Originally <a href='http://www.sportsbookreview.com/forum/showthread.php?p=25176977'>posted</a> on 01/20/2016:

For my 5000th post I would like to start a series of threads known as Words of Wisdom. While I considered offering something I’ve already posted or recent words of money management, I decided on something new.

I’ve often mentioned that it’s not that gamblers don’t know; it’s that so much they know is wrong. In the CFL thread poster Ra77er asked a question that offers a teachable moment.

Quote Originally Posted by Ra77er View Post
KVB why do you prefer median over mean? Mean is much more accurate right? I know this is random question in this specific thread but you spoke about this before with regards to modeling in the HTT.

More specifically mean would seem to account for anomaly type situation better than a median approach or no?
Remember, this is indeed a game of numbers if you want to succeed and many bettors, when comparing stats between teams, often use means instead of medians. Doing so will cause errors in your work.

The first lesson is in conclusions and behavior. I’ve mentioned in the past that there are as many ways to handicap as there are handicappers and bettors, when comparing teams, use all kinds of stats. When bettors use means when comparing teams, get false evidence, then fail, they often conclude that the statistics used were the wrong ones.

They move on to other angles, other stats, and then continue to use means and continue to fail. What they don’t realize is that it’s often not the stats used that is the problem, it’s the use of means. This cycle can go on for as long as a bettor lets it.

The mean and the median can be different numbers but are often close. First let’s address the problems with averaging stats.

Let’s keep it simple and, simply put, using means when comparing stats between teams will cause errors whenever both teams are above or below the mean. It fools the player because you can, at times produce a valid result, as when medians are used, but that is coincidental.

Let’s use an example. Let’s compare the Eagles to Falcons using yardage stats. Let’s say, for simplicity, that the average yards per game in the NFL come to 160 yards. We know it is closer to twice that but let’s keep it simple. In our scaled down example, let’s say the Eagles gain an average of 200 yards per game and the Falcons give up 200 yards per game.

To project the Eagles total yards gained against the Falcons, the natural tendency for the bettor is to take the average of the two: yards gained by Eagles plus yards allowed by Falcons divided by two, which would produce a prediction of 200 yards per game.

Does that sound right to you? Think about it. Using means, we are saying a better than league average Eagles team that’s “worth” 200 yards a game is going to be held to its “normal” 200 yards per game by a less than average poor Falcons defense?

That’s a mistake and it will show. Instead of trying another stat and again averaging the two, try this…

Add the yardage gained to the opponent’s yardage allowed and then subtract the league average.

Now we get 400 – 160 = 240. Let’s think about it again. Now we’re saying the Eagles good offense should get more yards against the poor Falcon defense than they would usually get against their average opponents.

That makes a little better sense. Now let’s talk about the advantage of medians. Ra77er brought up anomaly type situations and how the mean would take it into consideration. While this is true, is that what we want?

Let’s say, in our scaled down example that the wheels come off in a game and those pesky Falcons give up 400 yards. They are way behind, injuries, send in the backups, etc. and don’t give a shit. In the next three games let’s say they give up 180, 220, and 200 yards per game. If you average these four games you’ll come to 250 yards per game while the median is 210.

In a sport like the NFL teams tend to revert to a norm, whatever their specific norm is. By using medians, you can better remove those outlier results that aren’t exactly indicative of a team’s normal performances.

Winning gamblers have learned to compare what is relevant to staying ahead of the marketplace and often times those outlier results are more of an influence on perception than reality. This too should be taken into account.

One more note about means and medians. I mentioned that means can fool players because often times they can produce a valid figure, but it’s a mere coincidence.

Let’s say the Eagles gain 200 yards per game and the Falcons give up 120 yards per game. Averaging the two will yield the same result as adding them together and subtracting 160.

And that is why bettors can be fooled. When one team is above average and one team below average, you can get close to a valid figure, and this happens often. Eventually averaging between teams will produce errors while the method described above will avoid these errors.

Like I’ve said before, track your bets and why you made them folks; this can become the most valuable piece of information you can process. Not even the most advanced syndicates and bettors have that information.

That is, until we make you. But that’s for another thread.

May these words of wisdom put you onto some winners or take you off of some losers.