ALL-Ways Confusion

[PandaMan]

I've been reading the ALL-Ways newsletters, many which attempt to explain how to use the data provided by the software.

The problem is they spend a lot of time explaining but somehow don't manage to tell you exactly how to use the data. I'm not a moron, but I'm having a helluva time figuring out just how to use the info I'm buying and downloading from BRIS.

I looked at ALL-Ways some years ago, and this was a problem then. There was actually a book published that told you step-by-step how to handicap using ALL-Ways by a third party. The author claimed many people had the same issue, i.e. difficulty learning the software despite "detailed" instructions. ALL-Ways quickly took action against the book, wanting its clients to obtain information only from them. I'd love to get a copy of that book.

So am I alone in this? Is there any way to learn how to use ALL-Ways besides the frustrating newsletters they make available? It's obvious there are more than a few using it with success, so I feel like I'm missing something.

[gregrph]

I have tried to use this software a number of times. I love reading the newsletters and I believe that there is good information in there. My problem is that I have little tiime to sit down and digest it all. I have problems trying to figure out WHERE on the reports (and which reports to use!) to get the contenders. Also not too comfortable picking profiles to use. I think that if I tried to print out the reports for even one race, the output would be tremendous! The think is I would really like to learn how to use this program properly, but the information is too little and too far between (quarterly newsletters.) Also, the cost of data for 1 race is prohibitive if you are trying to learn! Thus, I only use it a few times a year. Triple Crown races. I did not use it for the Breeders Cup this year as I have never had any luck with it. I'm sure the problem is between my ears and not the program!


A dedicated forum for discussion of the program would be a BIG help! Users could help each other.

Right now, I will read with interest the newsletters, but probably not use it.
I still have the data files if there is a better explanation of how to start, which reports to use, WHERE TO FIND THE INFORMATION on the reports. MANY numbers with cryptic abbreviations! I did print out the manual one time but that's 100+ pages.

Greg

[Space Monkey]

I'll ditto all the comments from the first 2 posts. I tried it a couple of years ago and found it to be "information overload". The manual was humungous, the cost high, and the explanations confusing. I called a couple of times and they were helpful. But it became apparent to me that I had to sit down with somebody with knowledge of the program, who would walk it thru with me. I'm confident that just one day would have done it for me. I asked if they could hook me up with somebody in the Tampa area and they said flat out no. They don't have any groups or share members names as they are concerned about file sharing. OK, fine. I gave it up.

[PandaMan]

The frustrating thing is all this info and instruction is supposedly right there, but written in such a way as to make it extremely difficult if not impossible to understand. I know I'm not the only one with this complaint about AW.

They use some bizarre type of instructional slight of hand where it seems as though they're explaining things but really are not. That makes me crazy.

[Space Monkey]

I'm waiting for someone with a handle on it to tell us what we're missing. Threads still young.

Off to Ray Jay. Go BULLS

[garyoz]

IMHO a program with serious methodological flaws--and I (in conjunction with others) spent a very long time and many dollars trying to make it work. Too many highly correlated variables in poorly designed (and statistically flawed) "regression" models. You are just backfitting way too many variables that measure the same thing and determining beta weights as if they are acting independently in a bivariate regression--then combining them in a multiple regression equation. That is the major reason the oddsline is a joke.

BRIS and Frandsen were great at marketing the product and charging for data. I bought the first version of ALW when it came out--I think 1996--just as Gary Hall was signing a deal with BRIS. I've watched the iterations and paid for several upgrades over the years.

My advice would be to move to Horsestreet or HTR if you want to use datamining/databases as a foundation for your handicapping.

Personally I have given up on applying multivariate statistics to handicapping. There has been much posted on this board about this topic.

[Space Monkey]

You are just backfitting way too many variables that measure the same thing and determining beta weights as if they are acting independently in a bivariate regression--then combining them in a multiple regression equation

Gary, I respect your obvious high mathematical intelligence, but can you express yourself so an average joe can understand what your talking about?

[garyoz]

Once again, this is just my opinion-- First, there are too many variables. You really don't need 80+ variables. Many of the variables measure the same thing (for example back speed and back class are very highly correlated). The way I believe the modeling works is that each variable is measured individually against win/lose (a statistical violation unless you use something called logistic regression or probit analysis). So, in our example, back class will get a value in "predicting" a winner, then back speed wil get a value (these are called beta weights in statistics). Then the program adds them together, not accounting for the fact that they measure pretty much the same thing (this is called correlation). In multiple regression, you should enter them in a stepwise fashion and the second variable probably would not be significant and would not be included in the model . Otherwise, in effect, you double count.

There are other things that I was dissatisfied with--like way too much noise in all the models, contradictory findings, too many contenders, too many favorites, etc. I could go on. Twenty years ago, I thought that multiple regression could be an effective tool in handicapping, plus I like building models and doing the analysis. Thus, I was really psyched when ALW came out, and tried to make it work.

My criticism is based upon having spent thousands of dollars on the program and upgrades and data files during the late 90's and earlier this decade--you get bought-in after awhile, hard to cut your losses of cash and time. Also, don't get sucked in by backfitting. Remember, you can always build a model that fits historic data--the key is getting it to forecast the future and beat the takeout in this game.

Hope that helps-- Be interested to hear if anyone has a different view.

[CBedo]

I used All-Ways for some time about five or so years ago, and would echo most of what Gary said. To compensate for some of the backfitting, I woudl split data into multipe data (usually by date) and backfit to one and test others. I did have some success doing it this way, but you definitely have to be careful. Also, the success I had was finding contenders and maybe some overlooked horses, not using the odds line which never correlated to reality. Also, the angle assistant (I think that's what it was called) was fun to play with, but again, was just taking too many variables and factors and backfitting. I found that the best thing the software was used for was to use the regressions of the angles to possibly give me ideas for further research.

[Space Monkey]

Thanks guys.

In the few times I tried the program, I agree, the odds line was mostly out to lunch. I guess "information overload" not only pertained to the user, but to the data that went into the program too.

[zerosky]

I'm toying with the idea of logistic regression, using the export function to get at the 87 handicapping factors.
XLstat has a months free trial which is ideal for purpose.

Preliminary results for santa anita has whittled the 87 down to about 5 or 6 depending on the type of race.
The next step is to do a two step model incorporating tote odds,
unfortunatly the export does not include the odds therefore they will have to be entered manually....yuk!

The data files even with bulk discount is expensive so this is likely to be a one off experiment for me.

[HUSKER55]

I wish I could help but I can't. However, Dave Schwartz has a special on his programs. The program is easy to get up and going. I am an idiot but I was up and going in a week and Dave answers his own phone.

Give it some thought before you spend more time and money

BTW, there must be at least 20 free videos you can watch when you need assistance and his forums have someone who always responds.

Good luck

[garyoz]

Based on my experience, I'd go with Horsestreet or exported data from HTR if I were going to do statistical analysis and/or datamining. IMHO HDW has much better and cleaner data than ALW and it is much cheaper.

[PandaMan]

All I'm after are clear-cut instructions on how to use the data. That's been my big gripe with ALL-Ways. For the life of me I cannot understand their explanation of how to use the software. A big pile of numbers I don't know how to use won't do me any good.

I looked at HTR, and it does seem to be a serious, high-quality product. I briefly took a look at their forum as well as newsletters. I just don't want to get involved in something else that gives me a pile of data and then sends me hunting through hundreds of articles to learn how to use it effectively. All I want are concise directions to get me started handicapping with it. I can learn the more advanced aspects as I go.

[garyoz]

Originally Posted by PandaMan

All I'm after are clear-cut instructions on how to use the data. That's been my big gripe with ALL-Ways. For the life of me I cannot understand their explanation of how to use the software. A big pile of numbers I don't know how to use won't do me any good.

Yeah, agree, that's the problem... I don't think anyone can tell you how to use the output effectively. Maybe the best advice you are getting is to pass and move on.