We need to talk about ratings again

[dancariaz]

Example: Justin Verlander.

His stat line during last 2 weeks: 19 IP, 7H, 1 ER (1 HR), 4 BB, 34 Ks.

In this span his movements ratings fell (!) 3 (!!) points against both righties and lefties. Afaik movement ratings are based on HRs allowed. How does his rating fall 3 points while allowing 1 HR in 19 innings? Explanations please.

[bailey]

Maybe the formula would be too easy for some brainiacs to reverse engineer and give them a huge advantage, so to confound them a random number plus or minus is added to each result.

[dkgo]

The blend of current stats vs original projections isn't continuous. His season HR allowed is terrible. He's given up the most in the majors.

Basically he wasn't dropped far enough in the first place when initial ratings were the dominant factor and now that season long stats take up a bigger weight he is getting punished more for them.

Imagine someone who gives up 10 homers in their first game. They aren't going to drop 30 points in movement the first week of updates, but over time as the real stats make up a bigger portion of the weight then that rating can still drop even if he doesn't give up any for a couple weeks.

[Westheim]

Originally Posted by dkgo

The blend of current stats vs original projections isn't continuous. His season HR allowed is terrible. He's given up the most in the majors.

Basically he wasn't dropped far enough in the first place when initial ratings were the dominant factor and now that season long stats take up a bigger weight he is getting punished more for them.

Imagine someone who gives up 10 homers in their first game. They aren't going to drop 30 points in movement the first week of updates, but over time as the real stats make up a bigger portion of the weight then that rating can still drop even if he doesn't give up any for a couple weeks.

This, more or less. It was said that the live part of the ratings gradually increase throughout the season until at the end the live cards' ratings are 100% live. So currently they have about 30-40% "vanilla stats" still included, but those horrendous live stats continue to seep in, week by week.

Or in other words, he is still being punished for all the carnage he caused in April, May...

[dkgo]

For a clear example, say a player starts the season with an amazing 100 movement.

Week one he gets shelled. 10 HR allowed in one inning of work.

After week one, the real stats may make up just 4% of the overall rating (25 week season, in PT it is probably less since the final rating still will include some of the baseline I believe). If that 10 HR allowed correlates to a movement of 0 then the updated overall movement is 96.

Week two he is better. No homers allowed in a complete game. The player now has 10 HR allowed to 10 innings of work. Still a terrible number! Say this rate results in a movement rating of 5.

Real stats after week two make up 8% of the overall rating, so the player's new movement rating with these assumptions is (100 * .92) + (5 * .08), or just 92.4 which gets rounded to 92. His raw overall movement has dropped another 3.6 points despite throwing 9 innings of 0 HR ball because the initial terribleness has a bigger influence as well.

[One Great Matrix]

Originally Posted by dkgo

For a clear example, say a player starts the season with an amazing 100 movement.

Week one he gets shelled. 10 HR allowed in one inning of work.

After week one, the real stats may make up just 4% of the overall rating (25 week season, in PT it is probably less since the final rating still will include some of the baseline I believe). If that 10 HR allowed correlates to a movement of 0 then the updated overall movement is 96.

Week two he is better. No homers allowed in a complete game. The player now has 10 HR allowed to 10 innings of work. Still a terrible number! Say this rate results in a movement rating of 5.

Real stats after week two make up 8% of the overall rating, so the player's new movement rating with these assumptions is (100 * .92) + (5 * .08), or just 92.4 which gets rounded to 92. His raw overall movement has dropped another 3.6 points despite throwing 9 innings of 0 HR ball because the initial terribleness has a bigger influence as well.

Love that the example is 10 HR in one inning of work, ha ha.

Basically you can't look at the past week and expect a jump in a LIVE card from a good week because what will be added to the live is his whole body of work to this point in the season week after week. I think maybe, just maybe, the updates should be monthly (and more reflective of a player's performance for the month relative to where they were.) Of course, at this point in the season, even a month wouldn't necessarily make a difference. The change is something like the beginning of season rating + (to this point in the season) every week...

On top of this, trying to understand them entirely is just going to be a little much unless you've noted a definitive pattern, otherwise something is going to be going on that you just can't identify with any sureness.

[Westheim]

Originally Posted by One Great Matrix

Love that the example is 10 HR in one inning of work, ha ha.

Gotta be either Jeurys Familia or Edwin Diaz.

Edrys Dimilia.

Actual disturbing real life example would be Trevor Rosenthal with the Gnats. 15 walks in six innings? His calculated control rating was probably -24 at that point…