Live Help

You can now share systems with other Bet Labs members. By clicking on the Think Tank button from the Bet Labs Home page, you will be able to see all shared systems.

You can view an introduction video here: http://www.youtube.com/watch?v=3gTMohpHVDU

If you have any questions, please let us know.

asked 03 Mar, 13:11

SI-Travis's gravatar image

SI-Travis
468811
accept rate: 0%


« previous123next »

Hopefully my previous comment will give people an idea on where to start, and which filters they can use to create winning systems.

Someone named Troutman just posted 20 or so systems, and they have no predictive value whatsoever. This is exactly what I was talking about, how not to build a system. Cherry-picking individual pythagorean numbers, or individual ML prices, bet percentages, etc.. has no predictive value. None. Now, if you can create a range of numbers, that is something entirely different. O/U range, win percentage range, etc... These have a better chance of being predictive.

I have posted 6 systems. Lets take a look at a simple one, MLB. A 51% or better home dog vs a 60% or better opponent. Baseball is a great sport to build underdog systems, even the worst teams beat the best teams, often at inflated odds. So if I can find a good team that wins often, at a price - even good teams are dogs at times, that is my goal. So I filtered opp win % and found a threshold that works. 60%. Last year 2013 the record was 17-10 (ROI 33%) for a +892. Almost 9 units of profit. I'll take that every time. Now the important part...is that an anomaly or does it win every year or close to it??

I like to back test my systems for 5-6 years. 10 years isn't necessary, as systems can run their course. Over the last 6 years, it has won 5 years. The last 6 years profits are: 892, 1236, -277, 1186, 1121, 1201. The 3 years prior (7-9 yrs ago) to that were losses of 2-5 units. I'm fine with that, as the past 6 years have been phenomenal. Each winning year has hit at over 50% wins, if you can find a dog system with these win percentages you have a very good system.

There you have it. Not overly complicated, it makes sense, its rational, and it wins. Year after year. Thats what you are looking for when building a system.

link

answered 07 Mar, 00:27

smartplays1127's gravatar image

smartplays1127
1463310
accept rate: 0%

Let's look at another MLB system. A good team (55% or better), a dog, Lost last game by at least 3 (> 3), vs decent team (40-50% win).

Again, I'm looking for teams in the habit of winning, and I want a price. So I thought, what would make them a dog?? Well, how about coming of a loss. So I filtered Previous Game Margin and found that when they lost by 3 or more it was profitable. I wanted to improve on the ROI, so I then filtered the opponents win percentage. Makes sense, as better teams will offer a better price. Poorer teams will obviously offer lesser odds or our team may be favored despite losing by 3 last game.

The public will jump off a team that lost by a decent margin. Losing a baseball game by 3 or more doesn't inspire confidence by the general public. By using these 2 filters: Previous game margin and Opp win percentage, I was able to find a very nice situation that yields profits. Now, I could have filtered it even further: Road Dog, but home dogs are profitable too so I didn't include that filter. Again, does it win consistently or is it just a 1 year wonder??

Lets look at the past 6 years: Last 3 years a profit of 9.5, 17.5, and 3 units. The next 3 years lost 1,2,and 4 units. The next 3 years won 3.5,5.5,and 1 unit. Here is a situation where it isn't as "clean", but the last 3 years have won, the last 2 years it did very well. So I will again use this system and monitor. The bad years have only lost 4 units at most, which to me is worth the risk of picking up 10 or more units this year.

link

answered 07 Mar, 00:44

smartplays1127's gravatar image

smartplays1127
1463310
accept rate: 0%

Here's an NHL system: Winning Team (> 50%), Off a Shutout Loss, Road Dog.

I have no idea why games are high scoring or low scoring. You can look at all the stats you want: power play %, goals scored, allowed, penalties, etc... I don't want to turn handicapping into quantum physics, and that is what makes Bet Labs such a wonderful product.

I started with the premise of finding situations where the game went Under. I then thought, if a team just got shut out then they obviously are having trouble scoring. I know, no shit. So I then thought about the psychology of what that does to a team. So I figured, what if you are a good team, and just got shut out? You are used to winning, and getting shut out leaves a bad taste in your mouth. What's the best way to get back in the win column?? Not letting the other teams score.

I filtered the following: home fav, Visitor Off Points streak of 1 game no goals, visitor's win percentage (50% - 100%). A good team thats a dog is obviously facing a really good team. I figured coming off a shut out loss, the game would be a grind it out affair. The results:

Past 6 years 10-20 plays a year. Not a lot, but it wins every year but 1. It was 10-11 and lost 1 unit. The other years it won 3.77, 1.6. 7.7, 4.6, 2.3.

Now that may not be the most exciting system, but it wins. Look at these systems as if you were playing poker.

Some times a few players limp, you raise and you take down a small pot. Sometimes you get a couple callers, see a flop and take it down with a bet. Sometimes it goes to the turn, and it ends there and you take down a little bigger pot. You go to the river and take down another. And sometimes you go all in, get called, and take down a huge pot. The point is, every pot - big or small - contributes to your chip stack.

Every profitable system, whether they pick up 2 units or 5 units or 10 units or 15+ units contribute to your bankroll. You don't have to crush every system, you don't need every system to win 20 units or hit 70% or have a ROI of 50%. ROI IS important, as the better the ROI the fewer plays you need to gain a profit. Nobody wants to make 80 plays to pick up 2 units. If I can make 10-20 plays and pick up a couple units or a bit more, I'll take it.

link

answered 07 Mar, 01:05

smartplays1127's gravatar image

smartplays1127
1463310
accept rate: 0%

edited 07 Mar, 01:09

A couple notes:

1) I would agree that there is a lot of bad system pollution accumulating in the Think Tank. On one hand, I don't think there should be any kind of censorship; on the other hand, it would be nice to prevent someone from posting 20 small-sample systems in a day. Perhaps institute a +/- voting system, and then people can sort by net votes if they desire?

2) @smartplays11: One thing to note about your first MLB system is that its avg. margin of victory is -0.3 runs/game, which does not bode well for a 57% win pct. I.e, a 57% win pct. system should show about a +0.6 avg. run margin. That, combined with the small sample, suggests the system has won a lot of lucky games to achieve its record.

link

answered 07 Mar, 01:22

Mark%20P's gravatar image

Mark P
917713
accept rate: 0%

Small sample size??

216 plays in 6 years, avg is 36 plays a year.

I don't care what the average run margin is. It doesn't matter. If a loss is by a blowout, that skews the run differential. Again, this isn't quantum physics.

You failed to mention that the 57% win percentage is for an underdog. That is a HUGE edge. If you don't think a 57% win percentage for a system that plays strictly underdogs is good, then I have no idea what you are looking for in a system.

A lot of lucky games?? Not every game was a 1 run game, which if you are saying many were lucky, is a valid assumption. Nothing could be further from the truth. There were many many games with run differentials much bigger than 1 run. Netting the difference has absolutely nothing to do with the quality of the system. None.

Run margin has no bearing on the strength of this system.

Lucky? Thats an abnormal amount of luck to have these results.

link

answered 07 Mar, 01:28

smartplays1127's gravatar image

smartplays1127
1463310
accept rate: 0%

edited 07 Mar, 01:37

"Run margin has no bearing on the strength of this system. None." OK. Wonder why all of the most predictive power rating systems use margin of victory rather than raw won-loss numbers... Anyway, I'm not here to argue with anyone. Just found it funny that someone aggressively complaining about weak systems is championing 216-game backtested system with a MOV that is entirely out of whack with the W/L record. As they say, it takes two to make a market. Glad to have you on board.

link

answered 07 Mar, 01:38

Mark%20P's gravatar image

Mark P
917713
accept rate: 0%

You're comparing the systems I've posted to the ones I've mentioned as being weak?? Seriously??

I know all about power ratings, they have nothing to do with finding a profitable system. Apples and Oranges.

All systems are back tested, how on earth can you determine if they are profitable if you don't??

Again, MOV has nothing to do with it. We aren't creating Power ratings, which use MOV. We are finding profitable situations by W/L. You don't get bonus money for margin of victory, so why would it matter?? If I win a game 5-2, win another 6-1, and lose another 9-0, whats the collective MOV?? I win 2 games by a cumulative score of 8 runs, lose the other by 9. That's a negative MOV. Yet, you think that makes a difference?? It doesn't.

Also, you don't need 100 plays a year to be a valid system. You need consistent winning results and logic. Call it arguing, debating, I don't mind having a discussion. I'm here to help, not kiss people's asses. I don't sugar coat things, I call it like I see it.

We can agree to disagree, but I think you are completely misguided applying MOV to the system, to me its irrelevant.

Curious as to what some of the SO people think regarding MOV??

link

answered 07 Mar, 01:49

smartplays1127's gravatar image

smartplays1127
1463310
accept rate: 0%

I'm quoting your entire comment below, because you've significantly edited each of your previous comments. I would have to guess that SportsInsights thinks it has relevance, or why would they bother to calculate it? It's certainly not apples and oranges to compare predictive power ratings to predictive betting systems. Same concept: using past data to predict future outcomes. MOV rating systems always outperform binary W/L systems in predicting the future. Why? Because they partially neutralize the element of chance that is inherent in every sporting event. As it turns out, run/point-scoring/prevention during blowouts has predictive value. Looking only at wins and losses ignores this data. It all evens out over massive samples (thousands of games), but not necessarily over a couple hundred games (as your system proves). No one is creating betting systems for the sake of measuring the past; they are trying to project what will happen going forward. Ignoring a truly anomalous MOV number is very similar in effect to overfitting: past results look great, but the system will likely fail going forward.

You wrote:

You're comparing the systems I've posted to the ones I've mentioned as being weak?? Seriously?? I know all about power ratings, they have nothing to do with finding a profitable system. Apples and Oranges. All systems are back tested, how on earth can you determine if they are profitable if you don't?? Again, MOV has nothing to do with it. We aren't creating Power ratings, which use MOV. We are finding profitable situations by W/L. You don't get bonus money for margin of victory, so why would it matter?? If I win a game 5-2, win another 6-1, and lose another 9-0, whats the collective MOV?? I win 2 games by a cumulative score of 8 runs, lose the other by 9. That's a negative MOV. Yet, you think that makes a difference?? It doesn't. Also, you don't need 100 plays a year to be a valid system. You need consistent winning results and logic. Call it arguing, debating, I don't mind having a discussion. I'm here to help, not kiss people's asses. I don't sugar coat things, I call it like I see it. We can agree to disagree, but I think you are completely misguided applying MOV to the system, to me its irrelevant. Curious as to what some of the SO people think regarding MOV??


You wrote:

*Small sample size??
216 plays in 6 years, avg is 36 plays a year.
I don't care what the average run margin is. It doesn't matter. If a loss is by a blowout, that skews the run differential. Again, this isn't quantum physics.
You failed to mention that the 57% win percentage is for an underdog. That is a HUGE edge. If you don't think a 57% win percentage for a system that plays strictly underdogs is good, then I have no idea what you are looking for in a system.
Run margin has no bearing on the strength of this system. None.*

link

answered 07 Mar, 02:15

Mark%20P's gravatar image

Mark P
917713
accept rate: 0%

You're entitled to your opinion.

If these systems lose this year, it will be the first time in 9 years that collectively they will lose. As for measuring the past, if a system has shown a history of losing, thats a good thing??

Can you give an example of a good system??

link

answered 07 Mar, 02:44

smartplays1127's gravatar image

smartplays1127
1463310
accept rate: 0%

With a 10 year data base, very few if any systems will have thousands of games to reference. So why should Bet Labs even exist if you need thousands of games??

If you want to make this quantum physics, that is certainly your prerogative. I have used Bet Labs systems to show a significant profit, and see no reason that should change. If you would like to pan my systems and sample size, thats your right. It doesn't mean you're correct, given the results I've seen using them.

And no, my comments haven't been "significantly" edited.

Amazing you pan my systems, when the think tank is littered with systems that aren't relevant. About 95% of them.

This is my last post. I'll give you the last word. My systems work, the past proves it, my bankroll proves it, and no amount of quantum physics or any other method you want to use - MOV which doesn't prove anything despite your pronouncements otherwise - proves differently.

Enjoy the think tank, littered with useless systems. I was trying to help, but obviously it isn't needed. Why don't you post some successful systems, and show people how to do it the right way. Apparently you have the answers. Please show us all how its done.

For all your criticisms, you haven't offered up 1 system.

link

answered 07 Mar, 02:45

smartplays1127's gravatar image

smartplays1127
1463310
accept rate: 0%

edited 07 Mar, 03:18

Your answer
toggle preview


Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here

By RSS:

Answers

Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "title")
  • image?![alt text](/path/img.jpg "title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Tags:

×15
×1
×1

Asked: 03 Mar, 13:11

Seen: 942 times

Last updated: 15 Mar, 13:58