Page 3 of 4

Re: STRATEGY!!!!

PostPosted: Thu Mar 13, 2008 7:46 pm
by BitterDodgerFan
another thing going against playing pitching matchups is that you never know how long the starter is going to stay in the game. sometimes they get chased early in the game and the guy you benched mashes the bullpen.

Re: STRATEGY!!!!

PostPosted: Fri Mar 14, 2008 12:09 pm
by swyck
Apollo wrote:
This is a much more reasonable strategy in H2H than in Roto, though I'm not in huge favor of it in either format. In Roto, benching Matt Holliday due to a slump runs the risk of missing out on valuable statistics that he's putting up even during his slump vs replacement, as well as stats you miss out on in the time period after he's out of his slump and you insert him back in your roster vs. replacement. Last year, for example, if you started Matt Holliday every single game, and never removed him from your lineup, he added to your Roto scores .340, 36 HR, 137 RBI, 120 R, 11 SB. In my opinion, benching him during his 'slumps' increases the odds at losing out on his stats than adding those of a replacement level player, in the short run.


I disagree with this attitude. First of all, last season's Matt Holliday is a bad example (though I know you weren't the one who brought him up) because he never had a three-week slump. But say for the sake of argument that you know Matt Holliday will hit .340 with 36 homers, 137 RBI, etc. etc. Now, it's mid-July, and he's mired in a two-week long 8-for-52 slump. You have an average guy -- say young Jeremy Hermida, or maybe Raul Ibanez -- sitting on your bench, and he's batting .360 with 3 homers in the last week. I feel like you have to put him in. Just ride him for a few days. As soon as he puts up a couple 0-fers or Holliday hits a home run, you put the stud back in. You miss out on one good game, tops. There's no risk there.

Sure, you can get .340 with 36 homers by never taking Holliday out. That's great. But why settle for that? If you're careful during slumps, maybe you can get .344 with 39 homers. And every little bit helps.

What are your thoughts, bigh0rt?


Very wrong. You're not taking regression to the mean into account here. :-?

If Holliday is in a slump for a week, you've already had that slump in your lineup. If you now bench him, he may come out of it immediately, and you lose those numbers. Same with Hermida's hot streak. He's already had that streak, there is no reason for him to continue it. IMO you are more likely to underperform then to overperform since you are chasing (or avoiding) performance that has already occurred. You have no way of knowing when a slump will end or a new one start. Over the long run (the season) the better player will outperform, period.

You can't time slumps. A lot of people lose money in the stock market thinking they can time things. Same thing.

Re: STRATEGY!!!!

PostPosted: Fri Mar 14, 2008 3:52 pm
by fezzik
Last year I drafted Jason Bay at the beginning of round 3...was it a mistake to leave him in my starting lineup until Sept...I mean...he's a stud, so I had to play him. Right? ;-7

I agree though that in general you have to let your studs play...I almost never bench them (if they're on the DL I may...maybe). But there are times where they look so lost that it needs to be done (ie. Hafner, Bay, etc.). Hopefully they find there way back home to studliness in 2008. B-)

Re: STRATEGY!!!!

PostPosted: Fri Mar 14, 2008 4:36 pm
by Bloody Sox
GotowarMissAgnes wrote:
BitterDodgerFan wrote:how about sitting your non-stud lefty against a tough LHP? like hawpe, should he be benched because of the R/L splits in favor of someone who is facing a more favorable matchup? there are some players that will be benched against lefties often, so it works out. or if you have a wily mo pena facing a lefty or matt stairs facing a righty, then this would be the time to play them in favor of someone like hawpe facing a lefty, no?


Yes, I definitely use L/R matchups for my non-studs. Lefty hitters often drop further than they should in drafts, and that's who I usually round out my bench with. Match-ups based on hot/cold streaks or individual player matchups, however, is a futile exercise, imo.

ABSOLUTELY. You sit Hawpe 100% of the time against lefties. Similarly, even though he can be called a stud since he's going in Round 3-4, I sit Granderson against lefties every time as well. There is a big enough sample size to warrant doing so, and you're only sitting him maybe 20% of the time.

Re: STRATEGY!!!!

PostPosted: Fri Mar 14, 2008 5:33 pm
by Apollo
swyck wrote:Very wrong. You're not taking regression to the mean into account here. :-?


I'm as big a numbers guy as anyone. I believe in regression to the mean, I have an engineering degree, I read firejoemorgan, blahblahblah. But we're not talking about the long-term here. We're talking about single days, and the most basic principle of regression to the mean is that it only works over the long term. That's why "he's due for a hit" is dumb. And, though of course neither of us can prove it, I think you're kidding yourself if you think that immediate past performance has no correlation at all with immediate future performance.

Over the long run (the season) the better player will outperform, period.


This is correct, of course, and almost condescending. But of course I'm not advocating benching your stud for the season. I'm advocating benching him for a couple days if he's gotten on base twice in the last two weeks.

You can't time slumps. A lot of people lose money in the stock market thinking they can time things. Same thing.


This analogy is broken for obvious reasons. It might hold true if, say, I was going to trade away Holliday in the middle of his slump and then pay twice as much to get him back. But the great thing in this case is that you can get your stock back -- for free! -- as soon as it starts going up again. So here's a better analogy: Imagine that you had a historically strong mutual fund that has been losing money for the past six months. Now, say you could bench that fund's production for another mutual fund that is also fairly strong, but which has been making 30 percent interest over the last two quarters. And as soon as the original fund gets back in the black, you can go right back to it with no ramifications! What a deal!

Okay, my analogy sucks too. But it sucks slightly less.

If Holliday is in a slump for a week, you've already had that slump in your lineup. If you now bench him, he may come out of it immediately, and you lose those numbers. Same with Hermida's hot streak. He's already had that streak, there is no reason for him to continue it. IMO you are more likely to underperform then to overperform since you are chasing (or avoiding) performance that has already occurred. You have no way of knowing when a slump will end or a new one start.


I think the major disagreement here is that you think eight consecutive 0-for-4s will have no relation at all to what Matt Holliday will do the next day* -- in other words, that slumps are merely statistical anomalies where the coin comes up heads 6 days in a row. This seems very unlikely to me. Now, there is probably no way of proving either one of us correct or incorrect (unless Bill James has done a study on this already, which would be awesome). So in the absence of data, I'll pose a rhetorical question:

In your heart of hearts, if you have Matt Holliday coming up and the pennant on the line, do you want him to be stuck in a 2-for-26 slump?


*Out of curiosity, I skimmed Holliday's game logs and found that he had back-to-back oh-fers only three times all season, and never once had three in a row. That's insane!

Re: STRATEGY!!!!

PostPosted: Fri Mar 14, 2008 8:57 pm
by tmlfan4ever
I really think you're playing with fire if you tinker with your roster that much. If you like a guy, get him on your team and keep him in the lineup. 10,20,50 ABs is nothing in baseball and moving a guy in and out of the lineup is a dangerous practice. However, I do believe in rotating starters based on matchups, getting relievers in to eat some innings and benching guys when they are on, say, a 50AB cold streak.

Re: STRATEGY!!!!

PostPosted: Fri Mar 14, 2008 8:59 pm
by tmlfan4ever
I think the major disagreement here is that you think eight consecutive 0-for-4s will have no relation at all to what Matt Holliday will do the next day* -- in other words, that slumps are merely statistical anomalies where the coin comes up heads 6 days in a row. This seems very unlikely to me. Now, there is probably no way of proving either one of us correct or incorrect (unless Bill James has done a study on this already, which would be awesome). So in the absence of data, I'll pose a rhetorical question:

In your heart of hearts, if you have Matt Holliday coming up and the pennant on the line, do you want him to be stuck in a 2-for-26 slump?


*Out of curiosity, I skimmed Holliday's game logs and found that he had back-to-back oh-fers only three times all season, and never once had three in a row. That's insane!


You just contradicted yourself. You said that a guy going o'fer several times in a row means he is more likely to continue, but you also identified a hitter who has proven that he can beat a cold streak after back-to-back o'fers. If you take holliday out of the lineup after a couple of bad games and he goes 2-3 with a HR and 3 RBI... jokes on you.

Re: STRATEGY!!!!

PostPosted: Fri Mar 14, 2008 9:02 pm
by GotowarMissAgnes
I think you're kidding yourself if you think that immediate past performance has no correlation at all with immediate future performance.


Actually, you're kidding yourself. This has been studied over and over and over and over and over again, and if you have not seen the research, I suggest you take a look on "hot hand" research that's readily available through any major net or academicsearch engine.

What it clearly shows is that recent good or bad performance adds NO value to predicting what will happen next. Zip. Zero. Zilch.

Even in the midst of a big slump or streak, the best prediction of a player's performance over the next few games is an unadjusted long term average of the player's past performance.

Re: STRATEGY!!!!

PostPosted: Fri Mar 14, 2008 9:05 pm
by tmlfan4ever
GotowarMissAgnes wrote:
I think you're kidding yourself if you think that immediate past performance has no correlation at all with immediate future performance.


Actually, you're kidding yourself. This has been studied over and over and over and over and over again, and if you have not seen the research, I suggest you take a look on "hot hand" research that's readily available through any major net or academicsearch engine.

What it clearly shows is that recent good or bad performance adds NO value to predicting what will happen next. Zip. Zero. Zilch.

Even in the midst of a big slump or streak, the best prediction of a player's performance over the next few games is an unadjusted long term average of the player's past performance.


Bingo. Hitters hit. Unless you're working in a guy from the Waiver wire on a thursday or tuesday (with a less than full slate of games), get your best hitters in the lineup and keep them there. Period. +1

Re: STRATEGY!!!!

PostPosted: Fri Mar 14, 2008 9:27 pm
by OneLoveBoomer
Whatever the "research" is, I do think there is something to the "hot hand" in the short term. If, on a Tuesday or Thursday, I decide to add another batter in an H2H league to keep up with an opponent, I have often had great success with sorting by stats from the last 5 days. I recommend this for sure. Shelton was good for this a couple years back :-b

I wouldn't sit Wright in favor of Encarnacion. One potential difficulty I'll be dealing with this season, however, is what to do with Lastings Milledge. In a league where I'm full and he's on my bench, I'd love to work him in. Will I put him in for Thome or another similar type player when they're in an awful matchup, like against Sabathia or some such pitcher? Yeah probably.