cricrate: new cricket ratings website

[weldone]

lol, almost feels like you moved from a more rigorous to a less rigorous field

not necessarily

Some Financial Engineering jobs can be relatively light (e.g. if portfolio is light on exotic derivs); and data scientists do some heavy stuff in some companies (e.g. Google)

 

[viriya]

On Dravid's 180, I do think it's slightly underrated partly because even though I give credit for comebacks, I calculate it based on the % difference between the two first innings scores, which ignores the magnitude of the runs deficit.
Aus made 445 and India made 171 (38%)

Hypothetically if in another match Aus makes 250 and India gets bowled out for 95 (38%), I consider both situations as similar. The reason is for ratings purposes ratios are much safer to work with than raw differences to avoid edge cases giving too much credit (say two first innings scores of 700 and 150 - the comeback factor would outweigh the whole formula if I used raw difference here).

Anyway, I do think the actual run difference matters, so going to make some changes to include it to a certain degree. This will incidentally also help Laxman's 281 rating.

Made some more tweaks to consider actual runs behind in 3rd innings and runs to chase in 4th innings to give more credit to greater deficits. Unfortunately Dravid's 180 still misses out in the top 1000 but Laxman's 281 jumps from #77 to #16.

Interesting but I don't love it. When there are 3 (4) results possible, I don't think you can call Test 2 a decider when there are more results possible ( NZ win, draw or tie) that would make it a non-decider. This basically rewards those who improve as the series goes on and play in smaller 2 or 3 test series then the countries who play more 4 or 5 Tests Series.

Reduced credit for decider performances significantly.

Also fixed some bugs in the runs percentage and match status factors.
Some of the major changes:

Batting performances:
cricrate | Performance Ratings - Test Batting
Mahela Jayawardene's 123 jumps 140 places to #24
Brian Lara's 153* jumps 5 places to #1
Justin Langer's 127 jumps 242 places to #61
VVS Laxman's 281 jumps 61 places to #16
Dave Nourse's 93* jumps 205 places to #58
Lindsay Hassett's 102* jumps 36 places to #14
Saeed Anwar's 188* drops 19 places to #27
Jeremy Coney's 111* jumps 18 places to #8
Brian Lara's 196 drops 53 places to #78
Kumar Sangakkara's 156* drops 2 places to #3
Aravinda de Silva's 143* jumps 68 places to #35
Steven Smith's 199 drops 19 places to #29

Bowling performances:
cricrate | Performance Ratings - Test Bowling
Jermaine Lawson's 6/3 drops 15 places to #21
Imran Khan's 8/60 drops 55 places to #80
Shoaib Akhtar's 6/30 jumps 116 places to #53
Malcolm Marshall's 7/22 drops 18 places to #28
Vernon Philander's 5/15 jumps 77 places to #43
Andy Caddick's 5/16 jumps 85 places to #54
Lance Gibbs's 8/38 drops 21 places to #36
Frederick Spofforth's 7/44 jumps 15 places to #14

 

[Daemon]

I can definitely live with Lara's 153 being #1

 

[Spark]

I noticed there's no decay for players who miss a Test through injury (which is why Steyn is still ranked #1), what's the thinking behind that?

 

[viriya]

I noticed there's no decay for players who miss a Test through injury (which is why Steyn is still ranked #1), what's the thinking behind that?

The initial thought process was that it was unfair to penalize players for missing a match here and there for rest or injury. To avoid long-term injured players there's a Time window where they have to have played a test to show up on this current list.

It's also because indivial innings ratings are used to rate current ratings and adding a separate decay factor messes a bit with the performance - current - career ecosystem.

 

[Spark]

I agree with that, but Steyn has barely played in over a year. Seems odd that he would still be ranked #1.

 

[viriya]

I agree with that, but Steyn has barely played in over a year. Seems odd that he would still be ranked #1.

The window is actually 1 year. He played a Test match in November.

 

[Spark]

Oh yeah I'm aware, which is why I'm wondering if there's a better way to do it than a hard cut-off.

 

[Daemon]

It's hard imo. Steyn on Day 2 of the test before he fractured his shoulder looked as good as he's ever been. On the other hand some people come back from injuries looking horrible.

Also, don't past performances diminish with time anyways? In that case the injured player is already being penalized in a way because he doesn't have the weight of recent performances to back his rating.

 

[viriya]

It's hard imo. Steyn on Day 2 of the test before he fractured his shoulder looked as good as he's ever been. On the other hand some people come back from injuries looking horrible.

Also, don't past performances diminish with time anyways? In that case the injured player is already being penalized in a way because he doesn't have the weight of recent performances to back his rating.

The diminished returns only happens on matches that he played. So that November test is his most recent match with the same weight as a non-injured player's more recent last test.

It probably makes sense to add a small decay based on each match missed - even if a player is rested or dropped and not injured he still takes time to get back to "match ready" once he gets back.

 

[indiaholic]

I think there should be a decay factor built in.. Sometimes players may miss games because the team doesn't feel they are good enough for the conditions and may be dropped.. Definitely need a penalty for that. Plus remaining fit is pretty much in the top 3 attributes for a sportsperson. Won't complain if people drop in rankings due to injuries.

 

[Days of Grace]

You were discussing support and partnerships earlier. I have a factor in my ratings that counts the highest partnership a batsmen was involved in. Therefore, innings like Andrew Jones in support of Crowe and Watkins in support of McCullum get credit.

 

[Daemon]

The diminished returns only happens on matches that he played. So that November test is his most recent match with the same weight as a non-injured player's more recent last test.

It probably makes sense to add a small decay based on each match missed - even if a player is rested or dropped and not injured he still takes time to get back to "match ready" once he gets back.

Instead of basing it on each match missed, would basing it on time make more sense? For example NZ have ages off between now and their next Test iirc, so it would make sense for their performances now to lose that little bit more of relevance come 2018, even though none of the players have missed a game.

 

[viriya]

Instead of basing it on each match missed, would basing it on time make more sense? For example NZ have ages off between now and their next Test iirc, so it would make sense for their performances now to lose that little bit more of relevance come 2018, even though none of the players have missed a game.

That would indirectly benefit the Big 3 because they tend to play more Tests. I don't think a time based decay is fair.

 

[OverratedSanity]

That would indirectly benefit the Big 3 because they tend to play more Tests. I don't think a time based decay is fair.

Should be the number of matches his country plays that he misses.

 

[viriya]

You were discussing support and partnerships earlier. I have a factor in my ratings that counts the highest partnership a batsmen was involved in. Therefore, innings like Andrew Jones in support of Crowe and Watkins in support of McCullum get credit.

I recall you had a factor that gives more credit to a top score if the second highest score is low?

This is something I haven't really thought through.. run-making becomes easier if you have a steady partner right? So why should you get more credit for being second fiddle to a big partnership? I guess in terms of winning/drawing a match a big partnership matters.

 

[viriya]

Should be the number of matches his country plays that he misses.

Yea so a pure match based decay would do that effectively.

 

[Daemon]

That would indirectly benefit the Big 3 because they tend to play more Tests. I don't think a time based decay is fair.

The match based decay hurts the big 3 as well then using the same logic.

 

[OverratedSanity]

The match based decay hurts the big 3 as well then using the same logic.

As it should if the player is missing games for his team.

The time delay punishes the player for no fault of his own.

 

[Daemon]

As it should if the player is missing games for his team.

The time delay punishes the player for no fault of his own.

Neither through any fault of their own. An injured player gets punished more if he plays for big 3 compared to Bangladesh.