NAQT "Power Percentage" Stat Quirk

[quizbowllee]

I'm not sure if this is an easy fix, but an issue with NAQT's "Power Percentage" stat was brought to my attention. Like many coaches (I think), I have some players who are obsessed with statistics. Since we compete in ASCA (Alabama Scholastic Competition Association) events, many of the tournaments we compete in do not use powers/negs.

After our recent county tournament (IS-214A), Brodie mentioned that his "Power Percentage" was about to go down. I asked him why that would happen, since we don't use powers. He informed me that tossups gotten at a tournament that doesn't use powers still count as "unpowered tossups" on NAQT's stats page. I thought there was no way that was accurate. But, upon looking at the stats, he appears to be correct. I'm confused as to why that's the case, as it does seem to seriously negatively skew the stats for players who participate in tournaments where powers aren't used.

Can this be fixed? It it worth fixing? Has anyone else noticed this?

Just wondering.

[Important Bird Area]

Thanks for letting us know about this- our tech team will look into it.

[jonah]

I agree that this is weird. It's definitely technically possible to fix this, but I'm not sure it's a good idea.

If we changed things so that only tournaments that used powers were considered in the power percentage calculation, then the table would be internally inconsistent. For example, when playing for West Point High School, Brodie currently has 210 powers and 816 gets, leading to a power percentage of 20.5%. If you only consider tournaments that used powers, it's 210 and 475, leading to a power percentage of 30.7%, but we'd presumably still display his actual totals (210 and 816), possibly leading someone to wonder how we got 30.7% from 210 and 816. In theory, we could display a separate column for "gets at tournaments that had powers," but it's kind of silly.

This issue isn't unique to powers and power percentage; some version of it is likely to come up with any cases in which different scoring rules are used. For example, if a tournament used scoring values other than 15, 10, and –5, then their PPTUH would potentially be based on unreasonable comparisons. We "address" this by "reinterpreting" data into NAQT scoring rules when we aggregate results from multiple tournaments, but that's arguably wrong and can certainly produce some odd results.

There are also performance concerns. Without getting too technical, if we have to look in "more places" for information to build those tables — specifically, if we have to look up the scoring rules of each tournament being aggregated — it might get significantly slower and put significantly more stress on the database. Our stats-related pages and tools are already the most database-intensive part of our site, and I really do worry about making it worse.

I'm open to ideas on how to improve this, but I'm not convinced that it makes sense to do something or, if so, what. You can still click through to players' detailed history and get the tournament-by-tournament breakdown, which should produce faithful results, except to the extent that a few tournaments use different rules at different times within the same tournament (sigh).

[Halinaxus]

Has anyone else noticed this?

Just wondering.

I've known about this for a while. In Minnesota we have an annual speedcheck tournament (RAT-RACE) with 40-tossup games and no powers or negs. Because of this, it is common for the best players at Minnesota schools to have depressed power percentages due to racking up hundreds of tens at RAT-RACE over the years, similar to how Brodie's power percentage is depressed from playing ASCA tournaments. My own power percentage from high school is 8.7% lower than it should be (a non-trivial difference given how many NAQT tournaments I played) because nearly 300 of my tens came at RAT-RACE.

As Jonah points out, this is far from the only issue inherent with combining stats from as many tournaments as NAQT does into one database. For instance, theoretically there should be some sort of system accounting for the difference in difficulty between tournaments, but currently A sets are treated the same in NAQT's database as HSNCT (presumably because there isn't a great alternative).

I get around the above issues by keeping a personal spreadsheet for my stats. Perhaps that would solve your issue?

[meebles127]

Perhaps the true lesson to be learned here is that we should not place so much emphasis on individual stats.

[quizbowllee]

Thanks for all the replies. I wasn't sure if this would be an easy fix (clearly it isn't). Or if it's really even worth fixing (that's debatable, but I'm leaning towards "no" based on the apparent difficulty of doing so). I was surprised that it wasn't already a topic of conversation.

Perhaps the true lesson to be learned here is that we should not place so much emphasis on individual stats.

I guess. I personally don't put that much emphasis on individual stats. But, some of my players - and A LOT of other players around the country - do. Until this year, I never gave individual awards at our invitationals and - until a few years ago - often didn't even keep individual stats at all, unless it was an NAQT event. I'm still of the mind that this is a "team activity" first and foremost. However, many, many players and coaches over the years have requested that we start giving individual awards (and complaining that we didn't), so we starting doing that this year.

Regardless of the "emphasis," I still think that accuracy is important. I think this is another argument for just USING powers/negs. Unfortunately, we just haven't had the necessary votes and support for it in ASCA. So, for the time being, the majority of tournaments we attend will be "power-less" and our players' "power percentage" stats will be inaccurately deflated as a result. Is that a particularly big deal? To some, perhaps it is. I'm fairly indifferent, myself. But, it was brought to my attention and I thought I'd see what could be seen about it.

[AKKOLADE]

Perhaps the true lesson to be learned here is that we should not place so much emphasis on individual stats.

On the other hand, if we're doing individual stats, we should at least do them correctly.