Just a quick observation... I matched up the Beta rankings to the games this week, and there are 5 games that feature two teams in the top half of the league. That means 5 double-down bets, right? In case your curious, they are:
-dal (1) at det (4)
-pit (6) at ne (12)
-stl (15) at cin (7)
-min (11) at sf (8)
-ind (3) at bal (9)
Basically we are saying that these teams all score about the same amount each game, so it should be easy to pick who will score more, and by how much more, right?
An example would be if DAL (1) scores 30 ppg and det (4) scores 20 ppg consistently, we can reasonably expect a score of 30-20, and bet accordingly. Of course in this example, this game appears to be a risky bet, since the spread is dead-on at 10.5.
Showing posts with label G-SPAM. Show all posts
Showing posts with label G-SPAM. Show all posts
Wednesday, December 5, 2007
Thursday, November 15, 2007
Week 11 - Damnable Saints
Excelius-
Let's talk Saints vs Texans...
NO is ranked as a more volatile, less productive (alpha, sharpe, beta) team than HOU. Why do you not only pick the Saints, but suggest doubling down on them?
Let's talk Saints vs Texans...
NO is ranked as a more volatile, less productive (alpha, sharpe, beta) team than HOU. Why do you not only pick the Saints, but suggest doubling down on them?
Friday, November 9, 2007
Week 10 - Check your pants, is it really spam?
Did you run the G-SPAM without my apple sauce modification? Were the results any different?
I think something's off here. Factoring in volatility should increase the success rate. It's only logical. If a team plays consistently at some level, you should know how to bet their game. The application must be flawed.
Basically I'm thinking that:
1. If the game is A vs B, and
2. A plays at consistent level X, while
3. B plays at consistent level X+1,
4. if you bet B, you should expect to win.
Solution #1: Pick games in which both teams have high-ranked Beta Values (low volatility), then pick the winner based on the box's predictions. In this case we have past performance predicting future outcome, with a low possibility for variance from past performance. Overall this should yield good results.
Solution #2: Since Alpha and SHARPE ratios factor in performance and volatility in the form of a neat ratio, you should be able to pick any game and pick the team with the higher ratio as the winner. I suggest using a sample size of 10 games only, going back further than that may include irrelevant information.
*Performance = Performance vs. the average nfl team , so each team's ratios will reflect strength of schedule.
Solution #3: Creation of the Mad Capper Ratio (MCR)! When the box spits out an expected margin of victory vs. the NFL Average (MV), divide it by the Beta Ratio (BR) to come up with essentially an inflation adjusted MV, the MCR! Comparing the two teams' MCRs should give you the winner.
MCR = MV/BR
I think something's off here. Factoring in volatility should increase the success rate. It's only logical. If a team plays consistently at some level, you should know how to bet their game. The application must be flawed.
Basically I'm thinking that:
1. If the game is A vs B, and
2. A plays at consistent level X, while
3. B plays at consistent level X+1,
4. if you bet B, you should expect to win.
Solution #1: Pick games in which both teams have high-ranked Beta Values (low volatility), then pick the winner based on the box's predictions. In this case we have past performance predicting future outcome, with a low possibility for variance from past performance. Overall this should yield good results.
Solution #2: Since Alpha and SHARPE ratios factor in performance and volatility in the form of a neat ratio, you should be able to pick any game and pick the team with the higher ratio as the winner. I suggest using a sample size of 10 games only, going back further than that may include irrelevant information.
*Performance = Performance vs. the average nfl team , so each team's ratios will reflect strength of schedule.
Solution #3: Creation of the Mad Capper Ratio (MCR)! When the box spits out an expected margin of victory vs. the NFL Average (MV), divide it by the Beta Ratio (BR) to come up with essentially an inflation adjusted MV, the MCR! Comparing the two teams' MCRs should give you the winner.
MCR = MV/BR
Week 10 - If There's SPAM In My Pants, Clap Your Hands
So I made the quick switch per Mad Capper's suggestion - use the stats and compare apples to apples. After a week of work, it's official: I suck. Or, at least, SSPAM (Stupid Super Pick Averaging Machine).
Using the stats that correct for the average, the long term results were SIGNIFICANTLY worse.
The ole Box is looking better and better... we'll see how we do this week.
Using the stats that correct for the average, the long term results were SIGNIFICANTLY worse.
The ole Box is looking better and better... we'll see how we do this week.
Week 10 - Making G-SPAM apple sauce
Crystal Clear, Excelius!
I, myself, prefer kama to karma - like kama sutra gigity gigity.
Back to football- What do you think about altering the G-Spam to calculate volatility in relation to how a team performs against the average - exactly what the box does with stats only - instead of calculating the volatility of points scored only?
It seems that with this method, you would be taking strength of schedule into account more directly (since both teams' performance would be defined not in a bubble, but vs. the average team). Basically what I'm saying is that it doesn't matter so much how many points scores from week to week, what matters is margin of victory in combination with the strength of the opponent, exactly what the box considers.
Then, instead of averaging the box and the G-SPAM (two different systems), it's all rolled into one. Sorta like making apple sauce out of two apples instead of apples and peanuts (ew, peanuts).
Keep Capping, baby.
I, myself, prefer kama to karma - like kama sutra gigity gigity.
Back to football- What do you think about altering the G-Spam to calculate volatility in relation to how a team performs against the average - exactly what the box does with stats only - instead of calculating the volatility of points scored only?
It seems that with this method, you would be taking strength of schedule into account more directly (since both teams' performance would be defined not in a bubble, but vs. the average team). Basically what I'm saying is that it doesn't matter so much how many points scores from week to week, what matters is margin of victory in combination with the strength of the opponent, exactly what the box considers.
Then, instead of averaging the box and the G-SPAM (two different systems), it's all rolled into one. Sorta like making apple sauce out of two apples instead of apples and peanuts (ew, peanuts).
Keep Capping, baby.
Week 10 - You Dare Question My Salted Meat?
You ask good questions, grasshopper, let me see if I can explain.
First, all of the financials (Sharpe, beta, and alpha) are based on pure points. They do NOT take into account the statistics that the original (and still sexy) Box uses, they only look at week to week output of points regardless of the team they are playing. So if DAL scores a rolling average of 25 points per game (NE is somewhere around a ridiculous 39 points per game on the season), and last week they scored 25 points, they would show a lower volatility, especially if they did that consistently. If DAL (beta of 1.39) plays NYG (beta of 1.12), the beta module will pick the team with lower volatility (basically, whichever team is closest to 1.0), which in this case is the NYG.
Sharpe and alpha are different ways to measure essentially the same thing: how good is the team relative to their volatility/risk? Despite NYG having a slightly better volatility, DAL has been PAYING OUT HUGE on it's moderately higher risk. DAL has an alpha of 8.78 to NYG 2.85, and DAL has a Sharpe of 0.99 to NYG 0.53. That makes sense when you think about it - despite NYG playing more consistently, they have played more consistently at a lower level than DAL. One week, DAL may score 50 points, and the next 20, but they are still outputting higher scores than NYG.
Now, what I may do to make the modules run a little more evenly is to find out which teams consistently cover rather than just score. I tried this earlier this week, and it got sticky. It may be worth a revisit. The financials right now are great ways to modify your thinking, ESPECIALLY on moneyline bets, since it's gives you a benchmark against which you can measure the schizophrenic teams. This chart may better illustrate that, noting that interesting schizo teams include ATL (beta rank 8th, Sharpe and alpha rank 30th), DEN (beta rank 5th, Sharpe and alpha rank 29th and 28th), and KC (beta rank 7th, Sharpe and alpha rank 28th and 29th). Basically it's showing you that a.) these teams perform consistently week to week, and b.) they perform consistently BAD week to week.
What I've done with these rankings is AVERAGE the scores in with the original Box (they are making sweet love). It's a weighted average, so the original Box is weighted 2/3, the financials are 1/3 (historically it seems to pick up a few games, but I have to finish the back test). This explains why the pick is still NYG, because they are actually statistically better and lower volatility, versus a high risk/high return play of DAL which is weighted less. What the averaging has accomplished is switched the following picks from the original Box:
STL +11.5 switched to NO -11.5
ATL +4 switched to CAR -4
Both of these games were on the cusp statistically, but the weighting pushed them over the edge.
Are you humbled by my intellect? I thought so. (For all this boasting, I figure I should lose 100% of my bets this week - karma's a hideous bitch goddess).
First, all of the financials (Sharpe, beta, and alpha) are based on pure points. They do NOT take into account the statistics that the original (and still sexy) Box uses, they only look at week to week output of points regardless of the team they are playing. So if DAL scores a rolling average of 25 points per game (NE is somewhere around a ridiculous 39 points per game on the season), and last week they scored 25 points, they would show a lower volatility, especially if they did that consistently. If DAL (beta of 1.39) plays NYG (beta of 1.12), the beta module will pick the team with lower volatility (basically, whichever team is closest to 1.0), which in this case is the NYG.
Sharpe and alpha are different ways to measure essentially the same thing: how good is the team relative to their volatility/risk? Despite NYG having a slightly better volatility, DAL has been PAYING OUT HUGE on it's moderately higher risk. DAL has an alpha of 8.78 to NYG 2.85, and DAL has a Sharpe of 0.99 to NYG 0.53. That makes sense when you think about it - despite NYG playing more consistently, they have played more consistently at a lower level than DAL. One week, DAL may score 50 points, and the next 20, but they are still outputting higher scores than NYG.
Now, what I may do to make the modules run a little more evenly is to find out which teams consistently cover rather than just score. I tried this earlier this week, and it got sticky. It may be worth a revisit. The financials right now are great ways to modify your thinking, ESPECIALLY on moneyline bets, since it's gives you a benchmark against which you can measure the schizophrenic teams. This chart may better illustrate that, noting that interesting schizo teams include ATL (beta rank 8th, Sharpe and alpha rank 30th), DEN (beta rank 5th, Sharpe and alpha rank 29th and 28th), and KC (beta rank 7th, Sharpe and alpha rank 28th and 29th). Basically it's showing you that a.) these teams perform consistently week to week, and b.) they perform consistently BAD week to week.
| Sharpe | Alpha | Beta | |
| Team | Rank | Rank | Rank |
| ARI | 18 | 24 | 6 |
| ATL | 30 | 30 | 8 |
| BAL | 26 | 27 | 20 |
| BUF | 27 | 26 | 14 |
| CAR | 24 | 25 | 3 |
| CHI | 23 | 23 | 13 |
| CIN | 9 | 7 | 19 |
| CLE | 4 | 3 | 31 |
| DAL | 2 | 2 | 10 |
| DEN | 29 | 28 | 5 |
| DET | 7 | 8 | 18 |
| GB | 8 | 9 | 25 |
| HOU | 11 | 6 | 26 |
| IND | 3 | 5 | 23 |
| JAC | 19 | 20 | 16 |
| KC | 28 | 29 | 7 |
| MIA | 15 | 11 | 32 |
| MIN | 14 | 18 | 15 |
| NE | 1 | 1 | 2 |
| NO | 12 | 14 | 9 |
| NYG | 6 | 10 | 11 |
| NYJ | 25 | 16 | 30 |
| OAK | 20 | 22 | 22 |
| PHI | 17 | 17 | 29 |
| PIT | 5 | 4 | 21 |
| SD | 10 | 15 | 12 |
| SEA | 13 | 13 | 24 |
| SF | 31 | 31 | 4 |
| STL | 32 | 32 | 27 |
| TB | 22 | 19 | 17 |
| TEN | 16 | 12 | 28 |
| WAS | 21 | 21 | 1 |
What I've done with these rankings is AVERAGE the scores in with the original Box (they are making sweet love). It's a weighted average, so the original Box is weighted 2/3, the financials are 1/3 (historically it seems to pick up a few games, but I have to finish the back test). This explains why the pick is still NYG, because they are actually statistically better and lower volatility, versus a high risk/high return play of DAL which is weighted less. What the averaging has accomplished is switched the following picks from the original Box:
STL +11.5 switched to NO -11.5
ATL +4 switched to CAR -4
Both of these games were on the cusp statistically, but the weighting pushed them over the edge.
Are you humbled by my intellect? I thought so. (For all this boasting, I figure I should lose 100% of my bets this week - karma's a hideous bitch goddess).
Week 10 - Excelus Incredulus? G-SPAM response.
Response to the G-SPAM (baby)...
I love the idea of accounting for volatility - this would put some statistical analysis behind my "multiple personality" teams. However, some of the results make me question whether or not the parameters you've set are giving you what you think you have.... ok, confusing sentence, but basically i'm saying, are you sure the G-SPAM is giving you a better indicator of future performance than the original box?
The original Box picks a team to cover the spread based on past performance against the average team. The G-SPAM averages ratios that factors in volatility, hopefully reducing the risk of multiple personality teams.
However, why would the G-SPAM pick NYG +1.5, when both the alpha and SHARPE ratios pick Dallas? Since the Alpha and SHARPE ratios take into account peformance against the average team AND volatility, it would seem that if they agree on Dal, then Dal would be the pick when averaged. This is not the case. Any theories?
I love the idea of accounting for volatility - this would put some statistical analysis behind my "multiple personality" teams. However, some of the results make me question whether or not the parameters you've set are giving you what you think you have.... ok, confusing sentence, but basically i'm saying, are you sure the G-SPAM is giving you a better indicator of future performance than the original box?
The original Box picks a team to cover the spread based on past performance against the average team. The G-SPAM averages ratios that factors in volatility, hopefully reducing the risk of multiple personality teams.
However, why would the G-SPAM pick NYG +1.5, when both the alpha and SHARPE ratios pick Dallas? Since the Alpha and SHARPE ratios take into account peformance against the average team AND volatility, it would seem that if they agree on Dal, then Dal would be the pick when averaged. This is not the case. Any theories?
Subscribe to:
Posts (Atom)