Imagine the following fictional scene: Steve Wynn walks into a boardroom deep in the bowels of his casino – a part of the property people like you or me don’t even know exists.
Seated around a long oak table are all of his top managers in very expensive suits. Mr. Wynn slams down his briefcase and says: “I better get some answers or heads are going to roll!”
All the executives are now fully attentive, waiting for the other shoe to drop. “Before this meeting,” Wynn continues, trying to control his seething anger, “I was watching surveillance on blackjack Table L, and a player won 19 of his last 27 hands, which means we lost 19 of the last 27 hands at that table! What are we going to do about this? What changes are we going to make to prevent this from happening again?”
As most of you know, this scene would never happen. It is quite instructive, though, to contemplate why not. For starters, Wynn knows that unless a player is a world-class card counter or some other type of cheat, the house has a mathematical edge in blackjack. Over the long run, the more hands that are played, the more money the house will win.
The main goal is to encourage as many hands to be played as possible. The inevitable short-term losing streaks are no more of a concern than an interception by Peyton Manning. It happens, but not enough to question the game plan. This certainty is an advantage that is hard to come by for a sports bettor. And dealing with such uncertainty is a great challenge.
Last week we discussed the basic requirements for a handicapper to have some idea of his true winning percentage. First, accurate record keeping is required. In addition, the sample size of bets needs to be big enough that streaks – good or bad – are not mistaken for the long-term truth. But what exact category of games do we consider together?
If a handicapper is a loser at baseball but 55% at NFL, do we separate those games in our analysis? The valid case could be made to do so, because the expertise required to handicap each is quite different. But now, we need to make sure the sample size of the smaller group of NFL-only games is big enough
What if this NFL expert is a coin-flip on pro football sides, but 57% on NFL totals? This leads to the same questions about separating results – and similar concerns about sample size.
In last week’s article, we explained many math guys want to see 600 bets before being confident results are long-term. Imagine how long it would take to reach 600 NFL total bets. Even with high volume of five per week it would take six full seasons to reach the minimum number for confidence in the results.
Except now we are faced with a number of other complications. How has the NFL changed during those six seasons? And how has the betting market changed? And how has the handicapper personally changed? Maybe age has taken its toll. Or maybe the joys and obligations of fatherhood have the handicapper taking shortcuts today he would not have even considered a half decade ago.
In the extended cut of Apocalypse Now it’s said you can never “step into the same river twice.” I believe it’s equally fair to say no two handicapping situations are identical. Variations between them increase over time, ultimately becoming different to such a degree that comparisons are no longer valid.
This inability to fairly compare almost always happens before there’s a large enough sample to be confident in the truth and accuracy of long term results. The sports bettor does not have the advantage of facing losing streaks with the same confidence and patience Wynn has about his blackjack tables.
Next week, we’ll discuss how this uncertainty can wreak havoc with a handicapper.
RJ Bell is the founder of Pregame.com - and co-host of FIRST PREVIEW, heard Monday through Friday at 10 am on ESPN 1100/98.9 FM. Follow on twitter: @RJinVegas. Discussion of this article continues at Pregame.com. Contact RJ at [email protected]