Statistical Handicapping help pls

infinii

Registered User
Forum Member
Mar 30, 2001
1,208
0
0
Toronto, Canada
What is the easiest sport to track in terms of stats?

I'd like to build a database and try to come up with systems, etc to work off of but the task if keep a database up to date is daunting.

I know you can use the Resources page here on MJ's to get lots of data but I am looking to keep the stats myself so I can develop a program that will look for certain situations, etc.

I would think that baseball is the toughest because of all the numbers involved.

Any opinions?
 

yak merchant

Registered User
Forum Member
Nov 13, 2000
966
4
18
Texas
Get my email from Jack if you want to discuss. Because no matter what it is I've tried it. Baseball is actually the most "True" when it comes to statistics Because the true results are in the data. A hit is a hit a double is a double, etc, etc. With football and basketball, there are things you can't see in the data. Is the data skewed because of things like Hail Mary's, Fouls at the end of game, etc. I guess it all depends on what you want to try to track and what you want to build your systems off of.
 

infinii

Registered User
Forum Member
Mar 30, 2001
1,208
0
0
Toronto, Canada
You bring up a good point. Because MLB tracks errors, it is reflected in stats. Whereas in hockey, a turnover which leads to a goal scored isn't noticeable in the boxscore.

I guess the kind of system I'm thinking if is one that hilights trends so one could look a bit closer at a game. You have to admit that there are games that are just too close to cap and some that have definite advantages. The games with advantageous trends are the ones that I'd like to be pointed out to me so they can be looked at more closely.

i.e. Did you know that Patrick Roy of the Col Avs is 8-0 now after a losing game in which he gives up 5 goals? This one cashed nicely last week.


I know RobertIrish has such a system but I'm not sure how willing he'd be in sharing his experience with me. I'm not implying that he's selfish, but cuz of his other ventures...I'd understand.

I think it's interesting when others post systems or trends that they use. Wouldn't it be neat to enter that trend into a system so it could be tracked and monitored automatically?

This isn't something to be relied on soley, but simply a tool to automatically bring out attention to a situation that was profitable in the past.

Aside from stats, I'd also think it would be interesting to track line movements and public concensus. This would be interesting for ppl who believe some games are *tampered* with.
 

yak merchant

Registered User
Forum Member
Nov 13, 2000
966
4
18
Texas
I'm not an angle or system player, I use the stats to match up team versus team. And I'm just now starting to build a hockey DB. I live in Texas and I know Hockey about as good as I know Mandarin Chinese, so me trying to get value out of hockey stats is kind of a joke, but I'll figure it out. And I'm still trying to automate the download of the closing lines for analysis. Should have that one figured out soon.
 

ryson

Capitalist
Forum Member
Dec 22, 2001
1,142
9
0
IAH
Just off the top of my head: import all of the data on a nightly basis (all stats is scalar data anyway) design a report package to display it in an x y axis graph. this will show you the trend of the data, (i.e lhp vs rhp, home record vs. away record etc). You would also need the ablilty to correlate the data from each team on the fly, I would suggest a jsp system to serve up the data on the fly, then you would be able to correlate team A vs. team B depending on each attribute. The graphs may get a little funky if you are looking @ batting averages vs. ERA, just think about how you want the data presented and build it from there.

just my .02
 

infinii

Registered User
Forum Member
Mar 30, 2001
1,208
0
0
Toronto, Canada
ryson,

I'm not really concerned with the presentation of the data.

I just can't think of an easy way to import the stats nightly.

Last year i wrote a perl script to parse baseball/basketball lines from html into xml, then imported them into a database. Lost that script in a computer crash but wouldn't be hard to rewrite. Parsing stats for each game is alot more work because there are so many #'s.

And if I cant automate this, I'll forget it because I don't have time to manually key in all the relevant stats.
 

ryson

Capitalist
Forum Member
Dec 22, 2001
1,142
9
0
IAH
there are many sites with the daily data in various delimited formats, set up some code to go grab it on a daily basis from a particular place (ftp on most). Put enough logic in your code to go to various places to get the data if your initial place hurls. If you are not concerned with the presentation of the data, how do you analyze it, just curious:D .....please clue me in on how you view your data I am very interested. You can use many ways to parse the data, if Perl is your thing that's perfect, I use java -- I just live in an OOP world if ya know what I mean.
 
Last edited:

ryson

Capitalist
Forum Member
Dec 22, 2001
1,142
9
0
IAH
infinii said:
ryson,

Parsing stats for each game is alot more work because there are so many #'s.

And if I cant automate this, I'll forget it because I don't have time to manually key in all the relevant stats.


It is really just one method, about 24 lines of code in a while statement for each stat.
 
Bet on MyBookie
Top