ISC - NSA rating comparison

[Bantman]

Just curious - how do people's NSA ratings compare to their ratings on ISC? I played in tournaments for a couple years (several years ago), but now play pretty much exclusively on ISC, and am wondering how my ISC rating would translate into a new NSA rating, were I to return to live-tourney play.

[Yarn Chief Sr]

Long story short, for ISC ratings above 1300, multiply by .92 to get an approximate NSA rating.

There's a lot more magic voodoo math mumbo jumbo to be done, but basically:

ISC spans from 0 - 2200 with the best legit player rated 2045 (Weinstein NSA 1897).
NSA spans from 200ish (a rare case, with 500 floor) - 2100 (these days) with the highest rated player at 2047 (Richards). (US is Cappelletto at 2009.)

Others I've tried before arriving here: subtract 500 from ISC and multiply by 1.25, but that doesn't seem to be nearly as right at the top of the ladder because all the best players are bunched together and constantly switching places. I used to say a good rule of thumb is to simply subtract 200 (again, for ratings above 1300). You'd have to get a whole lot more data to get a more acurate conclusion, especially since ISC ratings are way more volatile than NSA ones.

In a nutshell, I'd guess an okay approximation is:

ISC NSA
1300 1196
1350 1242
1400 1288
1450 1334
1500 1380
1550 1426
1600 1472
1650 1518
1700 1564
1750 1610
1800 1656
1850 1702
1900 1748
1950 1794
2000 1840
2050 1886
2100 1932

But imagine that .92 actually squeezing closer and closer to 1 the closer you get to 1300 (by what factor I don't know -- the multiplier should probably be derived as a function of your rating itself, much like in that awful gameshow with the briefcases of money that Shall Not Be Named, in that the multiplier used is constantly changing there too, and a function of what's left -- but that's even hairier).

Yes, I'm bored.

[dugy1001]

Keep in mind how differently people treat ISC than actual tournaments. I do ridiculous things on ISC because it doesn't matter--my rating on there usually fluctuates from 1800-2000 (high of 1989, 1896 right now). To me, ISC rating has 0 importance. Even among NSA players, there are people that cheat, people that pay no attention (me), people that play ridiculously (me), etc.

Ryan's math looks like an awesome approximation though. However it underrates many near the top end of the spectrum. Consider...

Name/ ISC/ Ryan's NSA Approx./NSA

Sam Rosin/1896/1790ish/1875
Jesse Day/1827/1680ish/1875
Conrad BB/1933/1780ish/1885
Frank Tangredi/1699/1564/1660

But then again

Ryan Fischer/1879/1730ish/1607

Perhaps you are underrated Wonderfish?

[Yarn Chief Sr]

Yeah, I don't get to play in nearly as many tournaments as you yanks -- that's my best explanation for that one.

The phenomenon where high-rated players are underrated at times is because of that ISC volatility. Conrad's on a downswing, usually rated 2050 or so, but he also plays a lot of blitz which inflates ratings, and explains why Jesse's rated a bit lower on ISC and he and Conrad are rated the same in NSA. Maybe take the average of your last 10 ratings and your peak, and that would make things a teeny bit more accurate. For what it's worth.

[synodalhaj]

Probably the biggest reason for divergence between ISC and NSA rating is that people can pick and choose both the time control and the opponents. Being able to pick your opponents and play only people you choose to is quite different from a tournament where you are automatically paired against opponents.

There are other issues, including cheating, not cheating but playing cheaters, etc. etc. which factor into ISC rating. I mostly like to play players who I know (or know of) on ISC at any rate.

[edhorch]

I think the table is pretty close ( at least for my ratings) but I use ISC as a practice area, and I play mostly void games--Zyzzyva is great for learning words, but it's nice to use ISC for trying out new strategies and such. Also, for some reason (probably just my own perception without a sufficient statistical sample), drawing on ISC seems to be more feast-or-famine than real-life games. It seems like I'll get bagged ten games in a row on ISC then I'll draw the bag on my next ten opponents, and that doesn't seem to happen in club or tournament games. So, my rating on ISC tends to fluctuate between 1200 and 1400.

Like a lot of people, I don't really care about my ISC rating.

[JasonV12]

I'm rated about 1400 on ISC and 1021 from the NSA... I don't know if this is very helpful but whatever...

[cesar]

i usually play better in real life than on isc, but i mostly use isc for blitz games, although i haven't been there much. i'm around 1850 there on average, my nsa rating hovers around 1700.

[dacrON]

I have all the rating, along with drrating and kunjoos on ISC, but our NSA ratings are usually in the 1700s or 1800s (or in kunjoos case perhaps even higher)

[JackJackson]

what of the 1300-1500 nsa rated players who hover around the 1900-2000 range on isc ? nathan james (volucris) comes to mind, many more........

[ar-raqis]

In my personal case, since you asked, it's a queer mixture of:

a) not getting to enough tournaments to boost my NSA rating
b) environmental factors (8.5 years of regular exposure to the ISC environment, tens of thousands of games there vs. only a couple hundred lifetime live games -- my skills such as scoring and tracking, as well as psychological centeredness in the tourney environment, are not at the level of my anagramming, word knowledge and board vision)

As far as ISC -> NSA conversion, I've used a wide variety of alternative formulas, such as (x-300)*1.1, but I don't think any formula, linear or otherwise, is likely to fit, given the environmental factors, differing skill sets, and other complications, most mentioned in this thread. The degree to which blitz inflates or deflates ratings is highly variable, and I'm still trying to figure that out.

About a year ago, I did a study of ISC vs. NSA ratings among players with ISC usernames listed on badqoph. I lost the data in a hard drive crash, but from what I remember, the most ISC overrated/NSC underrated players were me and Jarett Myskiw. (Some others were Samantha Southard, Brian Leavitt and Joey Krafchick.) My NSA rating has gone up by enough since then that I would no longer be in the mix though.

There were also a couple of examples of players who performed significantly worse on ISC than IRL, perhaps due to the reverse of my own environmental issues. I can't recall who these were.

-Volubag

[joethebar]

the joys of cyberspace... i find it amusing how "good" players are online. in a given tourney, i am amazed at how conservative and uninventive the same players are. i am not talking about the elite BTW.
i play even better in real life, and i suspect that a lot of people are using assistance. who really cares? they are the ones who suffer at the tourneys; not me. i am always a success and often in the money and they don't understand why they do so poorly. it's because they are not doing the homework; developing their brains, studying strategy etc.
the isc ratings don't mean much unless both a players ratings might indicate some integrity.

[3dnf]

Ted Barrett NASPA: 847 ISC: 1300

Now 956 NASPA and 1630 ISC

[CanadianBacon]

Yeah, my NASPA rating is 1171, while on Isc my peak rating was 1558. And I play tons of fast games, 3-5 minutes long. So if anything my isc rating might be innacurate. I would probably win more games that are 25 minutes long. I lose a lot of games from time-out (but it doesn't deter me from playing speedy games lol).

I haven't played enough tournaments for my rating to boost up. And losing games by 1 point ugh lol, even the tiniest mistake made a huge difference in my rating adjustment.

I think one of the biggest factors is challenge void games. No penalty for playing non-words can make your rating higher than it should be. Very few games in the seek graph use challenge mode, though, and bascially none of the naspa players in my area play on isc. So what can I do.

[soccerguy]

Unfortunately, Canada is not always an ideal place to gain rating quickly. Best way to do that is to study up, and then go to an open tournament in the States like the Dallas Open or the California Open (or to Toronto, if they're still doing theirs).

I strongly encourage you to NOT play void games if you want to improve in live Scrabble. Best thing for you to do would be to adjust your variables on ISC so that you are able to send out seeks to more people, and conversely see more seeks that other people send out.

[CanadianBacon]

I have it set to accept all challenges. I usually use the graph when I'm looking for a game. I would like to play more challenge games to get more comfortable with the concept of phony words, and when its a good idea to use them (I've spent a lot of time finding out when NOT to use them heh).