News:

Herr Otto Partz says you're all nothing but pipsqueaks!

Main Menu

Race strength estimation

Started by Duplode, April 01, 2012, 05:34:13 PM

Previous topic - Next topic

Which would be a better metric for the strength of a race?

Combined strength of all pipsqueaks
1 (14.3%)
Strength of top pipsqueaks
2 (28.6%)
Strength of midfield
1 (14.3%)
Some other function of pipsqueak strengths
1 (14.3%)
Something entirely different (say, pipsqueak activity)
2 (28.6%)

Total Members Voted: 7

Friker

eh? seems interesting.. let me think about it :)

Friker

Quote from: Friker on April 23, 2012, 03:38:02 AM
eh? seems interesting.. let me think about it :)

your system is "well-constructed" :) but still i would like to see some output from my thoughts. maybe in august ill write something :)

BonzaiJoe

I'm very impressed with the accuracy of that graph.

Average results are a problematic indicator because of the concept of listfillers, though. I would not like for my result on ZCT79 to be taken as an indication of my racing level, for example...

But by the looks of it, this problem doesn't undermine the algorithm.
But we can't be quite sure.


Duplode

#33
Quote from: BonzaiJoe on April 24, 2012, 12:39:03 PM
Average results are a problematic indicator because of the concept of listfillers, though. I would not like for my result on ZCT79 to be taken as an indication of my racing level, for example...

But by the looks of it, this problem doesn't undermine the algorithm.

To me it seems that listfillers should be somehow accounted for when calculating race strengths. If a top pipsqueaks spends a season sending only listfillers, his/hers participation ideally should not contribute as much for race strengths during that season. For rankings of pipsqueaks, though, it does seem fair to have a way to discount listfillers, though I have not settled on one thus far.

The real problem with these running average algorithms is related to handling different time scales. Should early races of a pipsqueak which improved slowly but steadily into a top pipsqueak weigh down the averages forever? One might think of dropping early races (say, by only including the past 60 races into the average) - but then, how do we deal with cases like Roy/Reiger?

(The above considerations made me realize that calculating dropping "extreme" results - the worst 25%, for instance - for each pipsqueak could solve some of these problems...)

Friker

Quote from: Duplode on April 24, 2012, 04:51:39 PM
Quote from: BonzaiJoe on April 24, 2012, 12:39:03 PM
Average results are a problematic indicator because of the concept of listfillers, though. I would not like for my result on ZCT79 to be taken as an indication of my racing level, for example...

But by the looks of it, this problem doesn't undermine the algorithm.

To me it seems that listfillers should be somehow accounted for when calculating race strengths. If a top pipsqueaks spends a season sending only listfillers, his/hers participation ideally should not contribute as much for race strengths during that season. For rankings of pipsqueaks, though, it does seem fair to have a way to discount listfillers, though I have not settled on one thus far.

The real problem with these running average algorithms is related to handling different time scales. Should early races of a pipsqueak which improved slowly but steadily into a top pipsqueak weigh down the averages forever? One might think of dropping early races (say, by only including the past 60 races into the average) - but then, how do we deal with cases like Roy/Reiger?

(The above considerations made me realize that calculating "extreme" results - the worst 25%, for instance - for each pipsqueak could solve some of these problems...)

This is another think I was thinkin about. Average is a very bad indicator. It should be a median or average from top 60 per cents of races.

CTG

It would be nice to see ZCT 129-149 in the stats. I think some of the ZSC2013 races were pretty hard.

alanrotoi

Is it possible to calculate zakstunts pipsqueaks performance per season?

For example an average between imaginary fastest lap and average of all final results.

It may say something about general perfomance including 1st place but for lack of stats we should do the same calculation between 1st place time and average 2nd-last place time.

CTG

Quote from: CTG on February 01, 2014, 12:28:47 PM
It would be nice to see ZCT 129-149 in the stats. I think some of the ZSC2013 races were pretty hard.

Duplode: could you please update those stats?

Duplode

#38
Quote from: CTG on February 06, 2015, 03:17:47 PM
Duplode: could you please update those stats?

Here they are! Fortunately it wasn't difficult to get the (clunky, messy, undocumented) code from three years ago to work again. Plot:



Z126-Z161:



Data, sorted chronologically...


1       "C01"   0.9689400477956864
2       "C02"   0.8169024658050857
3       "C03"   0.7478094743173973
4       "C04"   0.7949415262372359
5       "C05"   0.677639508017131
6       "C06"   0.7116337738910322
7       "C07"   0.8279904326789597
8       "C08"   0.8310508366154873
9       "C09"   0.9420818239435098
10      "C10"   1.0972778898408067
11      "C11"   1.2404899424930709
12      "C12"   1.1840474601972015
13      "C13"   1.124114155813747
14      "C14"   1.2114542271138526
15      "C15"   1.1864078905618902
16      "C16"   1.162727735690297
17      "C17"   1.171543414546985
18      "C18"   1.1089406905463877
19      "C19"   1.212109149940057
20      "C20"   1.0539799065366733
21      "C21"   1.247335334941021
22      "C22"   1.137807421789967
23      "C23"   1.0649935743157235
24      "C24"   1.1789687766726773
25      "C25"   0.9759483918260397
26      "C26"   1.2244299316696852
27      "C27"   1.0347448924325184
28      "C28"   1.0896220590430292
29      "C29"   1.186146422146186
30      "C30"   1.0322721266809785
31      "C31"   1.0980502752261478
32      "C32"   1.0881691700870795
33      "C33"   1.0591014424531762
34      "C34"   1.1968395184366045
35      "C35"   1.186675827050426
36      "C36"   1.0811583543707717
37      "C37"   1.077402305685727
38      "C38"   1.0920956472475025
39      "C39"   1.1355196539355568
40      "C40"   1.1009690283447056
41      "C41"   1.1069528673472575
42      "C42"   1.0586761961920326
43      "C43"   1.0881245289825938
44      "C44"   1.079502836318097
45      "C45"   1.1056138374370517
46      "C46"   1.123911480968403
47      "C47"   1.1487808697986865
48      "C48"   1.073530365396353
49      "C49"   1.0698461700981488
50      "C50"   0.9170253239754842
51      "C51"   1.2256388760218293
52      "C52"   1.2084359855101359
53      "C53"   1.0138805448345958
54      "C54"   1.230100830983952
55      "C55"   1.1811679701291025
56      "C56"   1.181138013114387
57      "C57"   1.095312753011811
58      "C58"   1.0410976347846719
59      "C59"   1.1889970978565068
60      "C60"   1.246705158723104
61      "C61"   1.2186523436514123
62      "C62"   1.0404809789236207
63      "C63"   1.2115048611023715
64      "C64"   0.998089447533628
65      "C65"   1.1043051801041592
66      "C66"   1.096916991872554
67      "C67"   0.958986680861299
68      "C68"   1.0337562866527779
69      "C69"   1.0087598620715927
70      "C70"   0.8650927921523183
71      "C71"   0.6838412095390317
72      "C72"   0.920230735989811
73      "C73"   0.6666372759742343
74      "C74"   0.7238444443202098
75      "C75"   0.6667446862628902
76      "C76"   0.6537072386584194
77      "C77"   0.6460955451900426
78      "C78"   0.6835887235118633
79      "C79"   0.9127183957000091
80      "C80"   0.8571497164717864
81      "C81"   0.9330118799582122
82      "C82"   1.0171968221057057
83      "C83"   0.9484466516560793
84      "C84"   0.8989854425929414
85      "C85"   0.9575127217390671
86      "C86"   1.033437523165465
87      "C87"   0.979245234382722
88      "C88"   0.9758955778420796
89      "C89"   0.9500002906190882
90      "C90"   0.9808312155955641
91      "C91"   0.90913697352392
92      "C92"   1.0336759602936316
93      "C93"   0.84678025173352
94      "C94"   1.00262565160693
95      "C95"   0.9616690341956202
96      "C96"   0.9988325365932506
97      "C97"   1.0004956921389636
98      "C98"   1.0443681022662894
99      "C99"   0.9522585176695119
100     "C100"  1.3864666915197195
101     "C101"  1.1344522542473525
102     "C102"  0.9229873062738267
103     "C103"  0.973934107144398
104     "C104"  0.9543003756122189
105     "C105"  1.0356259904068026
106     "C106"  0.8704116651381018
107     "C107"  0.9305455620003085
108     "C108"  0.9185732330463009
109     "C109"  0.8794617945694458
110     "C110"  0.9609050588121039
111     "C111"  0.9879179564228385
112     "C112"  0.8598870905243342
113     "C113"  1.049866032397705
114     "C114"  1.0198398085440483
115     "C115"  0.9288924939867218
116     "C116"  0.843910754800388
117     "C117"  0.6829431422227926
118     "C118"  0.7983255980605544
119     "C119"  0.8382726824222874
120     "C120"  0.6388754109549367
121     "C121"  0.7522345128087627
122     "C122"  0.9307755960280949
123     "C123"  0.8249698996825673
124     "C124"  1.0100723350771101
125     "C125"  0.9703784608556298
126     "C126"  0.9878406384791253
127     "C127"  1.0210024728292988
128     "C128"  1.0137533770991995
129     "C129"  0.9691319709413175
130     "C130"  0.8183069076896105
131     "C131"  0.960763406601733
132     "C132"  0.9561603424161117
133     "C133"  1.0682626021949784
134     "C134"  1.100843257219523
135     "C135"  1.0643773901560338
136     "C136"  0.8539884609442169
137     "C137"  0.9128046380265993
138     "C138"  1.1718845649163696
139     "C139"  1.07745641609288
140     "C140"  1.0582642884803737
141     "C141"  0.9688228344946141
142     "C142"  0.9564228675082869
143     "C143"  0.8956600329934347
144     "C144"  0.9702091411932694
145     "C145"  1.0715087582028076
146     "C146"  0.9864259959232726
147     "C147"  0.9639372970562176
148     "C148"  0.925672351676703
149     "C149"  0.9690298358171286
150     "C150"  1.0973689693676674
151     "C151"  1.0346525275472949
152     "C152"  1.131985734713242
153     "C153"  1.1141132736088906
154     "C154"  1.0153874087851067
155     "C155"  0.931052503803616
156     "C156"  0.9796519979983589
157     "C157"  1.0443901571811245
158     "C158"  1.0487149142639867
159     "C159"  0.9478919544339005
160     "C160"  0.8522174320964528
161     "C161"  0.8349361390430942


...and by strength:


1       "C100"  1.3864666915197195
2       "C21"   1.247335334941021
3       "C60"   1.246705158723104
4       "C11"   1.2404899424930709
5       "C54"   1.230100830983952
6       "C51"   1.2256388760218293
7       "C26"   1.2244299316696852
8       "C61"   1.2186523436514123
9       "C19"   1.212109149940057
10      "C63"   1.2115048611023715
11      "C14"   1.2114542271138526
12      "C52"   1.2084359855101359
13      "C34"   1.1968395184366045
14      "C59"   1.1889970978565068
15      "C35"   1.186675827050426
16      "C15"   1.1864078905618902
17      "C29"   1.186146422146186
18      "C12"   1.1840474601972015
19      "C55"   1.1811679701291025
20      "C56"   1.181138013114387
21      "C24"   1.1789687766726773
22      "C138"  1.1718845649163696
23      "C17"   1.171543414546985
24      "C16"   1.162727735690297
25      "C47"   1.1487808697986865
26      "C22"   1.137807421789967
27      "C39"   1.1355196539355568
28      "C101"  1.1344522542473525
29      "C152"  1.131985734713242
30      "C13"   1.124114155813747
31      "C46"   1.123911480968403
32      "C153"  1.1141132736088906
33      "C18"   1.1089406905463877
34      "C41"   1.1069528673472575
35      "C45"   1.1056138374370517
36      "C65"   1.1043051801041592
37      "C40"   1.1009690283447056
38      "C134"  1.100843257219523
39      "C31"   1.0980502752261478
40      "C150"  1.0973689693676674
41      "C10"   1.0972778898408067
42      "C66"   1.096916991872554
43      "C57"   1.095312753011811
44      "C38"   1.0920956472475025
45      "C28"   1.0896220590430292
46      "C32"   1.0881691700870795
47      "C43"   1.0881245289825938
48      "C36"   1.0811583543707717
49      "C44"   1.079502836318097
50      "C139"  1.07745641609288
51      "C37"   1.077402305685727
52      "C48"   1.073530365396353
53      "C145"  1.0715087582028076
54      "C49"   1.0698461700981488
55      "C133"  1.0682626021949784
56      "C23"   1.0649935743157235
57      "C135"  1.0643773901560338
58      "C33"   1.0591014424531762
59      "C42"   1.0586761961920326
60      "C140"  1.0582642884803737
61      "C20"   1.0539799065366733
62      "C113"  1.049866032397705
63      "C158"  1.0487149142639867
64      "C157"  1.0443901571811245
65      "C98"   1.0443681022662894
66      "C58"   1.0410976347846719
67      "C62"   1.0404809789236207
68      "C105"  1.0356259904068026
69      "C27"   1.0347448924325184
70      "C151"  1.0346525275472949
71      "C68"   1.0337562866527779
72      "C92"   1.0336759602936316
73      "C86"   1.033437523165465
74      "C30"   1.0322721266809785
75      "C127"  1.0210024728292988
76      "C114"  1.0198398085440483
77      "C82"   1.0171968221057057
78      "C154"  1.0153874087851067
79      "C53"   1.0138805448345958
80      "C128"  1.0137533770991995
81      "C124"  1.0100723350771101
82      "C69"   1.0087598620715927
83      "C94"   1.00262565160693
84      "C97"   1.0004956921389636
85      "C96"   0.9988325365932506
86      "C64"   0.998089447533628
87      "C111"  0.9879179564228385
88      "C126"  0.9878406384791253
89      "C146"  0.9864259959232726
90      "C90"   0.9808312155955641
91      "C156"  0.9796519979983589
92      "C87"   0.979245234382722
93      "C25"   0.9759483918260397
94      "C88"   0.9758955778420796
95      "C103"  0.973934107144398
96      "C125"  0.9703784608556298
97      "C144"  0.9702091411932694
98      "C129"  0.9691319709413175
99      "C149"  0.9690298358171286
100     "C01"   0.9689400477956864
101     "C141"  0.9688228344946141
102     "C147"  0.9639372970562176
103     "C95"   0.9616690341956202
104     "C110"  0.9609050588121039
105     "C131"  0.960763406601733
106     "C67"   0.958986680861299
107     "C85"   0.9575127217390671
108     "C142"  0.9564228675082869
109     "C132"  0.9561603424161117
110     "C104"  0.9543003756122189
111     "C99"   0.9522585176695119
112     "C89"   0.9500002906190882
113     "C83"   0.9484466516560793
114     "C159"  0.9478919544339005
115     "C09"   0.9420818239435098
116     "C81"   0.9330118799582122
117     "C155"  0.931052503803616
118     "C122"  0.9307755960280949
119     "C107"  0.9305455620003085
120     "C115"  0.9288924939867218
121     "C148"  0.925672351676703
122     "C102"  0.9229873062738267
123     "C72"   0.920230735989811
124     "C108"  0.9185732330463009
125     "C50"   0.9170253239754842
126     "C137"  0.9128046380265993
127     "C79"   0.9127183957000091
128     "C91"   0.90913697352392
129     "C84"   0.8989854425929414
130     "C143"  0.8956600329934347
131     "C109"  0.8794617945694458
132     "C106"  0.8704116651381018
133     "C70"   0.8650927921523183
134     "C112"  0.8598870905243342
135     "C80"   0.8571497164717864
136     "C136"  0.8539884609442169
137     "C160"  0.8522174320964528
138     "C93"   0.84678025173352
139     "C116"  0.843910754800388
140     "C119"  0.8382726824222874
141     "C161"  0.8349361390430942
142     "C08"   0.8310508366154873
143     "C07"   0.8279904326789597
144     "C123"  0.8249698996825673
145     "C130"  0.8183069076896105
146     "C02"   0.8169024658050857
147     "C118"  0.7983255980605544
148     "C04"   0.7949415262372359
149     "C121"  0.7522345128087627
150     "C03"   0.7478094743173973
151     "C74"   0.7238444443202098
152     "C06"   0.7116337738910322
153     "C71"   0.6838412095390317
154     "C78"   0.6835887235118633
155     "C117"  0.6829431422227926
156     "C05"   0.677639508017131
157     "C75"   0.6667446862628902
158     "C73"   0.6666372759742343
159     "C76"   0.6537072386584194
160     "C77"   0.6460955451900426
161     "C120"  0.6388754109549367

CTG

Quote from: Duplode on February 07, 2015, 06:38:53 AM
Quote from: CTG on February 06, 2015, 03:17:47 PM
Duplode: could you please update those stats?

Here they are! Fortunately it wasn't difficult to get the (clunky, messy, undocumented) code from three years ago to work again.

Thank you very much! Great work!

zaqrack

#40
We had a great season start, the usual summer wind-down and a miserable season finish. Thanks to CTG's tantrum and my move back from China for the last one.  >:(

zaqrack


CTG

Quote from: zaqrack on February 09, 2015, 09:34:57 AM
Thanks to CTG's tantrum and my move back from China for the last one.  >:(

Look on the bright side: at least it improved the stats of Akoss Poo. :D

However, my own participation and activity had also an effect on the tendency. It will be published from home.