[Jay Edgar]

On the Best Bet threads there is more than one of my typically long-winded explanations of what these estimated pointspreads are, and on why I don't rely on them exclusively to pick games.

I have also prattled on at length about how the NBA lines at this late stage of the season seem very tight.

Nevertheless, it's worth noting that when the Jay Edgar estimated line enjoys a gap of more than 2 points from the final pointspread ( I have drawn the line at 2.4 points), the indicated team is still covering at close to 60%.

Here is the report card for the first ten days of March. (Note that I have made one alteration -- a common sense switch to a "no play" on PHX vs SA in light of the Nash injury.)

For some reason, it is usually NOT the case that a larger gap is more predicative of a winner. That is true again here as well.

DATE PLAY AGAINST GAP RESULT MARGIN

03-10 DEN +4 TOR 3.4 WIN 15
03-10 ORL +3- CLE 4.9 WIN 32.5
03-10 MIA -9 GS 3.1 LOSS 14

03-09 PHI +1 DEN 5.3 LOSS 3
03-09 POR +8- DAL 8.4 LOSS 8.5
03-09 PHX +4 SAN 8.6 no play 28

03-08 CHA -1- GS 5.6 WIN 1.5
03-08 WAS +9- MIA 2.9 WIN 4
03-08 BOS +2 PHI 7.6 WIN 9
03-08 IND +4 HOU 3.9 PUSH 0
03-08 MIN +4 UTH 2.8 WIN 1

03-07 BOS +7 WAS 3.2 WIN 8
03-07 TOR +6 CLE 3.9 LOSS 1
03-07 ATL +1- GS 9.2 WIN 9
03-07 POR +15 DAL 6.2 WIN 9
03-07 CHI -3 NJ 3.0 WIN 3

03-06 CHA +5- MIA 4.9 WIN 4.5
03-06 SAC +4 NJ 5.9 WIN 19
03-06 MEM +4- DEN 5.4 LOSS 9.5
03-06 ORL +4- UTH 5.3 LOSS 0.5
03-06 LAX +4 SAN 4.0 LOSS 2.5

03-05 IND +2 PHI 2.5 WIN 3
03-05 CHI +4- CLE 2.7 LOSS 14.5
03-05 HOU -11- POR 4.1 WIN 6.5

03-04 TOR +5 NJ 3.7 PUSH 0
03-04 ATL +12- MIA 3.9 WIN 10.5
03-04 ORL +10-DEN 8.0 LOSS 6
03-04 LAX +4.5 DET 3.5 WIN 13.5

03-03 ATL +4 SAC 3.1 LOSS 5
03-03 NY +2.5 CHI 4.0 LOSS 4.5
03-03 BOS +1 IND 4.0 WIN 2
03-03 CHA +11-DAL 6.2 LOSS 2.5
03-03 SEA +7- DET 5.2 WIN 5.5

03-01 SAC +3- CLE 5.5 WIN 10.5
03-01 BOS +5- MIA 9.3 LOSS 1.5
03-01 PHI +5- HOU 2.8 WIN 10.5
03-01 MIL +9- PHX 3.6 LOSS 3.5
03-01 CHA +5 UTH 2.4 WIN 2
03-01 NOK +8 LAC 5.1 LOSS 14

21 wins
15 losses
2 pushes
1 no-play

All of the estimated pointspreads were posted in the Best Bets threads as soon as they were available on the night before the game.

In the calculation I have used the final line as posted on Wagerline, although I am sure there are some variations. And of course the prudent lineshopper could have turned a couple of the losers and pushes listed above into winners.

[mad]

Nice work Jay Edgar. Are their any vaariations that produce better/different results at different times? Such as home/away, back2back etc or any other situations that might highlight weakness and/or strengths with your method. Bearing in mind a weakness could be used to play against.

[Jay Edgar]

Thanks mad, I never have tried to do that kind of breakdown in part because I haven't used these lines to dictate picks 100% -- rather I tend to take the numbers and then think about fatigue, motivation, matchups, shakeups or changes in routine for one team, etc.

One minor criticism I still have of my own ratings is that I think they may tend to reward good efforts on the road a little bit too much and good efforts at home not quite enough. I was going to go back into the formula and check about that.

The end result is that a team like SAC who just had a great road trip with no bad efforts is going to rate very highly, while a team that has been at home the whole time taking care of business against weaker opponents will not be able to keep up with SAC unless they are dispatching people by 20-25 or more -- which may not be a priority for them.

So that's one thing I try to watch out for (although SAC back at home last night did just fine, and was not a play because MEM has been so strong.)

[bigboydan]

this is very interesting indeed, thanks for sharing jay.

[imgv94]

Thanks JayEdgar for all your work. I think your cool!

[Illusion]

Excellent work Jay. It's much appreciated.

[fabric86]

Could you explain how the system works again? How do I calculate it?

[Jay Edgar]

Could you explain how the system works again? How do I calculate it?

I really can 't -- not because it's the formula for Coke, but because it's complicated. I wouldn't be able to show somebody how to get my exact ratings unless we sat at a table for a couple hours. And anyone would be better off doing their own because you learn so much while doing them.

But for mine the most basic concept is that each team gets a score from 0-100 for each game (six factors go into making up that score) with 50 as the dead middle average. Then once we're ten games into the season each team gets a power rating, which is their average game score over the last 10 games weighted for recency. 70-75 is a great power rating, and 20-25 is awful, and most are in between. Then for any upcoming game you can get a pointspread based on how far the rankings are from the league average and from each other Then one team gets their individualized home court ad, and I have an approximate number that I put into the thread. I'm also ranking the teams each night by power rating and putting the league rankings into the thread, so that you can get some sense of how good my numbers say the team is relative to the league. (League rankings also matter because a team gets a much better game score if, for example, it loses a close one to the #3 team than if it loses a close one to the #23 team.)

I really love the exercise of talking a look at every team every day and ranking them, even if the methodology is admittedly not perfect. I know that I have mindsets about certain teams that the numbers are constantly challenging -- the Knicks right now are a great example.