Monday's Data Mining...

Monday's Data Mining... FYI... The computer tells me that: The Vikings are 0-20 ATS since 1989 when they are off a two-TD-plus win in which they had fewer than 3 sacks and less than 20 sack yards, failing to cover by an average of 11.1 ppg. The SDQL text is: [B]team=Vikings and p:margin>=14 and p:sacks

what do the 1990 Vikings have to do with tonight's game?

[QUOTE=hardrock;29772]what do the 1990 Vikings have to do with tonight's game?[/QUOTE] This is a good question and I appreciate the chance to respond. The answer is, "Absolutely Nothing." The point here is that there is an easy way to perform research on past NFL results. It's free and it can be done over the internet. In the past, many professional handicappers paid tens of thousands of dollars PER SEASON to have a queryable database on their computer. They had to update the database with the results and lines themselves and the database was very primitive. Today, anyone on the planet with access to the internet can perform sophisticated research on past results in the NFL, NBA and in major league baseball using the SDQL -- and the database is updated automatically every day. So, if you think that a particular trend in performance is not useful, you can use the database to perform your own research. While the 0-20 ATS trend presented here is clearly the result of data-mining, it was active three times in 2009, so the question, "what do the 1990 Vikings have to do with tonight's game" is not that relevant. If I said, "the Vikings were 0-2 SU and ATS on non-Sunday games in 1990, so you should play against them tonight," then the question is on point. Prof Meyer

Please run the query for ALL NFL teams!! If it wins, even at 59% THAT is much more meaningful than a 20-0 trend. Note, the bye week has to make this one weaker also. I might add for EVERY NFL team trend, run it for the entire league.....if it doesn't show a good record for the league, it likely won't win for that team going forward. This trend DOES make sense in that a 'good' NFL team up by 14 should get 3 sacks vs. an oponent who has to throw every down......... Obviously, we had all week to query the trend, hear about Favre Emails, etc, NOTHING has changed. Obvious line move up. Betting the Jets-4.5 or -5 instead of the -3.5 that was out there all week is the type of bet that would have me shaking my head. Jets-3.5 good Jets -5 donk fest.

I totally agree with Fezzik on this one - I hate team specific trends, especially those that goes back 20 years. I ran the query for the whole league, both on all seasons, the last 10 seasons and since 2006. There seams to be a small bias towards Meyers point. Last 10 seasons its 154-178-1 ATS (70-95 since 2006). I cannot remember which lines the database is based on, so I would not play this, with the line move that has allready happened.

personally I dont think Jets -3.5 was a good idea and I'm happy to take Purple Favres +5

I will try to address some points brought up in this thread. I apologize in advance if the tone is pedagogical. I respectfully disagree that team trends cannot be useful. Teams that have had the same coach for a few years can have team-specific trends. One simple example that has gotten a lot of attention is the Bills after playing Miami. This is a team-specific trend. That said, if you ask a computer to find all situations in which a particular team is at least 12-0, a vast majority of the results will have no predictive value. But perhaps some will. The 0-20 ATS trend that started this was presented, “tongue-in-cheek.” I apologize if it was taken as a recommendation on the game. I titled the thread “Monday’s Data Mining,” to emphasize that the trend was computer-generated and I emphatically stated that we do NOT have a play either way on the game. I was not trying to make a point that this is a good trend to follow. I’m just presenting actual data. Here’s what I would like to accomplish… The goal of NFL handicappers is to try to figure out how a team will perform with respect to the line for the upcoming week. To do this, it seems to me that a first step would be to look at how the team has performed in the past in this situation. This is a universal practice. When a stockbroker wants to predict whether a particular commodity will go up in value, he looks at what has happened in the past under similar conditions. When marketers want to forecast sales, they look at what has happened in the past under similar conditions. It doesn’t always work, but we know we’re not going to win 100% of the time. This, however, is no reason to dismiss the results from the past with the sweeping declaration, “What does the past have to do with tonight’s game?” When I write up a game that has a 26-0 ATS system, a 17-0 ATS trend and a 14-0 ATS on a particular team and a set of 0-12 ATS, 0-9 ATS and 0-7 ATS trends against their opponent, I’m hoping that the play has a 56% chance of winning. When a team has covered, say, 17 straight times in a particular situation, I’m hoping that the parent distribution for the situation is something like 60%. In other words, I’m hoping that the 17-0 ATS situation was not a random fluctuation of a 50-50 event. To make this decision I try to determine when the trend makes good handicapping sense, whether it is recent and whether the margins are strong. If I have the PREPONDERANCE of the trends on my side, maybe I’m 55-60% to win. Maybe three of the eight trends I quote are garbage. That’s why I work so hard to uncover as much relevant information I can about a game. I’m not here as an authority on football. I’m here as an authority on searching historical results. When I present a trend, I’m confidently saying, “this is what happened in the past.” I’m not saying, “This trend will continue forward in the future.” My write-ups are provided as examples of what can be searched with the SDQL. I’m hoping that NFL experts decide to learn the SDQL and use it to perform their own expert analysis. At times it seems like it’s a confrontation between the “trend handicappers” and the “real handicappers.” I apologize for any part I had in this. I would like to work as a TEAM with the experts here to find inefficiencies in the market. I realize that I don’t have the NFL knowledge that many in this thread have. However, in my short time here, I have learned a lot from others and have adjusted my handicapping style accordingly. I believe strongly that someone with a knowledge of the NFL and years of handicapping experience AND the Sports Data Query Language would be a potent combination and one that would have a significant advantage over the linesmakers. Let’s team up and pick some winners. Regards, Prof Meyer

On team trends, you need to expalin to me WHY it would apply to just 1 team! THEN we will fire!! We all know for example the COlts are the Kings of mailing it in (preseason, week 17) when the games don't matter....

I'd like to know if any of this will give me the odds of getting head tonight? Probably not, but I took the Jets a few days ago -3 on a ridiculous buy.

[QUOTE=Prof Meyer;29766]FYI... The computer tells me that: The Vikings are 0-20 ATS since 1989 when they are off a two-TD-plus win in which they had fewer than 3 sacks and less than 20 sack yards, failing to cover by an average of 11.1 ppg. The SDQL text is: [B]team=Vikings and p:margin>=14 and p:sacks