Updated Projections

[Stat Boy]

Ray, I posted this in the discussion section of the article, but got no response from you or anyone else at HQ. As I said in the post, I've brought this up before and gotten no traction, but I think it would require minimal effort and offer a lot of help to users. I'll paste my question and solutions here. I'd really like to know if this could be at all viable, and wasn't sure if the site overhaul might help facilitate it. Anyway, here's what I wrote:

I've been requesting this for years and have consistently been told it's too hard or impractical for reasons that escape me. Perhaps the power of the new site might make it a possibility. But it's basically this:

Can you please keep a running list of all actual changes to the projections (which, as you point out, almost always involve adjustments to playing time)?

There seem to be 2 simple ways to do this:

1. Have a document that lists what players have had their projections changed by date. So you'd have all the players that had their projections changed on, say, 3/4 listed. Then the next day, on 3/5, you'd have all those players listed.
2. Conversely, you could just add a column to your 2025 Projections Datatable that says what date the most recent projection was updated on. So, say, for Shohei Ohtani it might never change. The date would just say 1/25 or whatever. But for a rookie that just made the team, you'd have 3/18 listed by their name when the adjusted projection was made.

Why am I so desperate to have this? Because I use Rotolab. And as much as I love and respect HQ's projections, I like to adjust them and take into account other site's projections as well and tweak things where I might disagree on the metrics or playing time. But what I have to do then is meticulously compare daily CSV lists to see what Playing Time changes and the rolling impacts they have, because I do care what HQ has to say.

It would make life so much simpler if you could just implement either version of the changes I'm talking about, so I could reorder the CSV list to just show me the most recent dates of changes in the projection. Just have whoever inputs the changes to a player also list the date the changes were made in a column. Or, in the other suggestion, just have a running list so that whenever a player's projections have changed, their name gets added to that day's list.

I honestly don't see why this should be hard or complicated.

[RAY@HQ]

I honestly don't see why this should be hard or complicated.[/I]

I get that you don't see it, but the fact is that it is complicated. There have been various asks for this over the years, enough that if it were easy or even just mildly hard to do, we would have done it.

The new site changed a bunch of things, but the projections process is pretty much the same as its always been. I'll bring it up for discussion again, and I'm not saying it's impossible or will never happen... but it's not easy.

[Stat Boy]

Couldn't you just add a column that lists the date a projection was updated? I'm sorry to be naive on how this might work, but that would seem the most straightforward way to address it. The thing that's hard for me is having to compare daily projections and look for any changes, etc. It's incredibly time-consuming, especially as draft day nears and spring training changes are happening. I assume whoever makes the projection changes has to input everything. All they'd have to do is input the date they make those changes in a new column. Is it that hard to add a new column?

[RAY@HQ]

That's not how it works. Sure, the simple case that you're probably imagining is that if someone changes, say (this just happened last week) Juan Soto's PT% from 95% to 100%, we could do something that flags that update and throws an X or a date next to Soto's projection in the CSV. But that's just one example of a projection change. Here are some others, just from the last week or so:

- I could update a player's baseline projection. Those happen offline and get uploaded to the site. Could I potentially flag every change I make in that process? Possibly.
- I changed the Royals team saves projection from 30 to 40 last week. That affected saves projections for all KC relievers, but I didn't touch any KC reliever individually.
- As playing time projections get updated, changes to OTHER players projections will cause dollar value changes for the rest of the player pool. Should those get flagged? How? I didn't touch anything to do with the player, but his value changed because the player pool itself changed.

I could go on. Hopefully this gives a sense of the scope of the problem.

Getting back to your original ask: you said you're using Rotolab. Once you override a projection there, you are excluding them from further updates. You're saying, "don't use the HQ projection, use mine". Which is great, we want you to do that! But then after you've put your stamp on the projection, you're still.... wanting to see if we change ours further? Why? (Oh, and now Rotolab needs to get looped into this whole "when did the projection last change" loop. The plot thickens...)

[ToddZola]

I've had to do this sort of thing. Using vlookups, the process takes 5-10 minutes, at most, for hitting and pitching. I can go into more detail, but I'll take the most current download and the one to which I am comparing. They don't have to be on the same worksheet, vlookup will work across spreadsheets.

On the first available column in the most recent, I do a vlookup using the ID and call in the plate appearances (or innings pitched) and then make that cell the current PA minus the previous PA. ZER) means no change. I'll also call in a decimal like OPS (if available) since it captures BA and SLG. If you want to check steals (or saves) you can do that as well. Pitching I'll do ERA and WHIP, or sometimes concatenate them.

Then you sort by the column of choice. Each vlookup takes seconds to set up, then fill down and voila.

[Bean Counter]

I use Rotolab & use one of the flags to note any player that I have done an override. As draft day approaches , I check the HQ's projections on the site to see if there are significant changes that I may have to account for. Takes me about 10 minutes each time. I suppose if you make a lot of changes it would take much longer, but I typically only make changes to a couple of dozen players.

[Stat Boy]

Ray, I hear what you're saying. Didn't realize that's how it worked. But to answer a couple of your scenarios: 1). You don't have to loop Rotolab in at all, IMO. That's a separate thing. 2). What I do is change player projections on Rotolab to my own by integrating HQ with other projection sites that I like that might offer some contrarian projections. I don't usually do this until the last week before my draft because I want HQ's projection to be as close to final as I can before I make the adjustments. I'd love to do it earlier, but that's where not wanting to check all the HQ changes every day comes in. The reason I go back and check HQ's site is because I DO trust HQ's PT projections and stat projections the most. So when I make my adjustments on Rotolab, HQ' is weighted the heaviest. But even if I were ONLY using HQ's projections, I might want to adjust BA by incorporating BA & XBA or adjust ERA by incorporating both ERA and XERA, etc. Or maybe I have a different sense of certain players' PT.

So, okay, if the column idea is too onerous -- what about a running list of players affected by changes. So in your scenario, where the PT is a rolling effect for a number of players, they'd all just go on a daily list. So March 3rd would have all the players listed who've been affected. March 4th would have all the players listed who've been affected on changes that day, et. al. Is that also hard to do? It seems anyone making the changes, say, to the Royals would know who was impacted by any adjustments in PT or whatever. There could even be a Google document anyone making changes could contribute to after they make them. Is that at all viable?

P.S. While I appreciate the vlookups suggestion Todd mentioned, I think he's more adept at spreadsheets than I am. I honestly couldn't even follow the basic explanation he offered. And that's on me, not him, BTW. What would be the biggest help to me is just to know what players have changed so I can quickly look them up and see how it will affect the projections I'm using in Rotolab.

[RAY@HQ]

Sorry, no, what you're describing just isn't viable.

[Stat Boy]

Just curious -- what are the roadblocks to that one?

[RAY@HQ]

You're asking 50 staffers to all remember to track their work in a google doc. It would never be reliable. If I were to do this, I would do it programatically to track changes in the system when they happen. But we don't have time or resources to attack that right now. I'm spending resources on LeagueSync, on the Custom Draft Guide, on our daily article email... all of which have been popular requests during the offseason. You're the only one asking for this right now.

[JonE]

P.S. While I appreciate the vlookups suggestion Todd mentioned, I think he's more adept at spreadsheets than I am. I honestly couldn't even follow the basic explanation he offered. And that's on me, not him, BTW. What would be the biggest help to me is just to know what players have changed so I can quickly look them up and see how it will affect the projections I'm using in Rotolab.

The basic concept is that VLOOKUP lets you take a value that appears on one spreadsheet and look for that same value on another spreadsheet. In essence, what Todd describes goes like this. Let's say you have the January 15 projection and January 16 projection in two tabs of a workbook, and then you add a third blank sheet.

a. Juan Soto's projection in the January 15 file was 46 home runs.
b. His projection on January 16 is 46 home runs.
c. In the blank sheet, use VLOOKUP to find the January 16 projection and subtract it from his January 15 projection. (Todd does it right there in the January 16 file, but making a third tab is simpler for a beginner.)
d. Repeat this for all relevant stats and players.

When you are done, that third tab will consist of mostly zeroes (unchanged projections), but you can quickly filter out the zeroes and spot the changes.

Stat Boy, you describe spending so much time working with daily projections that I think you will definitely benefit by taking the time to learn how to use VLOOKUP. Frankly, it should also help you with your comparison of HQ to other projection sources. A quick search of YouTube reveals dozens of tutorials lasting 10 minutes or less.

Patrick Davitt wrote a series of essays for HQ about how to use Excel for baseball, but I realized when I went looking for them that they are pretty old and didn't make the site conversion. Maybe Ray can dust them off and re-post them, or possibly ask Patrick to update them. IIRC they are pretty evergreen, but maybe I'm wrong.

Bean Counter's idea is a good one and you could also use that. It seems well suited to someone like me, who only spends two intense weekend days comparing HQ to other sites, making some modifications (hmm, three of my five sites like Cole Ragans more than HQ, let me bump him up), and then using the daily HQ updates in Rotolab for fine tuning.

[Stat Boy]

Ray, I think you're making what I'm suggesting sound much harder than it is. I don't know how the work is divided among the 50 staffers, but for argument's sake, say, one staffer is assigned to follow the Reds (among maybe other teams). They determine that Jake Fraley has moved from 40% playing time to 60% playing time and there are ripple effects for 5 other players because of this that is reflected in their PA going down and Fraley's going up. All the staffer would have to do is list Fraley and the 5 other players names as having changed on 3/18 somewhere on the site. And then whoever makes changes on other teams would add the names of players whose PT or projections have changed to that date on the list. That's all it is -- a list of what players changed on what date, nothing else.

But if that's an insurmountable challenge, my other solution still doesn't seem so onerous to me. Mostly, of course, because I don't know how the inner workings at HQ work and you do. That said, the earlier roadblocks you mentioned to having a column on the Projections that lists the last date a player was affected seems pretty basic.

Whoever uploads or makes changes to the site would also upload or makes changes to a new column that lists the date when they're making or inputting the actual changes.

So to be clear, of the three problems you mention:

1) The one about player pool valuations isn't affected or relevant. It's about stats and PT, not dollar values.

2) The one with changing a player's baseline projection only means you'd list the date anything changed with the player. You would NOT have to flag anything specific about the projection or various stats, just simply list in a new column next to the player's name the date a player projection was adjusted. That's it. So it would read. Soto, Juan. NL. Mets. 3/18.

3). And in the Royals Saves example, you'd just update the date where the closer and other pitchers were affected (i.e., anyone in the pen whose Saves numbers changed would have a new date in the column indicating something had changed).

Look, the bottom line is I'm the only one apparently asking for this (or at least publicly asking for it), and you don't seem to have any interest in doing it. But the problems you're mentioning seem solvable and shouldn't require much extra work. Again, I don't know the mechanics behind how things are updated or who updates them or what the division of labor is. But if you don't want to do it, then you don't want to do it.

[Stat Boy]

JonE, I'll look into the VLOOKUP you and Todd are fans of. But even what you're describing -- where you have to input stats and players -- sounds problematic. My goal is to quickly, at a glance, see whose projections or PT (usually they go hand in hand) has changed from from the last time I referenced the projections. Having to go player by player or stat by stat seems like it would take forever. Or maybe I'm not fully understanding the process or what you're saying -- which is entirely possible.

[ToddZola]

I don't have access to the current content, but if you send me two downloads for hitting and two for pitching, I'll add the formulas and explain them. You'll see it takes around 10 minutes (at most) and everything is at your fingertips. todd dot zola at gmail dot com

[RAY@HQ]

Look, the bottom line is I'm the only one apparently asking for this (or at least publicly asking for it), and you don't seem to have any interest in doing it. But the problems you're mentioning seem solvable and shouldn't require much extra work. Again, I don't know the mechanics behind how things are updated or who updates them or what the division of labor is. But if you don't want to do it, then you don't want to do it.

Look, I don't mean to be dismissive. I have a good understanding of the ask here. Yes, you're the only one asking for this right now, but this has come up a number of times over the years. And I've got a really good understanding of what it would take in our systems of processes. I've been pretty clear here that it's not easy. If it was, it would be done already.

I will have some conversations with our techies about what's possible. Things like outputting a daily or weekly list of playing time % changes might be very do-able. I realize that's not exactly what you're asking for, but it's a step in that direction.