Schedule Analyzer -- Ease

mymaus  5/30/2013 1:30 PM

There doesn't seem to be any Ease data available in the Schedule Analyzer. I'm trying to pick between 2 players whose projections are close and I decided I was going to go with the one with the easier schedule.

ken_admin  5/30/2013 1:56 PM

Yes, that's disabled currently, but I will enable it for tomorrow.

ken_admin  5/30/2013 4:07 PM

This will be deployed tomorrow, but you can view it on the preview site for now if needed:

http://preview1.baseballmonster.com/WeeklyTeamSchedules.aspx

mymaus  5/30/2013 9:26 PM

Thanks!  Unbelievably quick as always. I've been thinking about Ease a little. I assume that Ease is based on past performance, not future, right? Well....since you are already doing projections of the future, I wonder if you could use those to make Ease better. I know we would use Weekly projections for any short term projections within 2 weeks of the current date, but I wonder if we could use your ROS projections instead of past performance. To get pitching ease I THINK you could use the cumulative projections of all the projected stats of all of each team's players. Hitting ease would work the same way. This will be especially important after the trade deadline. When key people switch teams, team past performance may not be a good predictor of ease.

To make it even better, since you are already applying park adjustments to daily projections, why not use them for these ease projections as well?

What made me think of this is, thanks in part to BM I am doing very well in my H2H league. Believe it or not I am looking at the playoff schedule for my players and thinking about my roster as it pertains to my players during the playoff weeks. One of my players is Allen Craig. He has 4 games during the finals @COL. One would think that fact would be a huge advantage, but, believe it or not, Colorado pitchers have been suppressing HR's at home so far this season! I'm guessing if you look at your ROS projections for the entire COL pitching staff and then make a park adjustment you will not find the ease factor to be the same as it is shown now. If we are confident in your projections, why not use them for ease?  Thoughts?

ken_admin  5/31/2013 4:35 PM

Yes, the eases shown in the schedule analysis are currently based on past performance.  You're correct that using our projections could be a good source to determine future ease (for long-term projections).  We'll give this some thought, and see what we can do. Thanks

mymaus  6/7/2013 7:22 PM

Any more thoughts about this idea? I THINK it would be a relatively easy cange to make -- just cange your source of data from the past statistics to the projected statistics. I guess park adjustments may be a little harder, but since you are already doing it for daily projections I would think you would at least have a model for getting it accomplished.

I think this would make the EASE factor MUCH more accurate. If not, why don't we use past performance to rank instead of your projections?

Thanks again. Keep up the good work. 

mymaus  6/8/2013 9:56 AM

Reading a question in another one of your forums and I came up with what I think is an eaven better idea....

Why not adjust your ROS projections using your Ease factor. I would still use your projections (instead of past performance) and your park adjustment factor to calculate your Ease factor before applying it to ROS projections but, because I know this may introduce a loop of adjusmtents (i.e. using hitter projections to adjust Ease and using Ease to adjust pitcher projections and I am suggesting using pitcher projections to adjust Ease which will be used to adjust ROS hitter projections which I am suggesting to be used to adjust Ease which will be used to adjust pitching projections etc........) I suggest the following ...start with your current ROS hitter projections (i.e. not adjusted by Ease/park factor yet) and then make the Ease/parkfactor adjustment for ROS pitcher projections based on that.. Then for hitters use you current ROS pitcher projections (i.e. not adjusted by Ease/park factor yet) and then make the Ease/parkfactor adjustment for ROS hitter projections based on that.

You are already doing this type of thing for daily/weekly short term projections, why not apply it to ROS projections. Yeah, I know your daily/weekly projections rely on as accurate of a projection of actual lineups as you can get. But still....I think applying an agregated team Ease factor combined with a park adjustment  would signifantly improve your ROS projecitons. Right now I find myself struggling to compare an excellent pitcher for the Yankees (worst Ease based on past performance) with a very good pitcher for the Braves  (best Ease based on past performance).  Why not help us all out with that struggle?

ken_admin  6/8/2013 11:27 AM

Yes, it does make sense to use the ROS projections to create a ROS-ease.  Obviously, this will be an estimate since injuries and most call-ups will not be factored in.  We'll give users the option to view ROS or past-results ease (with date range).  It shouldn't be too hard to create a base ROS-ease based on projections, but adding the other factors may take some thought (since the current processing is already slow).  We'll update our progress here. Thanks

mymaus  6/8/2013 11:33 AM

Thanks AGAIN for the quick response. ROS ease would be great but what about the even greater idea of adjusting ROS projections using ease as I suggested above?

ken_admin  6/9/2013 9:18 AM

Yes, we will try to include those too, and the way you describe the process makes sense.

mymaus  6/9/2013 10:02 AM

Fantastic! I have no doubt that giving the option to incorporate projection-based Ease into the projections will make looking at trades and waiver pick ups better with more accurate projections. I am guessing it's going to give a little more value to NL East pitchers and a little less value to AL East pitchers becasue of their schedule. Logic tells me this, but it doesn't tell me how MUCH more/less I should weight them.  I can't wait to see the results.

I'll look to this thread to see when they are done or maybe you want to make a seperate thread announcing it. I really think other users will be excited -- anything to get more accurate projections!

mbuser   6/9/2013 12:37 PM

Just keep in mind the qualifiers that Ken mentioned previously when considering accuracy. We can really only project out current rosters for the most part, so the further out the less likely we'll be able to claim accuracy. Think about after rosters expand to 40, in particular - it's almost impossible for us right now to account for the adjustments to lineups and pitching staffs that happen at that point. And of course unforseen injuries will play a factor, as well.

mymaus  6/16/2013 8:56 PM

Any thoughts on ETA for this?

Matt...I understand your about be careful with this given the accuracy of rosters, but I guarentee make adjustments of Ease based on projections of those on current rosters is MUCH better than Ease based on past performance. I can use your same arguament about rosters expanding in Sept -- Ease would not take that into account.

As far as using Ease to adjust projecitons I think the idea of giving the user a choice is a good one. One very relavent example is Josh Johnson. Toronto faces one of the worse schedules ROS. This is not reflected in Josh's projections causing one to possbile overvalue him vs one of the Atlanta ptichers that have the easiest schedule. The nice think about using projection based ease to make an adjustment to projecitons is that it would immediately reflect changes in roster.....I assume come September you guys will adjust playing time projections based on changes to a roster. These playing time adjustments would be reflected in ease based on projections and thus in projections base on ease.  

I really can't think of an argument against this improvement to your projecitons.  Look forward to seeing them!

Thanks to the both of you for all of your hard work this year. BM has come a very long way since I first saw it 2 years ago.

ken_admin  6/17/2013 7:43 PM

This won't be in tomorrow morning's update, but it's next on the list and should be ready in a couple of days.

mymaus  6/17/2013 11:38 PM

Fantastic!! Let me know in this thread. It will be extremely interesting to see how applying porjection based ROS ease changes the projections!

mymaus  7/2/2013 10:18 AM

Any progress on the option to adjust projections based on ROS ease projections which would be based on base projections (see my full description from my post on 6/8/2013 9:56 AM above -- you thought it was a good idea in a subsequent post).

ken_admin  7/2/2013 12:54 PM

Progress has been made, but I've been sidetracked a little.  It's still coming :) Sorry for the delay.

mymaus  7/3/2013 2:58 PM

No need to apologize. You guys have been great!  I am going to be very intersted in the results of both the Ease based on projected instead of past stats and the new projections based on an adjustment made for Ease based on projected. For example, right now WAS is ranked in green for pitching ease in my 5x5. I suspect that ease based on projected stats of their current active roster will easily change to red since they are way more healthy now than they have been all season.

mymaus  7/6/2013 9:19 AM

To add to my previous post...I think LAD and Miami should definitely be better offensively than they were the first half. After the trade deadline there could definitely be changes in ease. There is no question that projections based on an adjustments made for Ease based on projected stats ot history will be better than what we have now.

mymaus  7/18/2013 8:01 PM

Any hope for any time soon? Sorry to be so impatient but I only have 6 more matches before playoffs. Thanks for all of your help this year.

ken_admin  7/18/2013 8:31 PM

It's turning into more than I thought it would be.  It's still the top priority for baseball features.

mymaus  7/19/2013 8:14 AM

Thanks!

mymaus  7/29/2013 5:33 PM

I know you've been busy...what if I reduce the scope of this work to an option to adjust pitcher (not hitter) projections base on opponent batter projections for ROS. You could use the same projected lineups and methodology that you use for Daily and Head to Head Daily projections and just extended out for the rest of the year. I know an outside source gives you projected starting assignments for the next 14 days. I would like this option to go ahead and project opponents for the rest of the year and then make adjustments to the pitcher ROS projections accordingly. I don't know if you can get projected opponents from your source or if you'd have to do it yourself. If you did it yourself I would just assume a pitcher pitches 1 of every 5 games. Better yet....let the 1 of every 5 games be the default and let the USER change the default and project different starts ROS by giving them a grid like you do in Daily Head to Head and having them place check marks in the starts he is projecting the pitcher to make.

To boil this down simply -- make your Daily Head to Head Template into a ROS ranking tool for pitchers. Do a projection of every pitcher for every game adjusting to the lineups. Use the projected start info from your source for where you have it and the default of 1 of every 5 starts when you don't -- allowing the user to change the starts.

Forget about doing the same sort of adjustment for hitters. It's not nearly as significant for them as for the pitchers.

Obviously there is a risk to projecting the rotations incorrectly, but you can state that up front when you announce the tool.

If this is still too much trouble, the next best thing would be to allow for the option to be used on your Weekly projections page. Again, simply use the same methodology that you do to adjust daily pitcher projections for opposing batters and park only extend those adjustments out for the rest of the year using the every pitcher pitches 1 out of 5 games assumption.

Thoughts?

By the way, when you currently adjust pitcher performance on opposing batters, do you use your projections for those opposing batters or the current season's performance? I would hope you use your projections.

ken_admin  7/30/2013 9:12 AM

Yes, we do use the projected hitter stats rather than the current season.

Using the actual projected starters and lineups for the ROS ease does get a little complicated.  For one, the processing for the weekly projections including all factors takes quite a while so applying that to the rest of the season would be very slow.  Projecting the rotations would be possible for the rest of the year, but they would likely be quite inaccurate so I'm not sure how many users would want to use it.  We've made the assumption that only short-term rotations have value.

Maybe there's an offline solution we can come up for you.  I can contact you directly to see if we can find with something useful.