Is this reasonable, or are my results off?

[mebaran]

Over a series of 643 bets (spanning 3 seasons) I've backtested, I achieved +102.4 units flat betting opening lines. In that same series using Kelly, I would have achieved +1634.7 units. Is this a ridiculous number, or is this in range of being feasible?

If I started the series of my bets with $100, I'd be up to $1734.70 after 3 seasons using Kelly (full Kelly). Has anyone using Kelly had, or heard of, results like this?

I'm considering pursuing this fully, as this has not been shown to be a fluke, but a trend over many seasons.

[Maverick22]

did you test your model/strategy using data, that wasnt used to create this model/strategy?

[mebaran]

did you test your model/strategy using data, that wasnt used to create this model/strategy?

Yes. Out of sample data was used to create the model, and 3 consecutive seasons outside that range were used to test. I assume, and it may be ill-advised to assume this, that 643 betting opportunities is a large enough sample size to start making me salivate with results like this.

But beyond data-mining issues, is there anything else that immediately comes to mind that I may be overlooking here? I've gone over the results numerous times, and have had some of my graduate math student buddies take a look as well without giving away what exactly the results belonged to, and so far, nobody has caught any mistakes.

My original post is more or less asking whether or not anyone has heard of these types of results being achieved. I'm sure if they are being achieved somewhere, nobody will ever hear about it, but nonetheless, I am interested in others' results with their "best" models being used with Kelly utility.

[Darkside Magick]

in a study by jane hung done at the university of washington in 2010, she applied full kelly at a series of bets with 55% winners with a random generator with -110 odds. she started with 100 dollars and at 1000 bets she was at over 3500 dollars. so yes i feel your results are accurate

http://www.math.washington.edu/~morrow/336_10/papers/jane.pdf

[sapidoc]

It seems like there is sort of two questions here...

1) +100 units over 643 bets is very good. Is it reasonable? Sure! Will it continue... Probably not. I'm just guessing though --maybe you did in fact create an amazing unbeatable model. There's not a lot of information posted to evaluate.

2) +1600 units using Kelly vs +100 straight betting doesn't really surprise me. Essentially if your 643 bets were large enough edges allowing you to make +100 units. Applying Kelly would definitely extrapolate this number. Kelly will also emphasize downswings just as much.

[mebaran]

in a study by jane hung done at the university of washington in 2010, she applied full kelly at a series of bets with 55% winners with a random generator with -110 odds. she started with 100 dollars and at 1000 bets she was at over 3500 dollars. so yes i feel your results are accurate

http://www.math.washington.edu/~morr...apers/jane.pdf

It seems like there is sort of two questions here...

1) +100 units over 643 bets is very good. Is it reasonable? Sure! Will it continue... Probably not. I'm just guessing though --maybe you did in fact create an amazing unbeatable model. There's not a lot of information posted to evaluate.

2) +1600 units using Kelly vs +100 straight betting doesn't really surprise me. Essentially if your 643 bets were large enough edges allowing you to make +100 units. Applying Kelly would definitely extrapolate this number. Kelly will also emphasize downswings just as much.

That's what I wanted to know, thanks. Initially I thought there had to be some error, but after hearing some of you guys, I'm easing up a bit.

[Pot luck]

+100 units over 643, if that were to be repeated indefinitely would correspond to an edge of 15.55%. Plugging that into the Kelly calculator http://http://www.sbrforum.com/betting-tools/kelly-calculator/ gives a median bankroll of 236650.3306% when bets are placed consecutively.

[statictheory]

Over a series of 643 bets (spanning 3 seasons) I've backtested, I achieved +102.4 units flat betting opening lines. In that same series using Kelly, I would have achieved +1634.7 units. Is this a ridiculous number, or is this in range of being feasible?

If I started the series of my bets with $100, I'd be up to $1734.70 after 3 seasons using Kelly (full Kelly). Has anyone using Kelly had, or heard of, results like this?

I'm considering pursuing this fully, as this has not been shown to be a fluke, but a trend over many seasons.

just curious, whos opener you looking at?

[mebaran]

just curious, whos opener you looking at?

Multiple, but in this scenario, I've used BM's