Measuring Lin's scoring consistency (as well as that of other Rockets)

Wow, that actually reached 7 pages when i pasted it in MS Word. I'll give it a read later. :grin:

 

the median would be more important than the mean here. Also it would be better to look at a scatter plot to see the shape of the distribution. If he is actually showing inconsistency the plot will likely be multimodal. But then it becomes as complex as the Jabberwocky's formidable lineup metrics, which 99% of the statistics for middle school kids on CF failed to comprehend.

What we really need is for Clutch to write an article telling us that stats is math, and Jeremy should sit behind Patrick because the data is clear. LOL banned?

 

what must I do to read this?

My attention span is very bad

 

I took an AP stats course in high school so i may be a little rusty, but if you're open to some suggestions, i can try and give you a possible significance/hypothesis test or other methods (though it'll probably have some big holes).

You could calculate a mean "shake" variable from a random sample representative of all players that play +25 minutes a game for each stat you choose to analyze (breaking it down by position and mpg might not be a bad idea, but then again, that kind of grouping would also assume similar roles per position when obviously, Ryan Andersen at the 4 has a much different role than Tjones, though this could also be an outlier and may not be significant in you analyses) . Get a mean and figure out what percentile all of our rockets players (that avg. 25+ minutes a night) land in for each stat. You could just get the percentiles for a specific player and his important stats (i.e. you might do steals, assists, and rbs for Pbev and points for Harden) and ignore others. If everything that i said is viable, then this analysis could compare our individual player consistencies with each other as well with the rest of starter (granted your shake variable is an accurate representation of consistency)

This could also be done for players that play less than 25 minutes and thus you could create reference for all of our bench players and their important stats. I would really be curious as to how consistent/inconsistent casspi's, brook's, and garcia's 3 point shooting is.

Not sure too much about an actual hypothesis test now that i think about it, but a chi-square test of homogeneity comparing players with similar roles could provide some insight, I'm pretty sure it could be useful but then again, i also don't remember how to do a chi-square test, lol

Also, a possible flaw that you might want to consider in your shake variable is that inconsistent play might not be a bad thing. A player that usually plays like crap most of the season but suddenly starts playing well might have a high shake variable, that might explain tjones' high shake variable. His recent play over last 10-15 games has been excellent, but the 10-15 games before that he was just good, so it could be a possibility

As i'm sure you know, statistics is always uncertain, and i'm rusty enough that i'm not fully confident in anything that i just wrote, so if you see flaws, alert me, and i might try some new ideas.

Main goal is to just help bounce some ideas for some more skilled people to take an advantage of, even a wrong idea good inspire a great idea. Hope this is helpful in some way!

 

Body language and facial expression reading - I find this must be regularised for cultural differences. for example, Asians are generally more reserved and less demonstrative. Therefore, many see this as "less engaged". Blacks are more expressive. But don't let false bravdo fool you into thinking someone is more engaged when in fact they are hiding behind a facade. That said, Linsanity was a period of high expressiveness from Lin, which was more an outlier than anything, so one cannot use that as a yardstick to measure his engagedness.

I kinda laugh when someone says Lin is meek, given all the odds that he has overcome, he is the last thing from meek. He is just not the chest thumping, look at me type that is prevalent in today's NBA which fans equate to as "aggressive". Lin plays his best when he is under control and almost languid in his facial expressions. When he is highly expressive, I find he is actually too jacked up and will turn the ball over too much.

 

I tried reading this while throwing my preconceptions out the window, but the conclusion to me keeps leading to "Lin isn't inconsistent, he's just consistently mediocre", following the stats to their conclusion. I'm not saying that's even what it is, but that is what you have me reading.

Summarising what the paper is saying, Harden is consistently great, Parsons is good but errs on inconsistency (although as said, he tends to have some wildly good games to skew his upper end), Howard is consistently good, Lin is consistently mediocre, Jones is inconsistently mediocre and the rest are all over the place.

 

None of Team 3rd and 4th scoring options can be consistent.

Plain and Simple.

Lin will be more consistent if he takes 15+ shots per game.

Haters complain that only when Lin plays well.

 

I also arrived at this conclusion. I'll go redo the gs method later. However, I also found that Parsons performance per 36 is on par with lin. He just gets more minutes. Howard and harden are our only two star level producers.

 

OP - don't underestimate your knowledge based on the two years of basketball, you've clearly displayed that you know a lot more about basketball than you think (and more than most members of this forum).

 

Yeah, the chi-square test was just bs, realized it after posting, but thanks for seeing some value in a not great post. I do have a couple of ideas to add to your response so here they are:

You definitely DO NOT want to compare every player in the league, too many people that play way to little minutes, (although i think you know that and it might just be that i don't know what a python script is). There are probably only about 180 players in the nba that average 25+ minutes (multiply 5 starters +1 sixth man times 30 teams) and you don't want to do more than a certain percent of all players(normally this would be the 10 percent rule, so that would be about 20 players, but i'm not sure this percentage applies very well to the nba). So really, the work could be very, very little, and the increased variation due to some sampling error might be a good thing (crazy idea! definitely not learned in my stats class!) as it could more clearly indicate trends that deserve exploration. A 20 player population is feasible for a large variety of stats, and again, this population has already been specified to address our main players so this size could be a great representation all starter/heavy-minute-players.

Again, just bouncing ideas and the main goal of this replay was to spark a more practical, time saving method of analysis, one that could be feasible and short and sweet

 

Please look this up . I am correct. I am sure. Google it. Non parametric tests are use when the distribution is not Gaussion (SIC), The poster is not using a non-parametric test. The semi-interquartile range is used for percentiles as the measure to express the spread of scores around the MEDIAN not the mean.

 

<iframe width="853" height="480" src="https://www.youtube.com/watch?v=nCFsOtSZpto" frameborder="0" allowfullscreen></iframe>

 

https://www.youtube.com/watch?v=nCFsOtSZpto