-
Notifications
You must be signed in to change notification settings - Fork 0
/
scratchpad.txt
78 lines (60 loc) · 4.04 KB
/
scratchpad.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
Jeremy Langford problem
... Backups who are extremely effective in taking over for the starter in short spells
end up higher ranked, than players who play a lot
... so the system has to take into account how many actual games he has played
... but the current system penalizes injuries (or games not played) so heavily, it creates a ranking that benefits positions where a player stays healthy, ie.QB.
... But if a single player stays healthy at RB or WR, it is apparent that that player produces a higher win pct over QB.
1) We can determine the most effective player when they played
2) We can determine the most effective player over the season, by marking 0 for games they missed
3) We have no system of determining what value to assign when he missed a game
- We can pull a random player from the backup ranks to also add what kind of effect his injury would had on team
A. Who made the most of their opportunities
B. who was full season best performing player? ...creates a ranking full-season
C. Who should have been flex? ...should be based on full-season testing, bc costing at a full season measurement
D. How many wins did a player add, when he was starting... meaning
E. Redo most valuable player, taking into account the drop off into back territory
Did last week's results agree with Monte?
1.cleanup
2.Stephen Gostkowski problem
Why is kicker so high in rank
3. separate past performance ranking from predicted ranked (removes injured players)
Grade Ranked
4. It took me an entire day to get the rosters... It took me 15 minutes to create the proxy for the database...
...create a tool to simplify creation of the tests, and the parse the data out.
...3 tab view. 1 raw html. 1 tokenized. 1 DOM tree.
...4th tab has 3 split view with DOM tree. upper left, we have listbox where we can create a record definition. upper right, we have a DOM tree. drag and drop DOM tree to record definition. have button run the parse and put it into datagrid in 3rd split view at bottom.
5. allow monte to detect when there is no change in Stats, so why re-run it? Give error message. Create a flag to override.
6. Website spell check. should only be checking literals.
7. Histogram maker
2015/12/6
8. command line (I think I had this idea already for IIS logs to see who had errors, but someone showed me logparser)
- from [filename] /as [alias] [filename] /as [alias] [filename]
- interpreter /library library recognize file and turn it into name/value pair
- | headers [column name1] [column name2]...
- | innerjoin [one table only] ON [join conditions] /where [where conditions]
- | leftjoin [one table only] ON [join conditions] /where [where conditions]
- | rightjoin [one table only] ON [join conditions] /where [where conditions]
- | select [column index/name]
- pick the column to display
- pick a function to apply
- parser?
- | where [condition]
- | groupby [column name] [column name] [column index] /avg [column name] [column index]
/sum [column name] /max /min /count
- | orderby [column name] /desc [column order]
- | tempfile [filename]... outputs filename
9. command line histogram converter / group by
actual draft
ADP
http:https://football.myfantasyleague.com/2013/adp?COUNT=250&POS=*&CUTOFF=5&FRANCHISES=-1&IS_PPR=-1&IS_KEEPER=0&IS_MOCK=-1&TIME=
how accurate are the ADP for ranking players?
-------------------------------------------
write another monte to take in variables that lead up to a game such as
player, team, pos, depth on team, coach, stadium, time, month, injured rating,
versus same ratings
output being points
see if monte can find coefficent for each possibility at each category
lynch, seattle, rb, starter, carroll, riverhead, 1:30p, oct, probable 3 days before game.
see if there is a linear regression factor for continuous variables, for each player... since they are the one's performing.
add support for error bars
http:https://stackoverflow.com/questions/20434344/error-bar-change-upper-lower-and-center-errors-independently-of-each-other