Losing Days And What They Mean

Our picks had an overall losing record for the first day of the 2011 NCAA Tournament. As usual, whenever our picks don’t have winning days, especially around tournament time, we receive a couple concerned emails or comments from our subscribers.

This blog post is meant as an open response to those emails and comments, and more broadly, an explanation of our perspective on the business of sports predictions and opinions on future games.

Being wrong is a reality that anyone who makes predictions has to deal with. Anyone who predicts things — sports handicappers, Wall Street stock analysts, even the weatherman — has to deal with the reality of being incorrect, when a bunch of people trusted them to get it right and often put their money on the line.

When it comes to sports betting, losing 45% of the time means that you’re doing a GREAT job — but it still means you lose almost half the time. And the unavoidable reality for sports predictors is that if someone starts using your advice at a random point during a given season, there is a decent chance that they start using your picks during a down streak.

There is nothing we can do about this. Variance in prediction accuracy happens. It’s a mathematical reality.

Our NCAA basketball ATS picks hit at over 54% for 3,500+ games during 2010-11, a long term success rate that very, very few humans can match. However, that long term success rate was made up of many winning and losing streaks along the way. We have no way to control when winning and losing streaks happen, and it’s not like we can just tell our math models “OK, now REALLY turn it on for the NCAA Tournament” or the NFL playoffs or other big games.

Other handicappers on the web try to dodge this fact with apologetic emails to their customers and claims that they “just hit a bad beat” and are “just about ready to heat up again” blah blah blah. It’s stupid and meaningless drivel. Losing streaks can always continue.

At Team Rankings, we’re not like other handicappers and we’re not going to dodge the issue. To us the message to our subscribers and potential subscribers is pretty simple: Don’t buy our advice if you’re not prepared for it to be incorrect sometimes. And if you’re not prepared to absorb some 0-5 days or 12-25 weeks or whatever the losing streak may be, then you seriously need to reconsider your money management strategies and whether you should be betting sports in the first place — especially if you’re trying to make money and not just have a little fun.

Let’s look at some data to further illustrate the point. Sharp bettors, as we mentioned above, are typically very happy with a long term 55% win rate for spread and totals bets at standard -110 odds. As long as you’re betting a decent number of games and winning at that rate, you’re making a solid return.

If you bet $100 on each of the sixteen Round of 64 games today, at the usual -110 odds, and you pick at a 55% long term win rate (again, very few people do), here are your chances of winning or losing certain amounts:

  • 1.5% chance you lose at least $800
  • 12.4% chance you lose at least $450
  • 25.6% chance you lose at least $250
  • 43.7% chance you lose money overall
  • 56.3% chance you net a profit overall
  • 36.6% chance you win at least $250
  • 19.8% chance you win at least $450
  • 2.8% chance you win at least $800

And if you’re worse than 55% long term, those expected returns diminish significantly.

The main point is this. Even at a 54% long term win rate, our ATS picks, for example, are going to lose you money almost half of the time, and when you are only considering a very small group of games as your sample size, there’s a non-trivial chance they do really poorly.

Our proven long term track record, the fact that our prediction methods are 100% data-driven and systematic, and our complete transparency with our results is how we have built a successful business in an industry that is pretty darn shady. But we absolutely cannot guarantee you winners at any point in time, especially across a sample size of 10 or 50 or even 100 games, no matter how important those games are.

All we can do is publish the smartest analysis we know how to produce. If at any point in time we feel like our objective computer projections fail to offer at least some sort of long term edge over Vegas, you have our word that we’ll stop selling advice. But you’ll get no get rich quick promises from us in the meantime. Consistently beating Vegas is an ongoing slog that takes a lot of work to do, and even longer to prove.

We’ll close this post with an email we wanted to republish. We welcome all feedback, so please feel free to add your comments below.

Date: Fri, Mar 18, 2011 at 9:16 AM
Subject: Re: Purchase Confirmation from Team Rankings (16-MAR-2011)
To: “TeamRankings.com” <support@teamrankings.com

Just want to say im dissapointed with the picks. 5-11 yesterday and maybe one 3* pick. I was also holping for real expert picks and not a computer generated pick. I wish I could have my money back and flip a coin.

Our response:

Thanks for the email. Sorry to hear you’re disappointed, and we certainly don’t like down days any less than you do.

However, if you’re judging us on the results of 16 games, you’re basically asking us to provide you guaranteed winners — and there is no way we can do that. Our NCAAB ATS predictions are 1931-1631-69 (54.2%) so far this year. That’s over 54% winners across a sample size of over 3,500 games:


Any professional gambler we know would be very happy with picking at close to 55% for that many games. With that said, across those 3,500+ games there were almost certainly many losing runs of 0-5, 2-10, 5-11, 13-30, whatever. That is just the unavoidable result of randomness and variance. Even a long term 60% picker — which, despite the false claims of sleazy handicappers on the web, almost no one in the world is — can EASILY go 5-11 over 16 games.

Regarding the point about “expert” vs. computer picks, it seems like what you should want are winning picks more than anything else. I’ve shown you our long term track record for this year above. However, how we do across the 67 games in the 2011 NCAA tournament will have a lot to do with randomness — they are just 67 games out of the 3,500+ we predict each year.

Because our long term track record is good, you have a better chance to have a winning tournament in 2011 using our picks compared to flipping a coin. But there is still a decent chance we have a losing tournament, and if you’re not prepared for that, then all I can do is encourage you to review your money management practices, because losing streaks can and will happen and should be expected.

Our main competitors are Internet handicappers who seem to think that the fact that they went 5-1 in their last six games is meaningful evidence that they are good cappers (it’s not), that claim that because they got their last four picks right they are “on fire” and have a great chance to get their next pick right (complete nonsense), who offer “guaranteed winners” (what good is getting $25 back for a bad pick after you’ve probably lost a lot more than that on the bet itself?), and who do detailed writeups of each game just because they know people like them and because they can use the writeups to “rationalize” their wins and losses. It’s all baloney. Winning at sports betting over the long term is hard and the sports books are in business for a reason.

We can’t guarantee you winners. All we can do is the smartest analysis we can, and our best efforts there all go into our mathematical prediction models, which are a lot more unbiased and sophisticated than the processes that self-proclaimed “expert” human cappers use to analyze matchups, no matter how much they write about why they made a certain pick.