Record Update - 1 Mar 2009
,
Labels:
in review,
ncaab week 10
Wow. The magnitude of this week's suck was greater than last week's success. New graph below.
Sunday, March 01, 2009
Subscribe to:
Post Comments (Atom)
8 comments:
maybe you should check if contrarian wagers are serially correlated?
I have to admit, your comment has me a little concerned. I might try to do a little ARMA modeling tonight to see if college hoops has been a random walk this year. If it has (and if 2007-08 was as well), that's bad from a long term perspective.
i think we have somewhat of different points here (possibly opposite).
what i was getting at was if winning % was tied to what last happened. this is somewhat out-there but i can come up with some reasons why it might be true.
a random walk would just be simple diffusion away from the mean. i would argue that this is true but with some constant for the "edge" that the contrarain gets that trends the diffusion upward over time.
i think i was implying that we could be "streak" shooters.
I knew what you meant. I took an implied leap of faith, something I shouldn't do. It clearly looks like some sort of time series model. However, to my poorly trained eye, it looked more like an ARIMA(0,1,1) model (random walk), rather than a ARIMA(1,0,0) model (serial correlation). My fiancee has to teach a class tomorrow night, so I am going to do the analysis then. I'm very interested in the results though and appreciate the comments.
Er, ARIMA(0,1,0), not ARIMA(0,1,1).
And note, that if it is ARIMA(0,1,0) with constant, than you are diffusing, but also growing.
good stuff
this could "prove" that contrarian wagers are profitable due to the growth aspect of the random walk.
why are you concerned that it could be a random walk? seems most likely that it is, but i don't see the negative in that if there still is growth.
what seems unlikely, but very easy to take advantage of is if it is serially correlated.
seems most likely that it is, but i don't see the negative in that if there still is growth.
I am concerned with your if statement being false.
Post a Comment