ATS records, predictive or not?

MrX · 02-09-08, 09:41 PM

More likely than the books wanting ATS records close to 50/50 is that the market overreacts to the natural occurrences of chance that lead to lopsided ATS records.

accuscoresucks · 02-09-08, 09:45 PM

Please note the change of address!

http://atsdatabase.com/

zerocage · 02-09-08, 09:45 PM

They are a smoke screen that is why the books readily give them to you to look at.

curious · 02-09-08, 10:05 PM

Originally posted by Data

Using ATS records (O/U records as well) is rather fascinating subject that deserve to be discussed thoroughly. I was hoping to get it going here.

I do not use ATS history but I am wondering if I am missing something.

My take on this is that it can be of any use only if some other outside conditions exist. For example, if the linesmakers were motivated in any way to produce the lines that make as many teams as possible finishing season with ATS records as close as possible to 50/50 then fading would make sense. Is that their goal? I doubt but this is not impossible scenario.

If we do not make such farfetched assumptions then we should think that the linesmakers make the lines in a way that taking either side would be 50/50 proposition. The ATS records play no role here.

The best "predictive" quality I have found is the biased line. What I mean by this is that the linesmakers, being human, let their emotion guide them and they set the line using an emotional reaction from the past.

I'll give an example of this. There was a stretch this season when Rice was always a +30 or higher dog. For good reason, they lost 4 of their first 10 games by 27, 27, 26, and 27 points. They covered most of the games where they were +30 dogs. Now Rice is a small to medium dog. They usually don't cover as a small to medium dog. So the linesmakers overreacted.

I think that ATS is actually a contraindicator if the team has been playing at the same kind of spread and winning ATS. What happens is that the linesmakers overreact. NE Patriots were a good example of this. They were undefeated ATS the first half of the season, then at the end of the season they went winless ATS. The spreads being given were not justified at the end of the season.

Some backtesting you can do is not to look at ATS record over an entire season. Look at ATS over different time slices. For winning teams you could use first half of the season / last half of the season. Look for teams that had win streaks ATS in the first part of the season, I think you will find that they suffer losing streaks ATS later on. Losing teams do the opposite. They lose ATS for many games in a row, then their spreads get larger and larger and they start covering them.

Something that I think is very worthwhile is to chart teams you are interested in using spread and win / loss ATS week to week. When I say chart I mean like you were charting a stock price. I would look for two patterns, a team that has a spread which is trending up while their win / loss record stays bad, at some point the spread will become biased and the team will start covering. Also, look at teams where the spread is trending down while their win / loss record stays great. By down I mean more and more negative. At some point the large spreads won't be justified and the team will start losing ATS. Now, since spreads vary depending on the strength of the opposing team you may have to adjust the spread for strength of schedule.

Data · 02-09-08, 11:01 PM

Originally posted by curious

Something that I think is very worthwhile is to chart teams you are interested in using spread and win / loss ATS week to week. When I say chart I mean like you were charting a stock price. I would look for two patterns, a team that has a spread which is trending up while their win / loss record stays bad, at some point the spread will become biased and the team will start covering. Also, look at teams where the spread is trending down while their win / loss record stays great. By down I mean more and more negative. At some point the large spreads won't be justified and the team will start losing ATS. Now, since spreads vary depending on the strength of the opposing team you may have to adjust the spread for strength of schedule.

What you say here is thay ATS could be a leadiing indicator of team's winning ability. I doubt this is true but I like keeping an open mind, so I will check this at some point. Thank you for contributing an idea.

RageWizard · 02-10-08, 01:31 AM

I do use ATS and O/U stats that I compile myself. I created a pivot table in excel that lets me narrow down situational meanings and in some cases the stats are just to good not to use. ie..The dolphins haven't beaten the Jets ATS since 2003. They were actually favored late this year so I hopped on the jesters. Another wild one that came to a screeching halt this year is the amazing run of overs at heinz field. Prior to this year there were almost 20 games in a row that went over. Hey don't quote me on the number but it was definitely above 12.

curious · 02-10-08, 01:40 AM

Originally posted by Data

What you say here is thay ATS could be a leadiing indicator of team's winning ability. I doubt this is true but I like keeping an open mind, so I will check this at some point. Thank you for contributing an idea.

Actually I don't care if the team actually wins or loses. I only care what they do ATS. I think that ATS is a leading contraindicator IF the kind of spreads the team is getting over time change. By charting the spreads you can see a trend line either up or down. Chart Rice's spreads and you will see what I mean. Rice's spread trends up then plateaus then starts trending down. They win ATS at the plateu and start losing ATS when the trend line has fallen quite a bit. If you see Rice +30 again this season bet the house.

I don't think I explained this very well. I think a detailed example will be better. I will try to put one together.

HedgeHog · 02-11-08, 07:34 AM

The predictability of past ATS records depends on how you use them. Team trends mean very little to me. The fact that College A has won and covered 9 of the last 10 times against College B is useless because the teams' personnel changes significantly every year, so you're comparing apples to oranges.

Where you can find sound angles is through logical extrapolation of ATS records that include all teams in that particular sport. An NFL example would be: Bet on any Underdog this week that lost outright as a large favorite (-7 or more) in its last game. This systematic play makes sense as your betting a team that is out of favor with a fickle public that puts too much emphasis on the last outing.

So IMO, yes ATS records can be used to handicap future contests. I have several contrarian angles I use, mostly in football and mostly on Underdogs , that as a whole grind out a profit each year.

louisvillekid · 02-11-08, 12:37 PM

i take ATS in consideration, but only on a certain range and in Conf. play. like if team A has covered a spread in a 3.5-6 range at 5-1 or 6-0 on the season, i'll take that into consideration,but if the spread is all over the place, like -1 to -25, i have to take into consideration who they've played, i still look at the current match-up though and see how they had fared in the past 2 games against that team. i also have to look if key players are out or back that would cause such a run. i know for college football i hate when i read stuff like so-n-so is 11-1 ATS in Nov. games on the road the last 3 years. i have no use for that info, cause i'm sure they didn't play the exact same teams every Nov. and on the road, cause in college, generally the teams swap home&away each season.

Willie Bee · 02-11-08, 03:39 PM

Originally posted by HedgeHog

The predictability of past ATS records depends on how you use them. Team trends mean very little to me. The fact that College A has won and covered 9 of the last 10 times against College B is useless because the teams' personnel changes significantly every year, so you're comparing apples to oranges.

Couldn't agree more, Hedge. LT and I have had what one might call spirited debate over chat about trends in general, and I simply won't buy into them as much as he does. Not sure that either of us are necessarily right or wrong, just take different approaches.

durito · 02-11-08, 03:46 PM

While personell changes are important-- I do like to look at records for coaches in different spots. For examples, after a win, after a loss, after a SU loss and ATS loss, after a blowout loss/win. Or versus conference teams, non-conference teams.

I've a lot of coaches that have fairly predictable patterns despite yearly changes in players. These aren't the type of angles that are strong enough usually for me to bet, but as supporting evidence they are nice.

I try to look for things that make sense. Nothing like 9-0 ATS on wednesdays. For example, John Grudens NFL teams (in both Oakland and Tampa) have been very good ATS after a loss, but not so good after a win. It seems that whatever he uses to motivate his teams work better after wins.

chemist · 02-14-08, 07:33 PM

Originally posted by durito

I try to look for things that make sense. Nothing like 9-0 ATS on wednesdays. For example, John Grudens NFL teams (in both Oakland and Tampa) have been very good ATS after a loss, but not so good after a win. It seems that whatever he uses to motivate his teams work better after wins.

That doesn't make sense to me. Is he trying to motivate them to lose? If they do better against the spread after a loss then the explanation may be in the spread rather than the team.

durito · 02-14-08, 08:42 PM

Well, the last word in my post should be loss.

It's a small sample size -- and yes the line is always a factor.

Nevertheless, for example, Gruden is 25-11-1 ATS as a favorite (or pk) in his career after a loss.

Quebb Diesel · 02-15-08, 12:52 AM

i think trends are highly overrated...to me its just like the roulette ticker...every game is independent of each other even though there are slight edges in coaching, match ups, and rivalries. trends may be a little bit more relevant to ML plays, but still...every game is independent of each other. high ATS success is bound to even out eventually...its like flipping heads 75 out of 100 times in one data set. the next 100 could be vice versa...i havent did much research on it at all...but i would like to devote some time examining ATS lines because i feel there could be weaknesses into a handful of ATS lines every day...finding a few biased ATS lines could be extremely beneficial on days where there are 35+ games.

JEMFS · 02-15-08, 01:11 AM

When I factor ATS I usually go back only 3-4 years to keep personnel and coaching rosters relevant.

If kept in the proper perspective I feel ATS trends can be helpful.

Wheell · 02-15-08, 01:55 AM

I can only speak of the analysis that I did but when I did ATS testing in the NBA, NFL, and NCAA I found almost no predictive value for ATS records. The only exceptions I found were of head to head matchups where for whatever reason one team had covered either 6 straight or was above 90% for any length of time of 10 matchups or greater. In those games you would beat the vig in betting for the ats dominance to continue.

You should note that while the overall sample size was quite large, the sample size of the predictive variables was quite small, and through sheer noise you might expect to find some predictive variables. I would generally ignore ats records. The market adjusts fast and it probably is noise.