Archived

This topic is now archived and is closed to further replies.

jsp2014

Hot Streaks -- A RotoWorld Study

26 posts in this topic

There is another topic about predicting hot streaks. rather than making assumptions, let's collect some data.

if you think a *batter* is on a hot streak or about to start a hot streak, post here. please provide some brief reasoning for why you think he's hot. I probably won't use random guesses but that will be determined ahead of time.

I'll probably search all Razzball mentions of hot shmotato as well.

from there, I will attempt to aggregate the data into a public Google Docs spreadsheet for the 2, 3, and 4 week periods following the hot streak mention. for a baseline, I will use the statistics that player collected from beyond the 4 week period. I think that is the best way to compare.

if anyone has suggestions on methodology, I'm open to the possibility that a better method exists.

2 people like this

Share this post


Link to post
Share on other sites

Josh Reddick: A career high BB%, a career low Swing%, making better contact in the strike zone and still hitting a nice amount of fly balls. He's now hitting .310 since he came back from the DL. Only 5 K's in 55 AB. His LD% this month is at 22%, right around career average.

Share this post


Link to post
Share on other sites

Predicting hot streaks is difficult to impossible. If you want to split hairs, here are some ideas I had on this topic:

What's useful to us is calculating the likelihood (http://en.wikipedia.org/wiki/Likelihood_function) that a player will outperform his current pace over a sequence of n games, where n is user-defined. Ideally, it would be a mathematical formula where you could input a batter's schedule, handedness of potential pitchers faced, past performance, and park location. Then it would be relatively straightforward to output a distribution on how the batter has performed, and where the batter is expected to perform, along with a confidence interval (or six sigma) on how much he will under or over-perform.

An interesting topic is how career numbers correlate with future performances for the average batter. If you can use career numbers (maybe with an age factor to predict drop off), then you can some pretty powerful information. Otherwise, if you go by current season, you probably need to wait till the all star break before getting enough history on how the batter is performing. Obviously, the formula gets weird when you are looking at someone who is injured or is recovering from injury, but I bet with some reasonable thought, even that could be factored in.

Just a few cents of my somewhat admittedly lacking knowledge of probability.

Share this post


Link to post
Share on other sites

One thing to consider is not just a hitter's baseline but that the quality of opponents vary per week. So a guy who played in Coors last week will be more likely on a 'hot streak' vs. a guy about to play in Coors.

I'd consider using the Razzball Hitter-tron (http://razzball.com/hittertron)* as the baseline for expected stats for upcoming games since it factors in the opposing pitcher (including handedness), park, and expected spot in the batting order.

* The Hitter-tron is run by me so bias noted.

4 people like this

Share this post


Link to post
Share on other sites

One thing to consider is not just a hitter's baseline but that the quality of opponents vary per week. So a guy who played in Coors last week will be more likely on a 'hot streak' vs. a guy about to play in Coors.

I'd consider using the Razzball Hitter-tron (razzball.com/hittertron)* as the baseline for expected stats for upcoming games since it factors in the opposing pitcher (including handedness), park, and expected spot in the batting order.

* The Hitter-tron is run by me so bias noted.

heyoo, the afro came out to play.

that's a good point that I'll keep in mind. likewise, a batter who has favorable matchups over the 2-4 week evaluation period is generally going to look better than he should. hopefully with a large enough sample, that part will mostly even out. I'll also look at wRC+ to at least negate the park factors.

that's a good idea using the Hitter-tron as a baseline. only problem is I'll have to make sure I grab the data at the appropriate times. we'll see how well I can stay on top of things.

Share this post


Link to post
Share on other sites

One thing to consider is not just a hitter's baseline but that the quality of opponents vary per week. So a guy who played in Coors last week will be more likely on a 'hot streak' vs. a guy about to play in Coors.

I'd consider using the Razzball Hitter-tron (razzball.com/hittertron)* as the baseline for expected stats for upcoming games since it factors in the opposing pitcher (including handedness), park, and expected spot in the batting order.

* The Hitter-tron is run by me so bias noted.

heyoo, the afro came out to play.

that's a good point that I'll keep in mind. likewise, a batter who has favorable matchups over the 2-4 week evaluation period is generally going to look better than he should. hopefully with a large enough sample, that part will mostly even out. I'll also look at wRC+ to at least negate the park factors.

that's a good idea using the Hitter-tron as a baseline. only problem is I'll have to make sure I grab the data at the appropriate times. we'll see how well I can stay on top of things.

I archive the hittertron projections so you can just let me know which player and which time frame and i'll custom pull it for you.

Share this post


Link to post
Share on other sites

One thing to consider is not just a hitter's baseline but that the quality of opponents vary per week. So a guy who played in Coors last week will be more likely on a 'hot streak' vs. a guy about to play in Coors.

I'd consider using the Razzball Hitter-tron (http://razzball.com/hittertron)* as the baseline for expected stats for upcoming games since it factors in the opposing pitcher (including handedness), park, and expected spot in the batting order.

* The Hitter-tron is run by me so bias noted.

Very nice. I love stat generation websites like this!

Share this post


Link to post
Share on other sites

Very interested to see how this goes. Great idea and pretty cool tool that Rudy has come up with.

Share this post


Link to post
Share on other sites

One thing to consider is not just a hitter's baseline but that the quality of opponents vary per week. So a guy who played in Coors last week will be more likely on a 'hot streak' vs. a guy about to play in Coors.

I'd consider using the Razzball Hitter-tron (http://razzball.com/hittertron)* as the baseline for expected stats for upcoming games since it factors in the opposing pitcher (including handedness), park, and expected spot in the batting order.

* The Hitter-tron is run by me so bias noted.

Hi Rudy! Glad to see you here. Razzball is an absolutely awesome fantasy baseball site. Thanks for all the great tools you guys keep putting out there (Stream-o-nator, Hitter-Tron, etc.)

As far as hot streaks go, I'll bet Aybar is about to light it up based on past years splits.

Share this post


Link to post
Share on other sites

"

Hi Rudy! Glad to see you here. Razzball is an absolutely awesome fantasy baseball site. Thanks for all the great tools you guys keep putting out there (Stream-o-nator, Hitter-Tron, etc.)

As far as hot streaks go, I'll bet Aybar is about to light it up based on past years splits.

"

Thanks for the kind words. I'm psyched to support anything that tests 'hot streaks' or 'hitter vs. pitcher' matchups. I ignore both of these for the hitter-tron and believe that - while they do exist to a certain extent - they provide almost no uplift in accuracy once you take batter/pitcher/park/playing time into account.

So open to providing any help I can on these topics :)

Share this post


Link to post
Share on other sites

How does everyone feel about Pedro Alvarez? He seems to go on short hot streaks with the BA. Last 7 days: 0.300, but last 14 days: 0.268, last 30 days: 0.240. He will hit the HRs but I've never seen him maintain that BA for a consistent period of more than 2 weeks.

Share this post


Link to post
Share on other sites

I wish I had an MLB package or have the time to watch games closely. I think that's really the only way to tell if a guy is starting to swing the bat well.

Yes I realize that we are not scouts and all visual assessment is subjective to a degree, but most of us have been baseball fans and fantasy ballers for years, so we should be able to tell a good at bat from a bad one

I also say welcome! to Rudy. When's the mustache himself gonna pay a visit? Or is he already amongst us? @@

Share this post


Link to post
Share on other sites

Michael Morse, even though he did strike out in his first at bat tonight. He's had a good three game stretch. He also was hitting a couple of foul balls that were loud. I could see him going on a home run binge like he did at the start of the season.

Share this post


Link to post
Share on other sites

Though I'm not a fan, Josh Hamilton is starting to get right in his swing. Watch for him to go on a 1-2 week tear.

Share this post


Link to post
Share on other sites

Curious to see how this shakes out, but didn't fangraphs show predicting hot streaks is mostly nonsense?

Share this post


Link to post
Share on other sites

I can't believe no one mentioned Puig.

Share this post


Link to post
Share on other sites

I can't believe no one mentioned Puig.

You think he's going to get even hotter?

I think one of the first things you'll need to do is define what a "hot streak" is. What are the parameters? How long? What statistical targets does one have to hit? How many ABs? Etc, etc.

Share this post


Link to post
Share on other sites

Cespedes would be someone. I need him to go on a hot streak so i'm curious...

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.