NRAA Glossary
Per reader request, a mini Net Runs Above Average Glossary.
Net Runs Above Average is calculated as follows:
[MLV(100/G) + (Rate2 - 100)] = NRAA/100 G
That number can then be divided by 100 to get the per game value of NRAA. That number can then be multiplied by however many games you wish to get rate n. I use per 100 games, per game, and per 162 in my spreadsheets, but any figure can be used.
MLV is Marginal Lineup Value. It is the cumulative version, rather than the rate stat MLVr.
Rate is the simpler version of the defensive metric; for the purposes of NRAA I use Rate2, which is adjusted for league difficulty.
There are other versions of NRAA though; eNRAA and pNRAA. eNRAA is Established Net Runs Above Average; this figures gives an established level of performance to better gauge future performances. Not as much of a projection as a guideline, with context taken into account. pNRAA is Positional Net Runs Above Average. This adjusts for position (really Marc? Does it now?). The formulas, for eNRAA and pNRAA respectively:
eNRAA
Firstly, find the EPL for MLVr:
MLVr EPL = (3 * PAx * MLVrx + 2 * PAy * MLVry + PAz * MLVrz) / (3 * PAx + 2 * PAy + PAz)
Where x = most recent year, y = second most recent year, and z = third most recent year
Then, find the EPL for Rate2:
Rate2 EPL = (3 * AdjGx * Rate2x + 2 * AdjGy * Rate2y + AdjGz * Rate2z) / (3 * AdjGx + 2 * AdjGy + AdjGz)
Where x = most recent year, y = second most recent year, and z = third most recent year
I use MLVr for eNRAA because I cannot get MLV from the PECOTA projection cards. I will need that information in order to work with the younger players who have not put in 3 years worth of time in order to weight their performance effectively.
After that, its time for a slightly different formula:
(EPL MLVr*100)+ (EPL Rate2 - 100) = NRAA/100 G
The same divide by 100, then by n rules apply to this formula.
Lastly, here is pNRAA:
[PMLV(100/G) + (Rate2 - 100)] = pNRAA/100 G
PMLV is Positional Marginal Lineup Value. This is the version adjusted for position, which basically gives centerfielders, middle infielders and catchers an extra boost, while taking away a little from corner guys. When comparing player X to player Y, as stated in comments, it has its uses. It also gives you an alternative to raw NRAA. This formula, just like the others, comes in the 100 and per game forms, as well as to n games.
I hope this will make it easier to follow the statistic, especially now since I believe it is done being tweaked. eNRAA on the other hand will most likely continue to be messed with until its time to project some performances. I should be uploading the spreadsheets once the website I would use starts working again, and then you can download them if you'd like. Or I can do this...
Net Runs Above Average 7-10-05 (Min. 150 PA)
Positional Net Runs Above Average 7-10-05 (Min. 150 PA)
If you would prefer to have the actual spreadsheet so you can sort stats, tell me.
0 recs |
10 comments
Comments
Bobby Crosby
by cephyn on Jul 11, 2005 12:56 PM EDT reply actions 0 recs
Crosby
by Marc Normandin on Jul 11, 2005 1:14 PM EDT up reply actions 0 recs
So in your opinion
Additionally, why MLV instead of MLVr? What's the thinking here?
Sorry I have so many questions. I love this site though. 8)
by cephyn on Jul 11, 2005 3:04 PM EDT up reply actions 0 recs
MLV over MLVr
Crosby is probably one of the best shortstops in the league; the reason he is ahead of Tejada is because of his defense. If you take a look at Crosby's rate, its well above average, while Tejada's is about 8 runs below per 100 games if I remember correctly. Crosby might not be this good, but then again, wouldn't that be nice? He has progressed to essentially what was expected out of him, which makes me giddy.
The reason the raw stats comparison to Jeter does not matter is thanks to Jeter's league average (or slightly below) defensive contributions, compared to Crosby's excellent contributions thus far. Things could change, but this is a rate stat after all.
by Marc Normandin on Jul 11, 2005 3:20 PM EDT up reply actions 0 recs
Hmm OK
Since I'm not familiar with the defensive stat, Rate2, how reliable/powerful a statistic is it? Should it be given this much weight for NRAA? I don't have a clue as to how Rate and Rate2 are really calculated. Are we basically assuming that just by placing a body at a position, we prevent 100 runs? But by placing Jeter at short, he only prevents 97 runs, whereas Crosby prevents 111 runs? Over how many games? 100? 162? And this is a reliable measure?
And the next question is then this:
You take MLV and multiply it by (100/G), thereby making it a rate stat. So why not MLVr? aren't you just recreating it? But anyway, you take this rate stat, and add the runs saved by defense, but the Rate2 number isn't ever adjusted for games played. In other words, Crosby isn't saving 11 runs per game and Jeter isn't giving up 3 runs per game, right? So when you go multiply out to project the 162 NRAA value, aren't you way way overvaluing the defensive contribution to the stat?
by cephyn on Jul 11, 2005 4:53 PM EDT up reply actions 0 recs
Rate2
111 is not how many runs are saved; the figure 100 is league average. 97 would be 3 runs below average defensively per 100 games played, and 111 would be 11 runs above average per 100 games played. Notice in the formula, the Rate2 portion is (Rate2 - 100). This would give us just the number 11 for Crosby, which is how many runs above average per 100 games he is.
By the way, MLV is used rather than MLVr because James Click, BP writer, told me that it would make more sense to divide the cumulative stat, which is what they have done, rather than to use their MLVr (rate) from the get go. So I tweaked the formula, and there is a difference in the numbers.
Anything else? Sorry if I wasn't clear on something. If I'm blatantly missing your point its because of my knee surgery and my heads sort of cloudy at the moment thanks to an IV full of goodies, not because I'm dumb. Thats just the rest of the time :)
by Marc Normandin on Jul 11, 2005 10:29 PM EDT up reply actions 0 recs
Finally
One Last Stupid Question: So if a player has a 0.00 NRAA, can we consider him truly an average player? And what I'd love to see is some look at historical players, particularly guys who were considered No-Hit/Good-Field guys - did they really save their team runs, or were they actually a liability with spectacular plays? Example with your data, Juan Castro and Alex Cora.
The other real world light bulb from this -- holy crap, now we know why the A's offense perked up with Crosby back. He's an 8-fold improvement over Scutaro!
by cephyn on Jul 12, 2005 2:28 AM EDT up reply actions 0 recs
historical data
I wasn't getting frustrated at all, I was actually excited someone was so interested in figuring it out.
0.00 is an average player, you are correct. I plan on simply keeping a leaderboard with a plate appearance cutoff throughout the season for easier to update spreadsheet purposes, but when the season is over I will calculate everyone (over 800 offensive players I believe, unless they included pitchers, then I may have a reprieve). For the historical seasons I will cover every offensive player as well, so we can begin to add together career NRAA for 1972 and up players.
by Marc Normandin on Jul 12, 2005 8:15 AM EDT up reply actions 0 recs
its good work
1)What stats are its "competition" (tell us similar things)
2)Why is NRAA better than the competition?
Abstract questions. But its a great stat. I hope someday I can look at some numbers and come up with something useful.
by cephyn on Jul 12, 2005 12:12 PM EDT up reply actions 0 recs
Holy Cow
by cephyn on Jul 12, 2005 12:32 PM EDT reply actions 0 recs















BtB on Facebook














