i've thot a ton and experimented a ton with all sorts of bet-sizing regimes over the last few years...
being an economist and noticing that countries that take a more simple (passive application of a formula) approach to their money-supply policy have a smoother growth and stability experience than those that more actively try manage interest rates... has biased me toward less active management/more deference to simple target-achieving formulas...
also the Kelly-Criterion is based on TARGETING (not risking) a % of your roll commensurate with your perceived expected value. In sport betting this is simply the ratio of your best-estimate win% vs. the book's implied win% buried in the odds...but due to the uncertainty of even your "best" estimate, i've concluded it's best to abandon trying to do this on an individual-play basis...and instead assume a flat % "edge"...as you operate, your ROI will objectively tell you just what that edge is...looks like yours is at least 5% over the haul...my version of low-variance Kelly (at one-fifth strength, say...to make ruination before doubling highly unlikely) then would prescribe targeting 1% on each and every play...cuz u'r assuming that all plays you approve of are in essence being peeled off the same giant never-ending deck of 5%-edge plays...
_____________
the funny thing u'll notice tho' is how little this'll change anything really for you...cuz when u're thinking in terms of how many units to RISK...you're actually already adjusting in similar ways...
your big dogs are always to risk low units...and your big faves usually risk much more...
risk 2 units @ 5...vs., say 8 units @ 2 look very different (like you're pounding the latter vs. the former) but in effect they're both the same in "ToWIN" terms: they both are targeting 8units of net gain...(the odds themselves dictate the difference in amount risked)
switching focus from the standard risk-based protocol to the "fixed target" protocol just simplifies and standardizes this adjustment process..."more likely to hit" fave plays are WORTH a larger risk to win the same target and "less likely to hit" dog plays simply shouldn't cost as much cuz the likelihood of cashing is lower...
if you think in terms of units risked, u actually have to compensate (with EV-estimating formulas or more intuitively)...but if u operate in terms of TO WIN it's all automatic
being an economist and noticing that countries that take a more simple (passive application of a formula) approach to their money-supply policy have a smoother growth and stability experience than those that more actively try manage interest rates... has biased me toward less active management/more deference to simple target-achieving formulas...
also the Kelly-Criterion is based on TARGETING (not risking) a % of your roll commensurate with your perceived expected value. In sport betting this is simply the ratio of your best-estimate win% vs. the book's implied win% buried in the odds...but due to the uncertainty of even your "best" estimate, i've concluded it's best to abandon trying to do this on an individual-play basis...and instead assume a flat % "edge"...as you operate, your ROI will objectively tell you just what that edge is...looks like yours is at least 5% over the haul...my version of low-variance Kelly (at one-fifth strength, say...to make ruination before doubling highly unlikely) then would prescribe targeting 1% on each and every play...cuz u'r assuming that all plays you approve of are in essence being peeled off the same giant never-ending deck of 5%-edge plays...
_____________
the funny thing u'll notice tho' is how little this'll change anything really for you...cuz when u're thinking in terms of how many units to RISK...you're actually already adjusting in similar ways...
your big dogs are always to risk low units...and your big faves usually risk much more...
risk 2 units @ 5...vs., say 8 units @ 2 look very different (like you're pounding the latter vs. the former) but in effect they're both the same in "ToWIN" terms: they both are targeting 8units of net gain...(the odds themselves dictate the difference in amount risked)
switching focus from the standard risk-based protocol to the "fixed target" protocol just simplifies and standardizes this adjustment process..."more likely to hit" fave plays are WORTH a larger risk to win the same target and "less likely to hit" dog plays simply shouldn't cost as much cuz the likelihood of cashing is lower...
if you think in terms of units risked, u actually have to compensate (with EV-estimating formulas or more intuitively)...but if u operate in terms of TO WIN it's all automatic
