Do you use Lay Kelly (staking plan)?

[London_Calling]

Because they ought to award medals for those who do . . .

I'm not quite happy with the quantity of data I have yet to settle on the 'expectation' value but I've looked at Lay Kelly - at the numbers - and it's . . pretty intimidating for a newcomer to the system.

I'm still keen to go for it when I have a little more data, just wondered if others here used it and what they're experiences are?

[jokerjoe]

Not 100% sure what you mean, but I always use Kelly staking if I can come up with my own probabilities. If laying you shouldn't lay more than one selection as two lays are not mutually exclusive. Also never use full Kelly, backtest to find what you're comfortable with.

If you want some discussion on Kelly staking I highly recommend Ganchrow's posts on SBR Forum http://forum.sbrforum.com/handicapper-think-tank/

[doris_day]

For some years I tried a variety of staking plans but I quickly discovered that level stakes is the easiest to handle and usually ends up with the best result.
I've done loads of analysis with The Staking Machine and it consolidates my opinion that if you can't make level stakes pay, then change your system.

[osknows]

Doris is right, essentially if you don't have an edge then no staking plan will work. I use percentage staking with a few alterations. This is a simple calculator I put together to work out odds required for given % strike rates and vice versa. It also works out profit based on number of selections http://www.mediafire.com/?020mh35b6yj866m

[London_Calling]

Thanks for everyone's comments. We may as well assume there is an edge and it pays at level stakes. The issue is then whether level stakes/percentage laying is more profitable than Kelly type proportional laying.

If osknows and doris have both put extended sequences through The Staking Machine (or similar) to come to the same conclusion it's significant - it would also make TSM redundant.

I suppose we also have a transparent example of percentage laying succeeding in Maria, as well.

Thanks again for the comments - this afternoon I'll be digesting (as best I can) the sheet in jokerjoe's link and the one osknows kindly put up.

It's a complicated old game . . .

[London_Calling]

Just to add one more thing; with regard proportional laying, I've never seen an argument for using (staking or liability) any given percent of your bank, but which is typically 1%, sometimes 2%.

Does anyone know where and why - it seems the amount of the bank to use is entirely arbitrary as long as its very small?

[osknows]

The lower the stake the less chance there is of going bust. Percentage staking theoritically means you will will never go bust which is probably why Betfair has a £2 limit. If the edge & average odds is known for a system, it's possible to run montecarlo simulations to estimate the possible ranges and probability of going bust/making profit.

This is a really good book which explains most of what you need to recreate the staking machine methods http://www.amazon.co.uk/Fixed-Odds-Spor ... 569&sr=8-1

[osknows]

The book essentially walks through each staking method (I promise I'm not making any money from my recommendation )

From memory, I believe the book advocates...

1. Kelly staking is the best method to maximise profit and minimise loss- but you need to be confident in the calculation of actual probability for it to work. Half Kelly staking significantly reduces risk of failure.
2. Percentage staking is potentially more profitable than liability staking and both are potentially more profitable than fixed level staking - BUT only because the increase in profit is offset at the expense of increase liklihood to lose.
3. No edge - no profit
4. Martingale system and progressive staking (loss recovery) are doomed to failure at some point

It doesn't cover Maria's system. My thoughts on it are it's actually quite ingenious as it uses both a ratchet system (i.e. fixed stake when losses reach certain level) and increases stake for smaller odds. Intuitively it feels like it adjusts the staking method based on current circumstances but I can't back this up with any data....

I'm also interested in another staking method (not sure the name of it) which advocates reducing the stake progressively everytime you make a profit then reset to a larger stake following a loss. This would seem to minimize losses especially when used in conjunction with a percentage stake. Again I have no data or experience to confirm how well it works.

Another consideration which is rarely mentioned is liquidity. If you can't get the required stake matched consistently then the staking method needs to be adjusted to account for the average available stake.

Interesting thread though....

[London_Calling]

One point I haven't quite grasped is this; if you have a strategy that seems to work and you know this because you've honed it and you have a history of actual bets (rather than backfitting) that have already produced a ROI/Yield, then you have the material to run through The Staking Machine to see what would have been the best staking plan?

At that point you already have the odds, the strike rate, even the actual sequence for this particular strategy . . .


My other point is more instinctive, I wonder if different back or lay strategies work with different staking plans and it's too generic to talk in general terms about proportional, level stakes, percentage, etc?

Agree with you osknows, it's an interesting chat!

[osknows]

That's the point of backfitting, you have number of selections, odds, and strike rate historically but all of these will vary in the future. The better your system/analysis the more able you should be to predict the future course of events. That's where montecarlo and other risk analysis helps; the variance in selections, odds and strike rate can be modelled to provide a view of likely outcomes with associated probabilites/propensities.