Ask John Ryan E-Mail Bag: Week 3

[John Ryan]

Welcome to the third installment of John Ryan's Mailbag. We have had some great questions asked over the first two week, all of which generated a lot of conversation.

This week I will continue to share my insight in this forum, but SBR has now upped the ante for us! While I will continue to select 3-5 of the best questions to feature in my weekly article, the poster will now receive a 50 point sportsbook free play for their involvement.

I will answer the selected questions in detail in an article that will be posted on www.sbrforum.com each Friday.
So now that the stakes have been raised, fire away.

[cant call it]

Become a Pro, John Ryan.

[Puppy]

why do i chase after every losing streak?

[JJJ]

This thread is stupid

[John Ryan]

why do i chase after every losing streak?

That is the human nature factor. You are not alone. Everyone hates to lose. No one has ever walked on the face of the earth lovig to lose.. Look at how Giselle reacted emotionally. The ket word is emotion in that sentence. So many people negatively effect their lives by acting by emotions that do not include the benefit of intellect. This is the bulls-eye where my passion for the markets and sports wagering is centered. Herd mentality runs rampant is so many parts of our lives. In is a enormous balance between fear on one end adn greed on the other..

Although your question is short it carries a ton of weight... You may just get that 50 point free bet from SBR. Thanks

[Swinging Johnson]

John,

I have an old 8X10 glossy of you posing with your neural algorithm software next to a Commdore 64. Do you have any recent posters for sale?

[Puppy]

That is the human nature factor. You are not alone. Everyone hates to lose. No one has ever walked on the face of the earth lovig to lose.. Look at how Giselle reacted emotionally. The ket word is emotion in that sentence. So many people negatively effect their lives by acting by emotions that do not include the benefit of intellect. This is the bulls-eye where my passion for the markets and sports wagering is centered. Herd mentality runs rampant is so many parts of our lives. In is a enormous balance between fear on one end adn greed on the other..

Although your question is short it carries a ton of weight... You may just get that 50 point free bet from SBR. Thanks

if i had a 1k bankroll and went on a losing streak down to $500 id rather have $0 at that point then just $500. because emotionally i would feel the same like i lost and it doesnt matter its all or nothing at that point

id rather just reload then play with half a bankroll

[2boca]

John, what is your best sport to bet on, purely from a profits perspective and what is it about that sport that makes it so?

[shaunovery]

Hi what's the easiest way to cap nba/ncca basketball totals

Is it trends possessions teams shooting defense etc

Shaun

[BHawksforLife]

How would somebody go about creating his own sports betting system?

I would like to create some sort of a simulator, but I don't know where to begin or any software that could help me.

What software do you use to keep track of all of the factors involved (i.e. Points scored, points allowed, minutes played, fg%, rest, etc.)?

Do you have any great reads that you would recommend for sports handicapping?

Thank you for having this thread. I plan on learning a lot.

[John Ryan]

John, what is your best sport to bet on, purely from a profits perspective and what is it about that sport that makes it so?

well, I could try to be funny and say the sport that is currently being played.. but, it is nearly always MLB and NBA and is attributed to the long seasons. NHGL can also be huge, but here my plays will hit roughly 50%, but because more than 75% of the plays are dogs, I am able to make a tidy sum... sort of analogous to playing Black Jack at the casino and instead of getting paid 1:1 for a winning hand, I am getting paid 1.65:1 on average..

My longest winning streak ever has been in the NBA and was 18 straight games back in 2002.

Thanks, Great question

[SBR Lou]

John,

Would you be willing to battle Justin7 in a one-on-one handicapping contest, held on SBR Contests for all to see?

[John Ryan]

How would somebody go about creating his own sports betting system?

I would like to create some sort of a simulator, but I don't know where to begin or any software that could help me.

What software do you use to keep track of all of the factors involved (i.e. Points scored, points allowed, minutes played, fg%, rest, etc.)?

Do you have any great reads that you would recommend for sports handicapping?

Thank you for having this thread. I plan on learning a lot.

I utilize professional programmers, which are not cheap, but are extremely good and understand the algorithms. Try browsing through this site. http://www.csail.mit.edu/

see what turns up..http://www.mathworks.com/help/toolbo...f/trainlm.html

[Landscaper]

John,

Would you be willing to battle Justin7 in a one-on-one handicapping contest, held on SBR Contests for all to see?

[John Ryan]

John,

Would you be willing to battle Justin7 in a one-on-one handicapping contest, held on SBR Contests for all to see?

sure.. why not... all in good fun..

[Landscaper]

John,

Would you be willing to battle Justin7 in a one-on-one handicapping contest, held on SBR Contests for all to see?

U should put up a prize for the winner though.. Cash is king

[John Ryan]

Hi what's the easiest way to cap nba/ncca basketball totals

Is it trends possessions teams shooting defense etc

Shaun

wish it was that easy... as you know basketball games are filled with runs by both teams... Those runs can 7-0 runs or they can be even a greater 14-0 run once in a while. But, they do occur.. Totals depend a lot on how a team reacts to these runs.. The team that just went wild can fall asleep at the wheel and allow the other team to come back.. or if the team that gave up the run will try too hard to get back into the game and make matters worse.

I think if you check out my Orlando play released today and read the research you can see why I like Orlando to win. I also think the 'over' will win too based on the research, but is NOT a play. You can see how the specific strengths of the Orlando defense matchup very well against the strengths of the Heat offense. So, it is not necessarily an 'under' play because of the advantages as the Magic can used missed perimeter shots to generate their own fast break scoring opportunities. Must consider both sides of the ball and how offense matchups affect defensive ones and for both teams. The sim does all that work for me.

[John Ryan]

Now, here is a quick example of the complexity of a simulator based on layers of variables,
Conjugate gradient backpropagation with Powell-Beale restarts
Syntax
[net,TR] = traincgb(net,TR,trainV,valV,testV)
info = traincgb('info')

Description
traincgb is a network training function that updates weight and bias values according to the conjugate gradient backpropagation with Powell-Beale restarts.

[net,TR] = traincgb(net,TR,trainV,valV,testV) takes these inputs,

net
Neural network
TR
Initial training record created by train
trainV
Training data created by train
valV
Validation data created by train
testV
Test data created by train
and returns

net
Trained network
TR
Training record of various values over each epoch:
Each argument trainV, valV, and testV is a structure of these fields:

X
N-by-TS cell array of inputs for N inputs and TS time steps. X{i,ts** is an Ri-by-Q matrix for the ith input and TS time step.
Xi
N-by-Nid cell array of input delay states for N inputs and Nid delays. Xi{i,j** is an Ri-by-Q matrix for the ith input and jth state.
Pd
N-by-S-by-Nid cell array of delayed input states.
T
No-by-TS cell array of targets for No outputs and TS time steps. T{i,ts** is an Si-by-Q matrix for the ith output and TS time step.
Tl
Nl-by-TS cell array of targets for Nl layers and TS time steps. Tl{i,ts** is an Si-by-Q matrix for the ith layer and TS time step.
Ai
Nl-by-TS cell array of layer delays states for Nl layers, TS time steps. Ai{i,j** is an Si-by-Q matrix of delayed outputs for layer i, delay j.
Training occurs according to traincgb's training parameters, shown here with their default values:

net.trainParam.epochs 100
Maximum number of epochs to train
net.trainParam.show 25
Epochs between displays (NaN for no displays)
net.trainParam.showCommandLine 0
Generate command-line output
net.trainParam.showWindow 1
Show training GUI
net.trainParam.goal 0
Performance goal
net.trainParam.time inf
Maximum time to train in seconds
net.trainParam.min_grad 1e-6
Minimum performance gradient
net.trainParam.max_fail 5
Maximum validation failures
net.trainParam.searchFcn 'srchcha'
Name of line search routine to use
Parameters related to line search methods (not all used for all methods):

net.trainParam.scal_tol 20
Divide into delta to determine tolerance for linear search.
net.trainParam.alpha 0.001
Scale factor that determines sufficient reduction in perf
net.trainParam.beta 0.1
Scale factor that determines sufficiently large step size
net.trainParam.delta 0.01
Initial step size in interval location step
net.trainParam.gama 0.1
Parameter to avoid small reductions in performance, usually set to 0.1 (see srch_cha)
net.trainParam.low_lim 0.1
Lower limit on change in step size
net.trainParam.up_lim 0.5
Upper limit on change in step size
net.trainParam.maxstep 100
Maximum step length
net.trainParam.minstep 1.0e-6
Minimum step length
net.trainParam.bmax 26
Maximum step size
info = traincgb('info') returns useful information about this function.

Network Use
You can create a standard network that uses traincgb with newff, newcf, or newelm.

To prepare a custom network to be trained with traincgb,

Set net.trainFcn to 'traincgb'. This sets net.trainParam to traincgb's default parameters.

Set net.trainParam properties to desired values.

In either case, calling train with the resulting network trains the network with traincgb.

Examples
Here is a problem consisting of inputs p and targets t to be solved with a network.

p = [0 1 2 3 4 5];
t = [0 0 0 1 1 1];
A feed-forward network is created with a hidden layer of 2 neurons.

net = newff(p,t,2,****,'traincgb');
a = sim(net,p)
Here the network is trained and tested.

net = train(net,p,t);
a = sim(net,p)
Algorithms
traincgb can train any network as long as its weight, net input, and transfer functions have derivative functions.

Backpropagation is used to calculate derivatives of performance perf with respect to the weight and bias variables X. Each variable is adjusted according to the following:

X = X + a*dX;
where dX is the search direction. The parameter a is selected to minimize the performance along the search direction. The line search function searchFcn is used to locate the minimum point. The first search direction is the negative of the gradient of performance. In succeeding iterations the search direction is computed from the new gradient and the previous search direction according to the formula

dX = -gX + dX_old*Z;
where gX is the gradient. The parameter Z can be computed in several different ways. The Powell-Beale variation of conjugate gradient is distinguished by two features. First, the algorithm uses a test to determine when to reset the search direction to the negative of the gradient. Second, the search direction is computed from the negative gradient, the previous search direction, and the last search direction before the previous reset. See Powell, Mathematical Programming, Vol. 12, 1977, pp. 241 to 254, for a more detailed discussion of the algorithm.

Training stops when any of these conditions occurs:

The maximum number of epochs (repetitions) is reached.

The maximum amount of time is exceeded.

Performance is minimized to the goal.

The performance gradient falls below min_grad.

Validation performance has increased more than max_fail times since the last time it decreased (when using validation).

Whewww!!!!! but when it essentially calculates 6 figures worth of data per second, it does not take all that long to get through a voluminous database of 'numbers'

[Swinging Johnson]

Now, here is a quick example of the complexity of a simulator based on layers of variables,
Conjugate gradient backpropagation with Powell-Beale restarts
Syntax
[net,TR] = traincgb(net,TR,trainV,valV,testV)
info = traincgb('info')

Description
traincgb is a network training function that updates weight and bias values according to the conjugate gradient backpropagation with Powell-Beale restarts.

[net,TR] = traincgb(net,TR,trainV,valV,testV) takes these inputs,

net
Neural network
TR
Initial training record created by train
trainV
Training data created by train
valV
Validation data created by train
testV
Test data created by train
and returns

net
Trained network
TR
Training record of various values over each epoch:
Each argument trainV, valV, and testV is a structure of these fields:

X
N-by-TS cell array of inputs for N inputs and TS time steps. X{i,ts** is an Ri-by-Q matrix for the ith input and TS time step.
Xi
N-by-Nid cell array of input delay states for N inputs and Nid delays. Xi{i,j** is an Ri-by-Q matrix for the ith input and jth state.
Pd
N-by-S-by-Nid cell array of delayed input states.
T
No-by-TS cell array of targets for No outputs and TS time steps. T{i,ts** is an Si-by-Q matrix for the ith output and TS time step.
Tl
Nl-by-TS cell array of targets for Nl layers and TS time steps. Tl{i,ts** is an Si-by-Q matrix for the ith layer and TS time step.
Ai
Nl-by-TS cell array of layer delays states for Nl layers, TS time steps. Ai{i,j** is an Si-by-Q matrix of delayed outputs for layer i, delay j.
Training occurs according to traincgb's training parameters, shown here with their default values:

net.trainParam.epochs 100
Maximum number of epochs to train
net.trainParam.show 25
Epochs between displays (NaN for no displays)
net.trainParam.showCommandLine 0
Generate command-line output
net.trainParam.showWindow 1
Show training GUI
net.trainParam.goal 0
Performance goal
net.trainParam.time inf
Maximum time to train in seconds
net.trainParam.min_grad 1e-6
Minimum performance gradient
net.trainParam.max_fail 5
Maximum validation failures
net.trainParam.searchFcn 'srchcha'
Name of line search routine to use
Parameters related to line search methods (not all used for all methods):

net.trainParam.scal_tol 20
Divide into delta to determine tolerance for linear search.
net.trainParam.alpha 0.001
Scale factor that determines sufficient reduction in perf
net.trainParam.beta 0.1
Scale factor that determines sufficiently large step size
net.trainParam.delta 0.01
Initial step size in interval location step
net.trainParam.gama 0.1
Parameter to avoid small reductions in performance, usually set to 0.1 (see srch_cha)
net.trainParam.low_lim 0.1
Lower limit on change in step size
net.trainParam.up_lim 0.5
Upper limit on change in step size
net.trainParam.maxstep 100
Maximum step length
net.trainParam.minstep 1.0e-6
Minimum step length
net.trainParam.bmax 26
Maximum step size
info = traincgb('info') returns useful information about this function.

Network Use
You can create a standard network that uses traincgb with newff, newcf, or newelm.

To prepare a custom network to be trained with traincgb,

Set net.trainFcn to 'traincgb'. This sets net.trainParam to traincgb's default parameters.

Set net.trainParam properties to desired values.

In either case, calling train with the resulting network trains the network with traincgb.

Examples
Here is a problem consisting of inputs p and targets t to be solved with a network.

p = [0 1 2 3 4 5];
t = [0 0 0 1 1 1];
A feed-forward network is created with a hidden layer of 2 neurons.

net = newff(p,t,2,****,'traincgb');
a = sim(net,p)
Here the network is trained and tested.

net = train(net,p,t);
a = sim(net,p)
Algorithms
traincgb can train any network as long as its weight, net input, and transfer functions have derivative functions.

Backpropagation is used to calculate derivatives of performance perf with respect to the weight and bias variables X. Each variable is adjusted according to the following:

X = X + a*dX;
where dX is the search direction. The parameter a is selected to minimize the performance along the search direction. The line search function searchFcn is used to locate the minimum point. The first search direction is the negative of the gradient of performance. In succeeding iterations the search direction is computed from the new gradient and the previous search direction according to the formula

dX = -gX + dX_old*Z;
where gX is the gradient. The parameter Z can be computed in several different ways. The Powell-Beale variation of conjugate gradient is distinguished by two features. First, the algorithm uses a test to determine when to reset the search direction to the negative of the gradient. Second, the search direction is computed from the negative gradient, the previous search direction, and the last search direction before the previous reset. See Powell, Mathematical Programming, Vol. 12, 1977, pp. 241 to 254, for a more detailed discussion of the algorithm.

Training stops when any of these conditions occurs:

The maximum number of epochs (repetitions) is reached.

The maximum amount of time is exceeded.

Performance is minimized to the goal.

The performance gradient falls below min_grad.

Validation performance has increased more than max_fail times since the last time it decreased (when using validation).

Whewww!!!!! but when it essentially calculates 6 figures worth of data per second, it does not take all that long to get through a voluminous database of 'numbers'

I am so turned on right now.

[Inkwell77]

John, with sample sizes being so small in the NFL are "basic" systems completely pointless?
If you just look at NFL games that have had lines moves across a 3 or 7 (starts and 2.5 or 3 and moves to 3.5 or 4 etc) over the course of the last 3 years you would be 87-67 ats (from my research) with this year being the true banger with a record of 30-16 ats. This is based on getting the bad number and not getting any early number before a line moves. Am I crazy to think that next year this trend stays below .500 ats?

[John Ryan]

I am so turned on right now.

well if that is the case, I now know why you have been wearing that bag over your head all of this time. Ha Ha Ha Ha

[Swinging Johnson]

well if that is the case, I now know why you have been wearing that bag over your head all of this time. Ha Ha Ha Ha

[Roxxyfish]

can i wager on Justin ?

John,

Would you be willing to battle Justin7 in a one-on-one handicapping contest, held on SBR Contests for all to see?

[ChuckyTheGoat]

Little did we know that John Ryan's simulator was like liquid viagra.

[GenosPicks]

John Ryan... Geno Bisconte of Geno's Picks here... first time long time...

Anyway now that The NFL is over and gambling lines are a more day to day thing i think we all become a little eager to make last minute NBA, NHL, or NCAA bets just to have a little action. Any thoughts on which of the three is the smartest league to gamble on or are they all pretty much similar and there is a different strategy to take.

Oh and nice line about Swingin's bag.

[Swinging Johnson]

John Ryan... Geno Bisconte of Geno's Picks here... first time long time...

Anyway now that The NFL is over and gambling lines are a more day to day thing i think we all become a little eager to make last minute NBA, NHL, or NCAA bets just to have a little action. Any thoughts on which of the three is the smartest league to gamble on or are they all pretty much similar and there is a different strategy to take.

Oh and nice line about Swingin's bag.

I just wish, for only a day, you and all of your kind (whom we call "faces") would have to endure the pain, bigotry and intolerance of being a baghead.

While our countenance may repulse you, our emotions are just as palpable and our yearning to be included is a daily struggle for which the battle is simply educating one "face" at a time.

So Geno, enjoy your ability to flash a big toothy grin and grow a beard on command. I envy you but wouldn't trade places. I sit here, my bag drenched in the tears of imperfection but blanketed by the outpouring of love from bagheads all across this great nation of ours.

[Swinging Johnson]

John, seriously good stuff last night with the Lakers and Kings, both lockline winners. Give this guy some love people!

[John Ryan]

John, seriously good stuff last night with the Lakers and Kings, both lockline winners. Give this guy some love people!

Thanks.

[John Ryan]

I just wish, for only a day, you and all of your kind (whom we call "faces") would have to endure the pain, bigotry and intolerance of being a baghead.

While our countenance may repulse you, our emotions are just as palpable and our yearning to be included is a daily struggle for which the battle is simply educating one "face" at a time.

So Geno, enjoy your ability to flash a big toothy grin and grow a beard on command. I envy you but wouldn't trade places. I sit here, my bag drenched in the tears of imperfection but blanketed by the outpouring of love from bagheads all across this great nation of ours.

I wonder if Therapy would be helpful.. any Dr. Phils in the audience?

You should rip the bag from your head and all of the pain, suffering, adn guilt, you have had to carry and walk proudly through the remainder of your life!!!! even if it is in the middle of the Resorts or Hilton casinos main ball rooms.

[BHawksforLife]

John, seriously good stuff last night with the Lakers and Kings, both lockline winners. Give this guy some love people!

Thank you for all of your hard work John. The detailed writeups on the site are excellent.

[John Ryan]

Thursday's article is here.

http://www.sbrforum.com/sports-betting/free-picks/john-ryan-s-mailbag-a-23298/

[GenosPicks]

I just wish, for only a day, you and all of your kind (whom we call "faces") would have to endure the pain, bigotry and intolerance of being a baghead.

While our countenance may repulse you, our emotions are just as palpable and our yearning to be included is a daily struggle for which the battle is simply educating one "face" at a time.

So Geno, enjoy your ability to flash a big toothy grin and grow a beard on command. I envy you but wouldn't trade places. I sit here, my bag drenched in the tears of imperfection but blanketed by the outpouring of love from bagheads all across this great nation of ours.

hahaha

For all the jealousy you have that i can grow a beard on command know that the hair that grows on my back is twice a thick half as appealing.. i may need a bag next year for the beach myself.

[John Ryan]

any further questions about the article this week? Thanks Geno... you don't need a bag.. just a razor... or that stuff called 'Nair"

you're tough enough to burn that hair off..

[Puppy]

John Ryan can i get that 50 pts i posted a unique question?

[John Ryan]

John Ryan can i get that 50 pts i posted a unique question?

you were not selected unfortunately.. I will ask them if they can add you.. but if not re-write the question again and submit for Week 4 that starts Monday..