Calculating your expected return/edge - why does win% make sense?

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • betso
    SBR Rookie
    • 03-18-07
    • 18

    #1
    Calculating your expected return/edge - why does win% make sense?
    To calculate my expected return/edge I take the average win of bets to return $1 (i.e. if odds on a single game is +300 then this would mean making a 25c bet to win 75c - yielding an input to the average calculation of either -$0.25, for a loss, or +$0.75 for a win).

    It seems alot of people count win% to be used as an input to a kelly or other bet sizing calculation. However this does not take into concern the actual odds on the bet made. So does people that track win% always bet -110 lines?
  • Dunder
    Restricted User
    • 10-26-09
    • 3345

    #2
    Not quite sure what is being asked here but to take your odds of +300 as an example. Obviously anything over a 25% win expectation would represent a +EV situation. Both the odds and the win percentage are key variables in any edge calculation (including Kelly).

    The win % is your calculated probability of winning this bet, it is not related in to previous bets on -110 lines.
    Comment
    • betso
      SBR Rookie
      • 03-18-07
      • 18

      #3
      Dunder, you answered my question. I missunderstood the win% calculation to just count the fraction of bets won and then applying this fraction wildly to any future bet irrespective of the odds (i.e. same fraction would apply to +300 as -110 lines).

      So just tracking win% and use this as an input to e.g. a Kelly calculation does not make any sense what so ever.
      Comment
      • u21c3f6
        SBR Wise Guy
        • 01-17-09
        • 790

        #4
        Originally posted by betso
        Dunder, you answered my question. I missunderstood the win% calculation to just count the fraction of bets won and then applying this fraction wildly to any future bet irrespective of the odds (i.e. same fraction would apply to +300 as -110 lines).

        So just tracking win% and use this as an input to e.g. a Kelly calculation does not make any sense what so ever.
        Here is how I look at it. The problem in your example IMO is that you are trying to use the same % for a wager at +300 and -110. This is where win% is an important element of Kelly. While you may have an average win rate over many different wagers, I am pretty sure that you win more of your wagers at -110 than you do at +300. Those wagers would and should have different %'s wagered on them. If you use an average win% with your average return, you will end up wagering too much on the +300 wagers and too little on your -110 wagers.

        To keep the math simpler let's use +300 and +100 and assume a 10% edge. For the +300 wagers you would have a 27.5% win rate. The full Kelly wager for a 10% edge at +300 is 3.33% of bankroll. The +100 wgaers would have a win rate of 55% for a 10% edge and your full-Kelly wager would be 10% of bankroll. If you combine those wagers (assuming equal # of +300 and +100 wagers) you would have a win rate of 41.25% with a 10% edge and a full-Kelly wager would be 6% of bankroll. This is where it goes wrong IMO as the 6% is too large for your +300 wagers and too small for your +100 wagers.

        I do not comingle my stats for my ATS wagers with my ML wagers to ensure that I am wagering an appropriate amount (I use half-Kelly) based on my risk/reward of that type of wager.

        Joe.
        Comment
        SBR Contests
        Collapse
        Top-Rated US Sportsbooks
        Collapse
        Working...