Which is better ? for MLB

[trevlyn1983]

Do you think using stats from past 7 days or from last month or from the whole season is better when trying to predict outcomes ?

for Offense ?
for Defense ?
Pitching ?
Bullpen ?

Thanks .

[DiggityDaggityDo]

Excellent question. This should be an informative thread... hopefully.

[illfuuptn]

a combination imo. Do I know it? No.

[mebaran]

I'm not exactly sure when in-season stats become useful, but I would shy away from using 7-day or 1-month stats. Think about it: If a hitter goes 12-20 with 5 walks last week, you can almost positively bet that he won't repeat that the following week. 25 plate appearances is SUCH a small sample size.

[bztips]

I use full year-to-date stats. Otherwise you end up chasing your tail IMO.

[trevlyn1983]

i figure using 7 days for Offense would help you see which bats are hot and which is not. Of course like you said that really doesnt mean to much since the hot could turn cold etc .

[underthe total]

imo
i use last 3 starts for pitchers and look for value

[durito]

none of the above?

[underthe total]

yes really none of the above, its all about price

[bztips]

yes really none of the above, its all about price

[underthe total]

[bztips]

ok, I'll spell it out for you -- In order to know whether a given price offers a +EV opportunity, you need to have some way of handicapping the game. If you're attacking it using a model or even just eyeballing some stats, the OP's question about the relevant time frame to look at is pretty important.

[Augustus]

all roads lead to Rome

[statnerds]

5 Inning?

Full game?

RL?

Total?

And fukk all that noise anyway. The Market is going to price in everything that is readily available for public consumption. But since I need someone out there doing the handicapping part, attack this project!!!

[ddesmara]

no way

[ColdBeerHere]

I'm using full season stats, and I am going to be adding bull pen stats soon...for the bp though i am thinking about using the last 5 to 7 days in hopes that tells me how tired the bp is....think i will use IP for 5 to 7 days, and whole season for all other stats

[evo34]

Do you think using stats from past 7 days or from last month or from the whole season is better when trying to predict outcomes ? for Offense ? for Defense ? Pitching ? Bullpen ? Thanks .

For starting pitching, you need to use more than current season (use quality pre-season projections adjusted to some extent by this year's peripheral stats). No real need to use defense explicitly in a model (should be lightly factored into your ERA projections for pitchers). By this time of year, you can get away with using YTD peripheral stats for bullpens and for hitting (adjusted for major personnel changes/injuries). There is a never a case where you should use recent (less than full season) stats. You hope that SPs have unusually good/bad last three starts, e.g., so that you can fade the trend by betting on a reversion to true skill.

[Augustus]

How do you import date from MLB?

[ColdBeerHere]

bump....any idea for # of days to go back to see if a bullpen is tired or not? thanks

[evo34]

Pretty subjective, but you prob. only need to concern yourself with guys who pitched 2-3 days in a row, and therefore might not be available. Longer-term fatigue issues may very well exist, but I don't think they are worth assessing unless you are conducting a thorough study.

Here is a nice source for recent bullpen usage:

http://dailybaseballdata.com/cgi-bin/bullpen.pl

bump....any idea for # of days to go back to see if a bullpen is tired or not? thanks

[ColdBeerHere]

Pretty subjective, but you prob. only need to concern yourself with guys who pitched 2-3 days in a row, and therefore might not be available. Longer-term fatigue issues may very well exist, but I don't think they are worth assessing unless you are conducting a thorough study.

Here is a nice source for recent bullpen usage:

http://dailybaseballdata.com/cgi-bin/bullpen.pl

thanks for the link and info.

I decided for now I will do a general cap for the whole bullpen and will go back 3 days...My thought here is that after 3 rested days they are good to go....

But I could see where I look into the pens more...few games on the schedule, playoffs, etc....that link you gave me will help for sure