Ping: craig_p

[ur_land]

My series recently had a discussion of how much our teams have improved and consequently how to measure improvement, but we were a little limited by the change in the hattrick rating systems and hatstats ratings. I know that the ratings are roughly doubled, but I know that's not exact, and I want as exact a conversion as possible from the old ratings to the new ratings.

craig, I know you've done a lot of work with ratings, can you (or anyone else) shed any light on this?

[Lone Star]

My series recently had a discussion of how much our teams have improved and consequently how to measure improvement, but we were a little limited by the change in the hattrick rating systems and hatstats ratings. I know that the ratings are roughly doubled, but I know that's not exact, and I want as exact a conversion as possible from the old ratings to the new ratings.

craig, I know you've done a lot of work with ratings, can you (or anyone else) shed any light on this?
I created a spreadsheet a couple of years ago that converted the text match ratings into a numeric value (before HatStats and the term 'NSI' ran rampant over the land). When the rating system changed I tried to figure out the conversion factor myself, but never got it exactly. The closest I got before giving up was to multiply/divide the total rating by 1.76.

As an example, in my last league match I got these ratings:

Midfield: formidable (low)
Right Side Defence: formidable (very low)
Central Defence: excellent (very low)
Left Side Defence: solid (low)
Right Side Attack: passable (very low)
Central Attack: solid (very low)
Left Side Attack: weak (very high)

According to my spreadsheet (and HatStats) this is a 252 rating. Divide by 1.76 and you get a rough estimate of 143 as the old rating. However, then I changed my HT preferences to show the old-style ratings:

Midfield: mythical
Right Side Defence: mythical
Central Defence: titanic
Left Side Defence: world class
Right Side Attack: brilliant
Central Attack: world class
Left Side Attack: excellent

I plugged this into my older spreadsheet and came up with 128. So I'm definately off somewhere with my conversion factor. However, by toggling to the old ratings and using the old spreadsheet I have an apples-to-apples comparison to older matches.

Looking at HatStats I guess I've improved. The last season of old ratings my club averaged 116.9, with an all-time high of 135 (in a MOTS).

Not sure if any of this helps you, though.

[Craig P]

They haven't exactly doubled, it's more complicated than that; I wouldn't bother trying to do a direct conversion (you can't do a direct multiplication, there's an additive factor as well). The two most feasible alternatives IMO are, go back to your old matches and update them to the new rating, or parallel calculate NSI from the current data. The parallel calculation is very easy -- take the new sector rating, subtract 1, divide by 2, truncate, then add 1, and that should give you the old-style rating. You should double-check this to make sure my off-the-cuff math is right -- I can confirm when I get a chance to check my code for doing this on the fly.

[ur_land]

Darn. Thought there would be an easy answer. Thanks anyway for the help guys--I knew there was a reason I supported both of you!

[Pal11]

My series recently had a discussion of how much our teams have improved and consequently how to measure improvement, but we were a little limited by the change in the hattrick rating systems and hatstats ratings. I know that the ratings are roughly doubled, but I know that's not exact, and I want as exact a conversion as possible from the old ratings to the new ratings.

Out of curiosity, are you working on some kind of analysis, or just looking to convert old ratings?

[ur_land]

Out of curiosity, are you working on some kind of analysis, or just looking to convert old ratings?

Yes.

I was actually going to do some historical stuff for all the teams in our series (including a few of the old ones), and try to put some graphs together, with the main point of making an ass out of Evil Runtities (since he's my biggest rival and all). But I really don't want to go back and go through each of the individual games for all the teams to get the hatstats ratings for season 22 and earlier. Since we have an average (in the old ratings) on the hatstats page already, I was hoping to tap into the brainpower around here without having to stress myself too much. If you want to work on a conversion, pal, I'll promise to do any analyses you want...... :eek:

[Pal11]

Yikes, that looks like a lot of work. :eek: I think it'd much easier to make fun of Runts by looking at where he is and where Cerrano is, considering they started around the same time. I think he's really got his act together now, with a solid training program and the patience to wait till his trainees hit formidible or higher before selling. I can't get him to add that 10th trainee, though... ;)

BTW, is this more of a friendly rivalry between you two or is it for real? I remember you two got into it last year before one of your matches.

[ur_land]

It's a friendly rivalry that sometimes spills into a real one (usually only when he trash talks my team for sucking while conveniently forgetting that his is only marginally better than mine despite him starting 3 seasons before me). I think of Runts as that younger brother-in-law that you don't mind seeing during the holidays, as he can be fun, but by the end of them you want to kick his ass.