My WNBA efficiency ratings

[frogsrangers]

Just wrote a program to calculate WNBA efficiency/tempo ratings like how kenpom does for college basketball. It's not quite as advanced as kenpom's yet but it should be good enough for the ballpark at the moment.

Glossary:
Offensive Efficiency = Points per 100 posessions
Defensive Efficiency = Points allowed per 100 possessions
Tempo = Possessions per game (average)
Pyth = Expected winning percentage

Team	Offensive Efficiency	Defensive Efficiency	Tempo	Pyth
Minnesota	105.73	91.21	81.05	.8197
San Antonio	101.59	94.74	81.07	.6716
Connecticut	99.74	94.21	83.21	.6421
Indiana	98.94	93.61	80.21	.6382
Atlanta	92.48	89.13	84.05	.5934
Los Angeles	98.41	96.14	83.92	.5595
Chicago	93.98	92.31	80.46	.5431
Seattle	91.14	94.00	76.88	.4215
Tulsa	92.56	100.98	83.96	.2906
Washington	90.33	100.24	77.58	.2559
Phoenix	91.55	101.75	84.83	.2530
New York	88.85	99.60	80.31	.2367

[frogsrangers]

Tonight's play:

Chicago Sky -5.5 vs. New York

Capping it as a 10 point Sky win

[EVPlus]

BOL with this play.

btw, I noticed in another thread that your model shows Conn will cover the -9. How confident are you about this?

I think the actual spread should be closer to -7.5

[bleek88]

and those are the exact way the standings should be.

[frogsrangers]

BOL with this play.

btw, I noticed in another thread that your model shows Conn will cover the -9. How confident are you about this?

I think the actual spread should be closer to -7.5

You are correct about the spread for Connecticut should be closer to -7.5, I actually have it at -5.5

But I am confident enough that I parlayed it with the Silver Spurs -4

My model shows all 3 favorites tonight winning by double digits and therefore covering

Now my model has both projected final score and projected Vegas point spreads.

Here are the projected scores:

Connecticut 93 Tulsa 77
Chicago 72 New York 63 (Chicago 79 New York 63 if Ephiphany Prince were playing)
San Antonio 85 Washington 67

But here were the point spreads my model created, vs. the actual spreads:

Chicago -9.7 (actual -5.5, +4.2 value points)
San Antonio -6.6 (actual -4, +2.2 value points)
Connecticut -5.5 (actual -8.5 or 9, +3.0 to 3.5 value points)

This is why I went with Chicago, you are getting the most value points according to my model

I just made the model today, so I wouldn't put too much stock into it yet, we will have to see how it actually fares with the results.

[EVPlus]

Great information and BOL tonight. I went small on the over for San/Was game.

The reason I do not want to lay so much for Conn tonight is situational. I think Tulsa has to be pretty embarrassed at having the worst record in the league. Also, they have tomorrow off whereas Conn has a game. Perhaps they will redline it all the way to the final buzzer whereas Conn may ease off the throttle...? Therefore, I see the potential for a backdoor cover now the lines are closer to 9.5 to 10.

[silvap]

Best of luck with the model sir, will follow you results. Cheers and thanks for sharing your insight

[MonkeyF0cker]

You're going to get destroyed.

[frogsrangers]

Chicago Sky was a dumb play, my point spread model doesn't take into account missing players, so it should have been a no play.

Nailed the other 2 games though.

Day 1 in the books lets see how it does in the long run. Today would have been a winning day.

[MonkeyF0cker]

Chicago Sky was a dumb play, my point spread model doesn't take into account missing players, so it should have been a no play.

Nailed the other 2 games though.

Day 1 in the books lets see how it does in the long run. Today would have been a winning day.

But here were the point spreads my model created, vs. the actual spreads:

Chicago -9.7 (actual -5.5, +4.2 value points)
San Antonio -6.6 (actual -4, +2.2 value points)
Connecticut -5.5 (actual -8.5 or 9, +3.0 to 3.5 value points)

[MonkeyF0cker]

I know this is really difficult to grasp, but if your line states that a team should be -5.5, how do you bet "-8.5 or -9?"

P.S. - You're going to get killed. Good luck.

[dynamite140]

Interesting numbers

[frogsrangers]

I know this is really difficult to grasp, but if your line states that a team should be -5.5, how do you bet "-8.5 or -9?"

P.S. - You're going to get killed. Good luck.

San Antonio and Connecticut were winners, Chicago Sky was a loser, but had I factored in Sky was missing its best player I would have taken it off

If a team's actual line is -5.5 and my model says the line should be -9.7, that means that -5.5 is a bargain. Likewise if an actual line is +10 but my model says the team should only be +5, thats a bargain too.

A model like this can't pick winners or losers. It is just another set of data, another tool to help make a pick. It won't make you a winner, it all depends on how you interpret the data and what you do with it.

[EVPlus]

Chicago Sky was a dumb play, my point spread model doesn't take into account missing players, so it should have been a no play.

Nailed the other 2 games though.

Day 1 in the books lets see how it does in the long run. Today would have been a winning day.

Congrats on your 2-1 night. I'm checking out for the night but before I do, I have one comment you can take or leave.

I've noticed many mathematical models do not factor in situational analysis and the human element. After all, these games are played by humans, officiated by humans, and cheered or booed by humans. Without an understanding of human nature, I believe most mathematical models are incomplete.

The good news is that it's your model and you can make adjustments as you see fit. Look forward to a possible discussion of the weekend's action.

[MonkeyF0cker]

San Antonio and Connecticut were winners, Chicago Sky was a loser, but had I factored in Sky was missing its best player I would have taken it off

If a team's actual line is -5.5 and my model says the line should be -9.7, that means that -5.5 is a bargain. Likewise if an actual line is +10 but my model says the team should only be +5, thats a bargain too.

A model like this can't pick winners or losers. It is just another set of data, another tool to help make a pick. It won't make you a winner, it all depends on how you interpret the data and what you do with it.

LOL. Uhh. Connecticut WAS -8.5, -9 ACTUALLY. YOUR LINE says they should be -5.5.

You seem a little confused.

[rfr3sh]

could be a good starting point
did u look into any injuries to see why your lines differed or just blindly make selections

[frogsrangers]

LOL. Uhh. Connecticut WAS -8.5, -9 ACTUALLY. YOUR LINE says they should be -5.5.

You seem a little confused.

Crap, I was reading my output table wrong. Yeah messed up on that one. Good thing I got lucky with it.

[frogsrangers]

Congrats on your 2-1 night. I'm checking out for the night but before I do, I have one comment you can take or leave.

I've noticed many mathematical models do not factor in situational analysis and the human element. After all, these games are played by humans, officiated by humans, and cheered or booed by humans. Without an understanding of human nature, I believe most mathematical models are incomplete.

The good news is that it's your model and you can make adjustments as you see fit. Look forward to a possible discussion of the weekend's action.

Human element is something you can't really legislate into any mathematical model. It's why models should at best be only a tool in aiding you make selections, not the sole factor in deciding selections. Think of it as a single criteria

For example tonight, even though my model said Chicago -9.7, that is factoring in Ephiphany Prince and her numbers into the equation. But since she didn't play, you have to look at that part of the equation as well. The smart play would have been to pass on Chicago because she is worth about 12 points to her team, wiping out the -9.7

A mathematical model should be one of the many factors influencing a pick. If you make it your sole factor you will lose. I think the best thing a model provides is looking for value. But as anyone can tell you value does not equal a winner.

Since profitability is all about finding an edge, if a model can help you gain that 1% or 2% edge, its worth it in the long run, even if the short term results look bad.

[Roxxyfish]

I went 4-0 again without any model,just intuition and knowledge

[frogsrangers]

I went 4-0 again without any model,just intuition and knowledge

I want to learn from the best

[frogsrangers]

Today I am parlaying the Indiana and Minnesota Money Lines for a -220 payout, and taking Indiana -9.5 as well as Atlanta -6.

My model is giving me money lines identical to the actual posted lines, so not much ML value today. Also its giving me spreads near identical to the spreads posted at the books, except for it gave me Indiana -14. So 4.5 value points on Indiana. My model also predicted a 16 point Atlanta win over Phoenix, which is why I am going with Atlanta -6.

My model also gave me totals identical to the posted totals. I must be using algorithims similar to what the oddsmakers use.