Does anyone want to join me helping test a value betting / dutching strategy?

[zippus]

Just looking for value bets where the market over/under-estimates a selection's true chances of winning. If late money movement set a signal that this may be the case, it's of interest.

[zippus]

By way of an example, I've looked at the performance of strong favourites at a specific race course. I've found that where favourites carry a pre-off book value % greater than the average book value of winning favourites for that race type, the horse wins 57% of the time. So if I take 1/0.57 = 1.74 odds, to my mind I may have value if I back such strong favourites at a price above 1.74. I know this is simplistic and working on averages but I think this may pay off in the long term but it's presently just a theory. (I do recognise that simply taking high book % favourites is akin to backing strong favourites at low odds. So it's important to find value to avoid falling into the trap of regular winning but with low returns).

Strong favs.jpg

[ShaunWhite]

where favourites carry a pre-off book value % (1) greater than the average book value of winning favourites for that race type (2), the horse wins 57% of the time.

So if I take 1/0.57 = 1.74 odds, to my mind I may have value if I back such strong favourites at a price above 1.74.

I know this is simplistic and working on averages

By "pre-off book value %" I'm assuming you mean price (well actually 100/price ie implied odds?)....if that's right then....

If figure(2) was say 2.5. Is the idea to back everything below 2.5 and above 1.74? What's confusing is that the 57% list (ie all horses under 2.5) contains a load below 1.74, so if you exclude those the set you're left with isn't the same set that should win 57% of the time.

It's only working on averages if you're comparing the same list of horses. ie if you back all the horses from list #1 (the avg 57% winners) at an average price better than 57%. I think what you've done is to find an average then assume anything above the average is value. But if you just take anything above average, that set of horses will have a different average.

It's sort of like saying the avg price of an iPhone is £500 so I'll buy all of them priced <£500. But you end up with all the busted ones worth nothing while missing out on the £600 phones that are worth £700. The cheap phones you bought aren't the same inventory as the ones that averaged £500. Maybe that's confused the issue.

[zippus]

Shaun,

I agree! I used the total numbers to explain the value principle and wouldn't simply back all over 1.74 because it's just an average as you say. Ideally, I'd do it on a race type basis - see far right column with '1'. I've compared the ave BSP for the strong favs with their actual win rate. So where I could have got a higher BSP and won more frequently than the odds would suggest, I'm suggesting there may be some value. The key problem with this approach is the lack of data points, as well as the fact that going 'high' could just leave me with the lemons.

I'm still pondering...