http://almostrandom.com/dietbet/attrition-oct14.jpg

Note: to fully view this image, right click on it and choose "View Image".  It should then be full-sized, but it will open in place of this blog post.  So open a second window with this post in it to read them side-by-side.

Okay, I haven't written about attrition in a while because it's a little bit more complicated and I'm not sure what conclusions can really be drawn from it.  Attrition is the number of people who leave the games over time.  Reasons I'm sure vary, some leave for injury or illness, some leave for financial reasons, and I'm sure some leave because they realize they just aren't going to win.  Whatever reason, they leave behind money, sometimes in a round pot depending on when they drop out and definitely in the final pot. 

This spreadsheet is a different snapshot of the main winnings tracking spreadsheet I keep, but one that only focuses on how the player totals go down as the games progress.  The first two sections at the top are from the other spreadsheet you've all seen - prizes per round and a running total of prizes per round.  The next section should also be familiar, it's the membership per round as calculated by dividing up the round pot (except for Round 6, where I use the posted number of qualifiers).  The rounds that haven't been completed yet, I just copy the last number down the column because it makes some my other formulas happier later.  So if you look at the Sept 23 game, it's only had one round, but I put the same number in each row. 

The fourth section is new - this shows the %age of player loss for a round (calculated compared to the previous round's membership).  Here you can see, that games lose between 2-4% of players each round.  Try as I might, I haven't found a reliable pattern to they way people drop out.  It's not that it's always more people in later rounds.  It's not that there's certain months (say summer months or big holidays) where the dropout rate is higher across all the games.   I think overall, we're dealing with numbers that are too small to have a pattern (see below).  Note the last row is always high - that's the %age of people who don't qualify for Round 6 with a 6% loss at the end of Round 5.  As always, the averages are at the right. 

The fifth section shows a running total of the attrition across the games.  Here's where I think it gets a little more interesting.  Clearly some games lose a lot of people at the beginning but then the losses slow down.  And some games are the other way around.  If you compare the March games, they had a similar pattern the whole way through.  But then look at the April 8 game.  A lot more people dropped out earlier on and throughout the game.  By the end of Round 2, the April game had lost 8% which matches the Round 3 loss in the March games.  By the end of Round 4, the April game had lost 14%, compared to about 10% in the March games.  And yet, the overall Round 5 loss is not very different among those games.  So that tells me that there should be more money left behind in the March games than in the April game - the longer players stay in, the more they contribute to the final pot.  Unfortunately, we don't actually see that - because despite all the people dropping out, the April 8 game had the lowest win rate in the entire game (81.45%), whereas the March 4 game was only a little lower than average (82.63%) and the March 18 game was much higher than average (85.29%) (These numbers aren't shown on this spreadsheet, but the average is 83.47%).  So the extra money might have made the pot a little sweeter, but it still comes mostly down to what percentage of people actually reach the goal. 

The sixth section shows in raw numbers how many people are actually dropping out of the game each round.  It's just simple math on the Membership section, but it's sometimes easier to see this way than to mentally do the subtraction on that section.  This is where you can see why I'm not sure we have enough people to actually see any patterns.  When 15 people drop out of game in a round, there's a lot of different factors for there to be any one trend. 

The last two sections are my attempt to make sense of the money left behind in the final pot by the people who leave.  A person who leaves in the first round is leaving behind their Round 1 stake of $12.50.  A person who leaves in the second round leaves behind stakes for Round 1 and Round 2, so $25.00.  People who make it through Round 5 but then don't qualify for the final round by making the 6% goal leave behind a whopping $67.50 each.  That's where the really big prize money comes from.  The first of the two sections is just a strict calculation per round with a total at the bottom, the second of the two sections is a running tally so you can see how it grows.  Note, this is the money of the people who don't play the final round.  The final pot includes this as well as the contributions of the people who do qualify and also their Round 6 contribution if they haven't pre-paid.  I have played around this these factors a few times and I haven't reliably figured out how many people actually pre-pay for the game, sorry.