Tuesday, February 14, 2006

Is One Steal Really Worth Sixteen Points?

About a week ago in the comments section of one of our posts, JM made the following comment: “According to espn.com's player rater averaging one steal is equivalent to 16 pts, one 3pt = 7 rebs, and one blk = 4 asts.” I thought this was a pretty interesting statement, and I thought it deserved a closer look.

Before I do, though, I’d like to point out to some of our more casual readers that there’s a lot of great stuff going on in the comments section, often better than whatever junk we’re spouting out here in the blog posts. For the full FBB experience, check out what some of our more talkative readers are discussing down there. It’s pretty impressive.

Anyhow, is one steal really worth 16 points? The answer: Yes … but no. Let me explain:

First, let’s discuss how (I would guess) JM came up with this statement, and then back it up a little bit. In order to have roughly no effect on their fantasy value (as in, it’s not a strength nor a weakness, and the player rater gives you a score of 0.0), a player has to have 16 points, or 1 steal. Other quick stat checks seem to back this up. There are 57 guys averaging 16 ppg, but that goes up to 78 if you go down just a bit, to 14 ppg. Meanwhile, 81 players have 1 steal or better. Additionally, if you look at your league’s standings, there is about a 16:1 ratio for most teams’ points and steals – some higher, some lower, but 16:1 is about right.

Of course, there are differences. For example, 25 times this year, a player has snagged 6 or more steals in a game. Nobody has scored 96 points in a game (16 times 6) – in fact, the hell nearly froze over when Kobe scored 81. But nonetheless, I think it’s fair to say that the value of one steal is of about the same value as 16 points, give or take a point or two.

However, here’s where I think the two stats really break apart from each other: Their predictability. When making a move in fantasy basketball, it’s not what have you done … or even what have you done for me lately. The question is, “what are you going to do for me next?” The fantasy owners who are best able to answer this question are going to be the ones on the top of the standings come the end of the year. And while points are pretty predictable, value in steals goes all over the place. Let me prove it to you:

I took last year’s steals leaders (by average), and compared their numbers to how they’re doing this year. The change was all over the board. Larry Hughes, Mike Bibby, Lebron James and Ruben Patterson were the biggest droppers, losing 1.39, 0.65, 0.61 and 0.61 steals per game, respectively. The biggest gains to be had were by Gerald Wallace (0.73), Brevin Knight (0.48) and Dwyane Wade (0.39). Now keep in mind, we’ve said that one steal is equal in value to sixteen points. So, Mike Bibby’s 0.65 drop in steals is really the equivalent of dropping 10.4 points, which is a pretty improbable number. What’s so confounding with steals is that other than Hughes, none of these guys switched teams in the offseason. None of them have seen drastic changes in minutes, either. Steals are just quite simply a stat that is tough to predict. In fact, the top 20 in steals last year saw their per-game numbers change (either going higher or lower) by an average of 0.39 steals (the equivalent of 6.25 “points”) from last year to this.

On the other side of the spectrum, we’ve got points. The top 20 in points last year (with the exception of Amare Stoudamire) have been remarkably consistent. The biggest changes have been Kobe Bryant (+7.6), Carmelo Anthony (+5.1), and Steve Francis (-4.9). What’s more, you can explain most of the changes: Kobe went crazy, and so did Stevie, just in opposite directions. On a whole, the top 20 in points saw their numbers change by an average of 2.56 points, just about a third of the variability of steals.

So what have we learned here today? In past value, sure, a steal is worth 16 points. But if you’re looking forward to what a guy will do in future games, steals are just SO much harder to predict that you can’t really count on them, where as scorers will find their points, one way or another. But I think this could be a major reason why points are "overvalued" while defensive numbers are slightly undervalued - their consistency from one year (heck, even one month) to the next. Thoughts?

16 Comments:

Anonymous Anonymous said...

i have been playing roto for 3 years now, and one thing about my teams is that they always dominate in steals and blocks, and they rarely dominate in points... i find it much easier to trade for players that excel in these categories compared to scorers (im talking about mediocre scorers, not those guys who go in the first 3 rounds), plus they tend to not screw you over in other categories like TO or FG%.
i do think, however, that bibby's steals numbers last year were a fluke, since he never really averaged a high number of steals except for last year. Also, Hughes went from a freewheeling, gambling oriented defense to mike brown's shut down defense. this might explain lbj's dip too?

11:26 AM  
Blogger T-Plan said...

Good points by Anonymous. You have to look at 2 things:
1. The style of offense/defense they play in (which changes a lot in the age of free agency)
2. Whether last season was an outlier. If someone leaps their steals from 0.8/game to 2.1/game, after averaging around 1-1.1 on his career, don't assume he'll keep up the pace.

My most trusted fantasy advice site uses the following numbers to normalize the cumulative categories (omitted TOs since they are completely unpredictable and a dumb category):

Blocks = 1 (all cats are in relation to this one)

Points = 22.39 (meaning, 22.39 points is the same as a block)

Rebounds = 8.34

Assists = 4.75

Steals = 1.45

3-pointers = 1.22

Essentially, they used the average stats for the top 200 players and compared relative value per category. Then they found the equivalent to zero on the index (similar to what the original post indicated) and built ratios between the categories in light of that number.

Since Blocks are the most rare and highest valued on a per-incidence basis, they built the ratios in relation to blocks.

Thoughts?

12:13 PM  
Anonymous Charlie said...

Extremely well-written article. Thank you guys for the time you spend catering to all of us fantasy nerds out there.

My team is in 2nd out of 12 because of defense. Points are worth 1, rebs=1.5, steals=2, assists=2, blocks=2.5. H2H average per player (8starters) per week. My highest commodities are Wade, Knight, Billups, Rasheed, Paul, Ilgauskus, Kirilenko. These guys have propelled my team because they aren't straight scorers. Each of them excels in at least one defensive category, and many of them excel in more than 1. Also, knowing assists were worth more than rebounds completely shaped my strategy. But I think it's more exciting to track players who help fantasy teams with more than just points anyway.

OK, I need some help with trading strategies. Help me decide which players will have reduced roles on their teams and/or reduced playing time in the 2nd half of the season. Chauncey and Rasheed, I'm split with them. Part of me says Detroit is so damn good that they'll get more rest, and Saunders has said he will reduce their PT. But at the same time, the team has started losing and they're coming back down to earth. This could mean they will take games more seriously since they're not winning every game as easily now, thus keeping in their big dogs longer. I see Chauncey maintaining his value on this team, along with Rasheed. But will I be screwing myself come fantasy playoff time when these guys may be resting more?

And what about Ilgauskus? I really want to send him packing. I think he's overrated, but very consistent. Problem is foul trouble combined with seeing only 29 mpg on a loaded Cavs team (plus they simply underutilize him). With his minutes, he's a tiny step down from the rest of the 2nd tier centers (Yao, Mehmet, etc.) but still consistent. Who can I expect in return for him? Should I respect his consistency more and stick with him? Perhaps package him with one of the previously mentioned players?

OK, hold on to Paul, or sell high? And Kirilenko? He's been unspectacular over the past couple of weeks. Thoughts?

1:44 PM  
Blogger Rook said...

Just a great article. Thanks.

4:39 PM  
Blogger JM said...

Nice posting. very interesting (and easily misunderstood) topic - particularly when certain leagues assign their own value to a type of stat, rather than treating them all equally.
I'll try to clarify where my original statement came from: “According to espn.com's player rater averaging one steal is equivalent to 16 pts, one 3pt = 7 rebs, and one blk = 4 asts.”

I went to espn.com's player rater:
http://games.espn.go.com/cgi/fba/playerrater
Their explanation for how their rater works:
"In each of the scoring categories, a number is calculated that represents the average stat total in that category. If a player has that average exactly, his rating for that category is 0.00. The numbers then represent how much a player is above or below that average. If the rating is positive, then that player is an above average fantasy player in that category. If it is negative, the player is below average in that category. All ratings are then added together to get a grand total player rating, and they are ranked accordingly."
I think that how BV explained it too.

I set the espn rater for per game "averages" rather than season totals. I checked nba.com for someone who averaged 1 stl/game i.e. B.Simmons. I then sorted espn.com's rater by steals, and found that Simmons had a value of "-.13" for that stat. I then sorted by pts, and looked for someone who had a "-.13" value,i.e.Ben Gordon, and then checked to see how many pts he was averaging: 16.0.

I did the same thing starting with 1 3pt/game, and converted to an amount of rebs/game. And then did the same for 1 blk/game, and converted to an amount of assists/game.

Does that make sense?

And I agree with BV - that steals are definitely harder to predict from year to year (compared to pts.)

5:08 PM  
Blogger bv said...

interesting comments. some quick responses:

anonymous, that's a very interesting take on LBJ's dip. Might it also be a reason for Donyell Marshall's lower defensive numbers?

T-Plan - i like the system. i think it goes right along with what i discussed in the article, but it has the same downfall - predictability.

Charlie - i'd say hold on to sheed, but see what sort of interest billups gets you. he's gotta be overperforming so far. Hold on to Z, too. If gooden gets moved for a more D-minded player, then more shots will come to Z. Yes, sell high on Paul, but hold on to AK. His value on the trade market isn't close to his real value.

JM - makes sense to me. I think we're on the same page.

5:34 PM  
Anonymous AJ said...

The problem with ratios in regards to fantasy stats is that they are not linear. Sure 16 points = 1 steal at a Zscore of -.13 (as reflected in ESPN's PR), but at a Zscore of 1 (21pts and 1.5 stls respectively), the ratio changes to 14:1 and it will continue to change as each stat increases. You cannot extrapolate those numbers to assume that 96 points = 6 steals.

6:53 PM  
Blogger bv said...

AJ,

the point i was trying to make is that there is some validity to the "luck" factor of getting steals. While I agree with your statement to a point, the standard deviations of points to steals are about 12:1, which is fairly close to the 16:1 ratio we earlier discussed. However, i'm not going to pretend like I know much about this kind of statistical analysis. z-scores and F-tests make me shiver.

7:42 PM  
Anonymous john said...

this is "anonymous" from first post, btw...
What i normally use in evaluating my selection is the ratio between the best in points and the best in steals. its less accurate but it makes rought cut approximations much easier. In a year, we'd probably expect that 30ppg would lead the league, and this i equate to 2.5 spg... And this is why i think it is much easier to get trades done when you are targetting steals/blocks than when you are targetting points... Especially when you consider that a guy whos averaging enough pt to get more than 2 spg, would probably average at least 11-12 ppg. (there are of course exceptions)

i actually think the 12:1 ratio would be better in judging relative worth of a player. We wouldnt put a player averaging 12 points in out lineup for his points, just as we wouldnt put a guy averaging 1 spg just for his steals. We would however, place a guy averaging 16 ppg in our line up for his points. that being said, i tried a small experiment within our league. I took the ratio between best in points and best in steals, 2nd-best in points and 2nd-best in steals, etc... and the ratio fluctuates from 14.6 to 15.6...

also, we have to think about it in the context of an entire season, not just on a game by game basis. (i guess this applies more to the roto players.) So, while day to day production of steals fluctuates a lot, in the end its the averages that matter. Of course, it is sometimes hard to swallow having to start a guy like childress for the entire season when there are always flash in the pan waiver wire wonders...

10:22 PM  
Anonymous john said...

on an unrelated note, (related to previous day's post) do you guys think shareef is still worth hanging on to? i mean, if their plan was to have shareef gain the 15 pounds before making him starter, then i might end up getting only a few weeks out of him, at most...

11:36 PM  
Anonymous bublitchki said...

Completely off topic here... but when trades happen, you gotta move quick!

D'ya think Chris Wilcox will finally have some value now that he's been traded to Seattle? I've always been a fan of his game and was somewhat baffled by his banishment to the Clipper bench. In those rare games where he was given minutes this year, he's produced some nice lines.

But the Seattle frontcourt situation he's landed in isn't exactly clear: Evans, Collison, and Fortson are all still on the roster. Will Hill give Wilcox some major burn at the 4? Evans, VladRad and Collison were all given their chance and all were found wanting. Maybe Wilcox will step in and get good minutes purely by default.

Despite all the questions, my hunch is that he'll be a worthwhile pickup. He's a onetime lottery pick and this is a contract year for him. But I'll defer the ultimate judgement on the former Terrapin to our esteemed hosts here, who wear their Maryland bias on their sleeves. Undoubtedly they can offer insights into his game that will far exceed my own half-baked analysis of his fantasy potential.

2:24 AM  
Blogger DM said...

Situations like the Wilcox -- and even the possible Darko one -- are always pretty tough. Quite likely, there's an owner in your league who has nothing to lose and a roster spot or two to play with, and he scooped up Wilcox right after news of the trade leaked.

Anyway, thoughts on Wilcox ... he belongs in that group with Stromile Swift and Melvin Ely, guys who should seemingly be fantasy studs but never get the chance because there's just something about them as actual basketball players that rubs their coaches the wrong way. Wilcox is a physical specimen, but he's prone to mental and defensive breakdowns. So at least he should fit in on Seattle. That's a really crowded front court situation, and I'd think that Collison, Swift and Petro are bigger priorites for the Sonics since they are locked in for longer. But you never know.

10:51 AM  
Blogger T-Plan said...

Good conversation and I found myself thinking the same thing last night when I learned of the trade. My thoughts . . .

Vlad Rad - if he will ever come around you'll know soon. He will be well-suited to an up-tempo game with a good point guard, a solid SG who can score, and a stud at the 4-spot. The Clips are the perfect situation for him. He'll start in place of Maggette for now, and possibly the rest of the way. Maggette's injury won't go away until the offseason regardless of rehab, so they'll need to limit his minutes like Grant Hill these days.

Wilcox - I agree he is like Ely and Swift in good and bad ways. Good raw skills but bad work habits and questionable attitude. He showed us last season that he can put up top 20 stats for a Center when starting. My guess? They'll experiment with him and Collison starting together. Swift and Petro then make for solid bench players, and they can learn from more established veterans. Wilcox has far more offensive skills than both of the 2 young guys, so the team would be crazy not to try him out at a starting spot. Reggie Evans will get traded before the 23rd, because they're desperate to bring in a PG behind Ridnour. The hottest rumor on this one has Evans going to the Bobcats, but info has been limited since the trade for Lonny Baxter. He ain't the answer, so they'll keep shopping IMO.

11:52 AM  
Anonymous Anonymous said...

Do you think someone at Yahoo reads FBB?

This was in the news and notes section for Kobe last night:

Feb 14 Bryant had just 23 points on 6-of-20 shooting, but he had three-three pointers and seven steals.
Recommendation: Seven steals is probably worth 50 or 60 points at least in your standard eight category league, so there's nothing to complain about here. He also had zero turnovers - for those who play in leagues that count them.

11:53 AM  
Blogger T-Plan said...

Nah, that comment was from Rotowire. Yahoo has agreements in place with Rotowire and Fanball to syndicate content into their fantasy leagues.

1:00 PM  
Blogger Biggs said...

The problem with ratios in regards to fantasy stats is that they are not linear. Sure 16 points = 1 steal at a Zscore of -.13 (as reflected in ESPN's PR), but at a Zscore of 1 (21pts and 1.5 stls respectively), the ratio changes to 14:1 and it will continue to change as each stat increases. You cannot extrapolate those numbers to assume that 96 points = 6 steals.

AJ, I agree with what you say. I think the key is to actually look at the standard deviation of a category. So, while the average number of points per game is equal to 16 and the average number of steals is equal to 1 (or whatever), looking at the number of points or steals at 1 or 2 standard deviations from the average will give you more information. I don't know the numbers off the top of my head, but I bet you will find that the number of steals per game trails off much more quickly than points per game. What that means is that while, on average, 16 points/game is equal to 1 steal, once you get out to the edges of reality, 32 points/game is equal to 3 steals a game.

11:46 PM  

Post a Comment

<< Home