's more than one way to skin a cat. Hope this makes sense.

[TomG]

My thoughts FWIW...

For starting pitcher, you'll need to go back through the last few seasons. Even near the end of the season, sample size is way too much to handicap starting pitcher based upon the current season.

The importance of using previous years stats for other positions such as hitting or even bullpen relief goes down in important since you'll have a decent sample into the season.

[newb411breaker19]

I don't know much at all about statistics and how a player regresses but it is obvious that it will always occur. For the offenses i like to look at the season to date work, specifically home or road so when I make my offensive power ratings I rank the teams home and road numbers individually.

For the bullpens I like to use the overall era, but I use Xera which is supposed to be more accurate than regular ERA.. I use XERA for pitchers as well.. I use a weighted average where I give a players last season and current season to date ERA a rating of 5 and a players ERA in the last 3 games a 3. Some would say this too much weight on current performance but so far it has worked for me.

I think this is an important question as I'm sure there is a much more efficient way to do this than I do. But for the hitters I would say that last years stats have less influence on today than the previous years stats on the starting pitchers