Haneyh's Review of Conquering Risk

[statictheory]

There are patterns?

I thought you were responding to me initially because you didn't see the sharing of valuable information as a problem at all.

I don't mean to be argumentative, but it appears to me that you're actually agreeing with my point of view.

im not agreeing that a book is giving away too much information. im saying the info in any book in regards to sportsbetting, poker, horseracing, trading, is limited at best.
im not agreeing with you regarding your speed figures because no horse that i know of read beyers book. so the only thing that would change is the prices one would get from speed horses right? but if calculating speed figures correctly was a true advantage and such a strong predictor of out come then you could make money regardless of the price. favorites have won around 30-33 % of the time forever. you are trying to say i think that speed was overlooked until beyer wrote a book. its hard for handicappers who use it because it has limited utility before or after beyer. its not the overiding factor of why a horse wins, .
and as in sportsbetting if your handicapper friends know that it isnt working and havent made an adjustment well, thats the definition of insanity isnt it?
I think that there are people that make money handicapping but they are looking at other factors besides speed, but there are very few that win that dont have an angle that they have found on their own. im kinda done with this , so good luck too you. if you are a horseplayer i would just say too you to start watching the board, look at the lines and start looking st them close once you have a thousand or so races. you might start seeing some things. good luck ps the favorites at 30-33% has been that way before and after beyer. you would expect an increase in favorite% coming in as a result , if beyers speed figures etc were truly having a large effect.

[roasthawg]

Ordered the book... gonna read through it and see if it gives anything important away. Winning is so tough that anything short of a blueprint to success will probably lead to more money for the books and sharps rather than vice versa.

[RockyV]

Rocky, I honestly don't think you'll get anything out of Yao or Wong's books. Better to wade through threads on good forums. Honestly, I'd find posters you think are smart, click on their names, and read through their posts to find good threads. It can be time consuming because there's a lot of noise, but some nuggets. Good luck.

Thanks. I'm doing that, but it helps me to sort of fit the picture better in my head the more I read. And yeah, I've waded through a lot of threads here...lots of good stuff posted by Ganchrow and others.

I read through the "Well" stuff on the twoplustwo.com sports betting forums too...that also seemed pretty useful.

[statictheory]

Ordered the book... gonna read through it and see if it gives anything important away. Winning is so tough that anything short of a blueprint to success will probably lead to more money for the books and sharps rather than vice versa.

+1

[CatPulp]

Hooray!!! My book arrived today!!

[rise]

good stuff will make me better in the long run

[George7904]

It is certainly an interesting read; but nothing that a mid-range to advanced player doesn't already know. I don't see how it gives away too much.

I would like to read an entire book based on his "grey page" stories. Those stories were the most interesting part of the book.

[roasthawg]

I agree with George... I didn't read word for word but I've skimmed through most of the stuff that interests me and thought it was a good book but not dangerous or anything. I enjoyed it... solid read.

[Lib208]

Sounds like a good read, i think its in the sbr store

[2Bdown]

i will be picking this up in the SBR store, $24.95 on amazon seems a bit steep for a paperback

[dataz722]

Question for those of you that have read this book, how much prior statistics knowledge do you need? I have taken some fairly advanced math classes previously and used to be very strong with math. I am sure that once I start getting back into it that it will come back pretty quickly. Only thing is that I have never taken any statistics before and really know pretty much nothing about it. I am sure I could learn it pretty quickly as long as I started with the basics. Does this book go over the basics or does it use the assumption that you already have a base knowledge of statistics.

[Justin7]

Question for those of you that have read this book, how much prior statistics knowledge do you need? I have taken some fairly advanced math classes previously and used to be very strong with math. I am sure that once I start getting back into it that it will come back pretty quickly. Only thing is that I have never taken any statistics before and really know pretty much nothing about it. I am sure I could learn it pretty quickly as long as I started with the basics. Does this book go over the basics or does it use the assumption that you already have a base knowledge of statistics.

I try to walk the reader through most of the statistics you need. You do need to understand algebra at a minimum to get everything out of the book.

[dataz722]

Sounds great. Algebra is no problem for me at all. I am a little shaky anymore with calc but anything less then that am I am fine.

Looks like I am going to be saving up points to get your book.

[statictheory]

I have a question. At dr bobs site he talks about the models most people make are wrong because they use regression and says it doesnt work because such things as turnovers cant really be forecasted since he views them as random events. How do you feel about this and do your models in your book use regression as the main tool?

[Justin7]

I have a question. At dr bobs site he talks about the models most people make are wrong because they use regression and says it doesnt work because such things as turnovers cant really be forecasted since he views them as random events. How do you feel about this and do your models in your book use regression as the main tool?

I agree that turnovers are difficult to predict going forward. A regression would tell you that.

Dr. Bob states on his webpage: Less sophisticated simulators that try to come up with a formula to predict future games tend to make the same mistake; they use regression analysis to find the correlation between different statistics and point differential. While that exercise is very useful for explaining which statistics impact a game's result, regression is not necessarily useful in using past statistical averages to predict future results since some important statistics simply don't correlate very highly to the future.

There are right and wrong ways to use regression. If you assume the observed patterns will carry forward exactly as they have in the past, your model will fail. If you study the data to determine how predictive past results are, you can use regression as a tool.


Regression is a chief part of development in 3 out of 4 models (the MLB model doesn't really use regression).

[statictheory]

I agree that turnovers are difficult to predict going forward. A regression would tell you that.

Dr. Bob states on his webpage: Less sophisticated simulators that try to come up with a formula to predict future games tend to make the same mistake; they use regression analysis to find the correlation between different statistics and point differential. While that exercise is very useful for explaining which statistics impact a game's result, regression is not necessarily useful in using past statistical averages to predict future results since some important statistics simply don't correlate very highly to the future.

There are right and wrong ways to use regression. If you assume the observed patterns will carry forward exactly as they have in the past, your model will fail. If you study the data to determine how predictive past results are, you can use regression as a tool.


Regression is a chief part of development in 3 out of 4 models (the MLB model doesn't really use regression).

makes sense, thanks

[cg432]

Just placed my order on Amazon. Cant wait for it to get here, as I am a novice for sure

[LegitBet]

can you name just a couple of other variables with similiar challenges that the turnover problem presents.
thanks

[Flight]

Any event that you would consider random. You can look at a teams turnover differential, but how much is random and how much of that is based on a team's skill? Just be careful about assigning all of a turnover's value to a team's ability (and repeatability) rather than randomness.

Consider the question: Does Adrian Peterson really fumble the ball more than league average (as certain sports outlets would have you believe)?

[Flight]

Another team attribute I consider to be random is penalties. (some may argue against me)

[Justin7]

can you name just a couple of other variables with similiar challenges that the turnover problem presents.
thanks

Turnovers are the biggest one. 3rd down conversions have a luck component also, but there are more factors going into the conversion rate than luck.

[pedro803]

injuries seems like a particularly unwieldy variable as well -- although I guess it doesn't play in the same way that your are discussing -- anybody got anything to say about handling injuries?

[Poogs]

I just finished up with this book. I thought it was excellent and would certainly recommend it to anyone who takes sportsbetting seriously at all. My only complaint was that I thought the prop section was a little too short. It has some great info, but I just wish there was more. Overall I think this has been the best book on sportsbetting Ive read; its like Weighing the Odds but much more up to date. Defeintly get this book.

[gilly6864]

i have read justin 7 book and if you have a lot of time to produce models and apply math then this book is for you by far. if you don't have time to invest hours into producing models that will work than the basic EV method is pretty workable. people with extensive math background can profit in sports betting no question. however, people who handicap by looking at basic stats, injuries, line movement do well to. making money gambling is difficult b/c the majority of people just bet with no reason to why a team might win and to boot they have no money management. money management is key, being patient not to bet the whole bankroll. you can make money betting sports, by doing basic homework on the games. i do believe you can over handicap a game by spending hours applying math concepts. find the balance when handicapping!

[TPowell]

should I purchase the Sharp Sports Betting book from Wong and Weighing the Odds from King Yao?? I also recently bought this book from the SBR store, waiting on it to get here

[Arkwright]

Sorry to bring up an old topic but I'm buying this book. However, how am I supposed to apply these models. For example, how am I supposed to implement the NFL points based model in the book if I don't have in depth knowledge of Excel. This is where I think I'm going to struggle. Should I be doing some sort of Excel course or something?

[hubie69]

[mathdotcom]

I agree that turnovers are difficult to predict going forward. A regression would tell you that.

Dr. Bob states on his webpage: Less sophisticated simulators that try to come up with a formula to predict future games tend to make the same mistake; they use regression analysis to find the correlation between different statistics and point differential. While that exercise is very useful for explaining which statistics impact a game's result, regression is not necessarily useful in using past statistical averages to predict future results since some important statistics simply don't correlate very highly to the future.

There are right and wrong ways to use regression. If you assume the observed patterns will carry forward exactly as they have in the past, your model will fail. If you study the data to determine how predictive past results are, you can use regression as a tool.

Just because a bunch of numbskulls don't use regression properly doesn't mean it can't do the above. The more stuff I read of Bob's the less and less I respect him (if at all). In fact the first bold highlighted part is a complete contradiction of itself.

If you run a regression of Wrigley Total Score on Wrigley Wind Direction and find it's significant, Bob would say "Good job, you've discovered that wind direction at Wrigley is a factor affecting the game's total!", and then next sentence says "But that probably won't help you handicap Wrigley totals next season." Of course it will. I believe he makes this statement because he usually sees 'modelers' coming up with completely arbitrary regressions with no theory behind them that typically amount to dubious correlations.

Now that I think about it, his statement is actually backwards. There are plenty of variables that are meaningless for "explaining which statistics impact a game's result", yet including them in a regression can still lead to significant coefficients AND effective predictions. Suppose you want to estimate baseball moneylines so you begin with starting pitchers. You run a regression of Runs Given Up on ERA of Team's Last Starting Pitcher. Clearly the ERA of the pitcher who started last night has no effect on the runs given up today. But what it will likely pick up is some sort of positive relationship, since the starting pitcher today usually has a higher ERA than the pitcher yesterday (except when you move from #5 to #1 in the order). This positive relationship is garbage, it's just correlated with the variable that does matter: ERA of starting pitcher today. Obviously this regression is going to yield pretty shitty predictions, but I'd rather defend the statement that "ERA of starting pitcher yesterday is predictive of the runs given up by the team today" than "ERA of starting pitcher yesterday explains how many runs will be given up today."

And the second bolded statement is completely wrong since that is exactly what you're doing, or at least trying to do if you're J7

[durito]

Sorry to bring up an old topic but I'm buying this book. However, how am I supposed to apply these models. For example, how am I supposed to implement the NFL points based model in the book if I don't have in depth knowledge of Excel. This is where I think I'm going to struggle. Should I be doing some sort of Excel course or something?

You can learn the excel skills necessary to replicate the models in this book in a couple of hours (assuming you have no previous excel experience). Only do it as an exercise though, if you expect to make money do not bet these "models".

[bruceBRUCEbruce]

any reason why Justin's book is no longer available via the SBR store?

[TomG]

J7 would you be willing to share some financial aspects of your book? What was your cut from the publisher? do you feel your publisher provided enough value to justify that amount? how many copies did they expect to sell? how many did you actually sell? how many copies did you sell in 2010 vs. 2011 vs. 2012 ytd?

[Justin7]

TomG,

A friend of mine owns a publishing company. This made things much easier, since the main publishers and distributors had no interest. I sold more than I expected, and it has been reprinted once. I made some money from sales, but I made much more from the information that flowed to me as a result of the book.

Ironically, the people I work with (that found me due to the book) have made it clear that they don't want me publishing any more books on sports betting.

[sharpcat]

TomG,

A friend of mine owns a publishing company. This made things much easier, since the main publishers and distributors had no interest. I sold more than I expected, and it has been reprinted once. I made some money from sales, but I made much more from the information that flowed to me as a result of the book.

Ironically, the people I work with (that found me due to the book) have made it clear that they don't want me publishing any more books on sports betting.

[raiders72001]

And it is always perfectly visible who is more or less drunk than others. Unless you have evidence that this guy was completely falling over when ordering his drinks, the bar is completely off the hook.

The bar isn't off the hook. That's why you buy dram shop insurance.

[big0mar]

Just because a bunch of numbskulls don't use regression properly doesn't mean it can't do the above. The more stuff I read of Bob's the less and less I respect him (if at all). In fact the first bold highlighted part is a complete contradiction of itself.

If you run a regression of Wrigley Total Score on Wrigley Wind Direction and find it's significant, Bob would say "Good job, you've discovered that wind direction at Wrigley is a factor affecting the game's total!", and then next sentence says "But that probably won't help you handicap Wrigley totals next season." Of course it will. I believe he makes this statement because he usually sees 'modelers' coming up with completely arbitrary regressions with no theory behind them that typically amount to dubious correlations.

I"m not sure how relevant this example is. One wouldn't use past wind direction to predict future wind direction.

Now that I think about it, his statement is actually backwards. There are plenty of variables that are meaningless for "explaining which statistics impact a game's result", yet including them in a regression can still lead to significant coefficients AND effective predictions. Suppose you want to estimate baseball moneylines so you begin with starting pitchers. You run a regression of Runs Given Up on ERA of Team's Last Starting Pitcher. Clearly the ERA of the pitcher who started last night has no effect on the runs given up today. But what it will likely pick up is some sort of positive relationship, since the starting pitcher today usually has a higher ERA than the pitcher yesterday (except when you move from #5 to #1 in the order). This positive relationship is garbage, it's just correlated with the variable that does matter: ERA of starting pitcher today. Obviously this regression is going to yield pretty shitty predictions, but I'd rather defend the statement that "ERA of starting pitcher yesterday is predictive of the runs given up by the team today" than "ERA of starting pitcher yesterday explains how many runs will be given up today."

I don't think that was the suggestion. I think the point was that if a statistic cannot reliably be predicted, it's impact on outcomes is irrelevant in this context. You use a very good example of mediating statistics. In football, penalty rate has only a very low correlation with outcomes. But it is very reliable in terms of predicting future penalty rate, which makes it useful.