Smartgambler
Pro-Punter

Go Back   OZmium Sports Betting and Horse Racing Forums > Public Forums > Sports and Gambling
User Name
Password
Register FAQ Search Today's Posts Mark all topics as read

To advertise on these
forums, e-mail us.

Reply
 
Thread Tools Search this Thread Display Modes
  #11  
Old 22nd March 2015, 05:37 AM
Rinconpaul Rinconpaul is offline
Banned
 
Join Date: Feb 2013
Posts: 1,645
Default

On a lighter side, here's a picture of Jarryd Hayne at his new club. Jarryd's the one in the background partly obscured by a bevy of body builder grand finalists! I think he's got some work to do, not only to learn his match plays, but to MUSCLE UP?

Photo courtesy of NineMSN
Attached Thumbnails
Click image for larger version

Name:  Jarryd Hayne.JPG
Views: 112
Size:  48.8 KB  
Reply With Quote
  #12  
Old 24th March 2015, 12:38 PM
walkermac walkermac is offline
Member
 
Join Date: Nov 2013
Posts: 930
Default

Round 3 re-cap. Using the "original" (one could say "mostly defunct") method:

Tips were 6/8:
Canterbury-Bankstown Bulldogs
North Queensland Cowboys
Parramatta Eels

St. George Illawarra Dragons
Melbourne Storm
South Sydney Rabbitohs
Newcastle Knights
Sydney Roosters



Bets were @ $22:
Parramatta Eels WIN @ $3.10
St-George Illawarra Dragons WIN @ $2.25
Cronulla-Sutherland Sharks WIN @ $3.80
Wests Tigers WIN @ $4.65
Penrith Panthers WIN @ $2.95


This Week
Out: $110
In: $49.50
55.5%LOT

Overall
Out: $292.24
In: $333.49
14.1%POT

New Bank: $1041.25 (stakes for next round: $20.80)

H2H S/R: 5/12 (41.7%)
Line S/R: 2/2 (100%)

In my original comment on staking I said that I would watch my imaginary bank dwindle away - that looks to be underway


Using the newer/Monte Carlo approach (one could say "hopefully soon defunct" as I'm intending to alter predicitons dependent on team line-ups), I got 7 from 8 tips right:

Canterbury-Bankstown Bulldogs
North Queensland Cowboys
New Zealand Warriors
St. George Illawarra Dragons
Melbourne Storm
South Sydney Rabbitohs
Newcastle Knights
Sydney Roosters


The value bets I highlighted were a bit of a mixed bag. Odds changed in the lead-up to some clashes: Melbourne -12 and Melbourne by 13+ both became value bets, as did Roosters vs Penrith game total over whatever it was. I believe it came to 7 winning bets vs 9 failing bets (ignoring the error I made in the Dogs/Manly match), and came out more or less square money-wise.

The Cowboys match hurt me most (accounting for 3 failed bets). That team has some real problems...as do Newcastle and St-George Illawarra, despite their wins.

One thing I discovered later during the round was that I'd made a bit of an error with my Monte Carlo simulation. I was pulling my random numbers from an unskewed normal distribution, but once I actually looked at past errors in predicting team scores I saw that it was otherwise (i.e. instead of the bell curve being symmetric, there was a "bulge" to the right, indicating teams were slightly more likely to score above my prediction, than below). Applying the change to the Roosters-Panthers, my estimates were within a handful of cents of what Centrebet was offering.
Reply With Quote
  #13  
Old 25th March 2015, 06:40 PM
Rinconpaul Rinconpaul is offline
Banned
 
Join Date: Feb 2013
Posts: 1,645
Default

This link might be of interest to you Walkermac:

http://www.sportstradingnetwork.com...hing-positions/
Reply With Quote
  #14  
Old 26th March 2015, 11:58 PM
walkermac walkermac is offline
Member
 
Join Date: Nov 2013
Posts: 930
Default

Here's some more NRL mathemagic from http://www.matterofstats.com/mafl-s...afl-and-the-nrl

Win % = 1 / ( 1 + (Pa/Pf)^1.89 )

where Pa = Points Against, Pf = Points For.

Derived from something called Pythagorean Expectation, you can determine the winning percentage for a team over a season only taking into account their Points For-Against. The linked blog attests that the same formula can be used to determine the % chance of victory in a match, given the relationship between the scoring abilities of Team A and Team B. If a team is expected to score 5% more points than their opponent, you can plug it into the formula to determine the probability of their victory (54.7%).

So taking that into account, we can work out what the line should be:

On Centrebet, Cronulla is $1.50, Gold Coast is $2.65. Presuming that the overround is evenly split (which it likely isn't; the home team is probably of greater likelihood of winning than the odds indicate), according to the bookmaker at least, Cronulla has a 63.85% change of winning.

If we plug that into our formula, we eventually get the relationship between the two as:

Points for Cronulla = 1.351356 * Points for Gold Coast

Now what? Well, the bookmaker also tells us there's a 50% chance of the total score being above 37.5. Using our ratio and the total volunteered by the bookmaker, we can solve simultaneous equations and discover that to comprise the 37.5 total, Gold Coast = 15.95 and Cronulla = 21.55. The margin between the two is 5.6. In 50% of the thousands of imaginary games these two sides can play (the 50% with points total greater than that specified), to maintain the points ratio this will be the smallest possible margin. So if you can find Cronulla -5.5 at greater than $2, the bookmaker is telling you you're going to win!

Guess what price Cronulla -5.5 is? $1.91, of course.

The bookmaker's line is the same as determined by the formula above (and the odds are lower than the total offered implies) for all games this round bar Sydney Roosters vs Canberra Raiders. In this case the calculated minimum margin is 17.4. Roosters -16.5 is currently paying $2.05 (Woohoo! Roosters -17.5 is greater still: $2.10).
Reply With Quote
  #15  
Old 27th March 2015, 08:45 AM
Rinconpaul Rinconpaul is offline
Banned
 
Join Date: Feb 2013
Posts: 1,645
Default

Interesting link Walkermac. Very honest assessment, that's what I liked about it. I loved this quote too, "in gambling against the house it's entirely possible to do better than you should - to exhibit something approaching skill and ability - and still walk away with a loss."

LOL
Reply With Quote
  #16  
Old 27th March 2015, 03:41 PM
walkermac walkermac is offline
Member
 
Join Date: Nov 2013
Posts: 930
Default

Speaking of walking away with a loss, here are the "Defunct" method's selections. I'm still working on player strength ratings for the method I have slightly more confidence in; hopefully ready in a few hours/days...

Tips:
South Sydney Rabbitohs
Canterbury-Bankstown Bulldogs
Newcastle Knights
Cronulla-Sutherland Sharks
St. George Illawarra Dragons
New Zealand Warriors
Sydney Roosters
Melbourne Storm

"Value Bets" (not recommended for investment)
Parramatta WIN @ $3.30
Wests Tigers WIN @ $3.40
Gold Coast WIN @ $2.65
St George-Illawarra WIN @ $2.30
Canberra WIN @ $5.50

The method is very much making selections on value rather than likelihood. Canberra, for example, is thought to get beaten by 24 points, but it contends the odds should be around the $3 mark. St George-Illawarra is the only pick to feature as both a Value Bet and a winner, so it's the only one I'd have any belief in (using this as a sub-selection method it's 6 successes from 9, thus far - and 2 of those failures were by only 1-2 points).
Reply With Quote
  #17  
Old 27th March 2015, 03:55 PM
darkydog2002 darkydog2002 is offline
Member
 
Join Date: Nov 1999
Posts: 6,615
Default

Is Jarrod the fat "orange boy " at the back.?
__________________
When you live live in clover . when your dead your dead all over.
Reply With Quote
  #18  
Old 27th March 2015, 04:16 PM
evajb001 evajb001 is offline
Member
 
Join Date: Oct 2009
Posts: 1,249
Default

walkermac, only just spotted this thread. I'm a South Australian so AFL is where my heart lies. Earlier this year I found that matterofstats website as well and saw the ELO style rating system that was applied to the AFL.

I decided to grab some data just based on venue, teams, margin, line and odds/winner. Put together my own really simple ELO type ratings and with some filters/adjustments found a pretty consistently approach to applying these ratings to turn a profit.

Now what i've done could possibly be backfitted or too simplified but just wanted to post here given it seems you've taken on the same kind of project only in a more sophisticated manner. For those interested, as it stands at the moment the following will be my bets for week 1 of the AFL season. Note all my backtesting was completed with closing lines so i'll be waiting to place any bets as close to game time as possible. Bolded bets are my standout bets that i'd place 2 units on whilst the rest i'd place 1 unit on.

Also a quick note, make sure you shop around for NRL/AFL bets. Some bookies offer $2 lines on NRL/AFL all year and many bookies also have offers like lead at half time but lose get your money back. I know these are from some of the corporate's that we don't like but could make a significant difference to your long-term profits.

Brisbane Lions @ the line
Hawthorn @ the line
North Melbourne @ the line
Port Adelaide @ the line
Richmond @ the line
Sydney @ the line (Note: this one is a bit iffy given the Essendon drugs scandal/verdict
West Coast @ the line
Gold Coast @ the line
GWS @ the line

Brisbane to win

I won't post any further in this thread about the AFL approach, was just good to see someone else having a crack at aussie sports ratings. Hope you continue to pursue it walkermac and have some success!
Reply With Quote
  #19  
Old 28th March 2015, 03:04 PM
walkermac walkermac is offline
Member
 
Join Date: Nov 2013
Posts: 930
Default

AFL?! Ew...

I'm from NSW and despite us having two teams now - it is two, right? - I've never followed the sport. I was still going to have a go at tipping using my NRL-developed techniques though; there's too many competitions around to not try. As you say, you only need a small amount of data - much of it not specific to the sport - to come up with a rating system.

I'm finding however that I'm putting more specialised elements into the NRL system - though I'm unsure as to whether they'll be of any great benefit.

I've implemented the "teamlist strength score" I hoped to, but I seem to recall reading on matterofstats someplace that who actually played made very little difference - the team that was playing was just as/more predictive. I suppose that all teams have structures that individual players can be swapped in and out of; all the better to cope with injuries, suspensions and whatnot. Perhaps it's a rare player who has the skill and ability to impose their will upon a game. Still, I'd be interested to discover what sport was under study to come to that conclusion. I can't imagine the NRL's situation is terribly common: given that they remove the star players from regular competition for a number of rounds over the State of Origin period! They also schedule City/Country, Aus vs NZ matches mid-season, giving star players an extra opportunity to get injured in what are likely to be high intensity games.

Another "feature" of NRL: scores aren't uniformly distributed. The score can go up in 4's, 2's or 1's - but the latter is somewhat unusual. It's far more likely to see a score of 20-12, than 21-13 (and you likely wouldn't see many odd-numbered scores when there's a large margin; players tend to only attempt a field goal when the scores are close - final scores aren't independent).

I was running my Monte Carlo Simulation using a Normal (continuous probability) Distribution to model the score prediction error, whereas scoring is a discrete distribution (teams cannot score negative or non-integer values, and some scores are likelier than others). When graphed the error certainly looked like a Normal Distribution so it likely wasn't making a huge difference to H2H selections, but anything regarding actual points scored (line betting, margins, totals) would have been affected negatively.

I've some ideas, but unfortunately I only remember enough mathematics to be dangerous! A better scoring model and a quicker way to test different configurations is on the "to-develop list", this week.
Reply With Quote
  #20  
Old 28th March 2015, 11:45 PM
walkermac walkermac is offline
Member
 
Join Date: Nov 2013
Posts: 930
Default

I might have put the "defect" notice on the frills-free version of ratings a bit early: the value bets are smashing it this round Three out of four, so far - and the one loss occuring only after going to golden point extra time.

Anyway... I've attached a graph of NRL scores attained 1998-2014 (over 6500 individual scores). Clearly, any score predictions would need to take this discrete distribution into account.
Attached Thumbnails
Click image for larger version

Name:  chart.jpg
Views: 93
Size:  54.8 KB  
Reply With Quote
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is On
Forum Jump



All times are GMT +10. The time now is 02:52 PM.


Powered by: vBulletin Version 3.0.3
Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.
2008 OZmium Pty. Ltd. All rights reserved . ACN 091184655