Hi All,

again might I remind us all that this thread is about Risk Management, not staking. Almost everything I've ever read about Kelly stipulates that it increases risk. Therefore, as a RISK MANAGEMENT technique, it's a poor one. (Please note that I'm not commenting on it's suitability or performance as a staking plan, just on it's risk managmenet capabilities).

To Crasher:

the retirement staking plan is ok as far as risk goes, but a better one is the one that EJ Minnis espouses occasionally over at Ausrace. The main difference between EJ's and the retirement is that the target is a miserly one PER RACE. Rather than trying to achieve a target per series of bets, it attempts to achieve a small target per race, and the target is suggested as around 10c per race. Therefore, while the entire series of bets are in front, it will not increase the multiplier.

The major problem I have with a lot of loss chasing plans is that they tend to increase the multiplier as soon as you get one or two losers in a row, regardless of whether you've you've just won 10 in a row. Each time you reach your target, you start a new series of bets, and if you lose the first one, you start increasing your multiplier almost immediately. With a "per race target" approach, it effectively uses past profits to soak up a few losses. I have actually written a program which implements this and have offered it freely to anyone who wants it.

TO DarkyDog2002..... thank you for your well thought out and insightful comments :roll:

Cheers,
Chris.

La Mer: I think Stebbo is correct - we appear to have gone a little off track on this discussion - perhaps instead of being name 'risk management', 'money management' might have been a better heading.

Anway, for my two bob's worth, whatever the benefits of Kelly’s Criterion in regard to optimizing a punter’s bank, in its truest form it fails to measure as a practical tool.

I know of people who use Kelly, but not one of them has claimed to be an adherent to the use of a full Kelly.

So where does it fail? On two out of four main points.

Barry Meadow a US pro punter and considered to be one of the smartest handicappers/money managers around offers a four point criteria for the correct use of money management:

a)keep records on all bets and expenses;
b)do not overbet in relation to your bankroll;
c)bet more when the value is greater; and
d)bet within your emotional threshold.

Clearly, for most punters, the Kelly criterion fails in regard to both b) and c) above.

As Becareful succinctly points out and as would be true of most of us, to bet(say)50 percent of your bankroll on one horse is way beyond our emotional threshold as well as totally overbetting in relation to the bankroll.

So what are some of the better methods of maximising profits (assuming a profitable set of selections)?

There are variations to the Kelly, such as the half-Kelly as well as another variation named the Super-Kelly (also known as the Hyper-Kelly). However, to operate effectively, Kelly betting requires a precise estimation of a horse’s chances and therein probably its downfall as even a small variation in a horse’s estimated chance can mean large errors in application.

I personally use a progressive level stakes milestone approach, that requires a 4 percent level stakes bet until such time a predetermined milestone level of bank is reached whereby 50 percent of the profits are withdrawn and placed into a reserve bank, betting then recommencing at 4 percent of the higher bank level, i.e.
$1000 commencing bank equals $40 bet (1000*4%)
When bank reaches $1250 - half the profits withdrawn ($125)
Recommence with new bank of $1125 equals $45 (1125*4%)
Continue in this fashion until emotional threshold reached, then go back to last milestone and recommence.

So long as the profits continue, this can go on ad infinitum, or until there is enough confidence to lift the emotional threshold.

What I mean by this it that there is little point in (say) starting at $40 with bets rising to $200 over time, if the emotional threshold is reached with $100 bets.

When the heart starts pounding, when a punter starts to regret the fact that a bet went down with such an amount on it that they say to themselves such things as ‘that was a week’s pay’, or ‘gee the amount of that bet would have bought a new TV’ then the emotional threshold has been over-reached and the punter needs to have a big rethink, remembering that they should only be prepared to lose what they can afford.






[ This Message was edited by: La Mer on 2003-10-21 15:45 ]

[ This Message was edited by: La Mer on 2003-10-21 18:48 ]

In my opinion any staking plan should do two things....

Maximise profit and minimise risk.

Therefore I think that staking plans are pertinent to the risk management argument.

Wow - there's a lot of good things on this thread already. Most of what follows is not new but perhaps just a different spin.

The following risk management plan comes from a source that will remain anonymous lest the censors razor is invoked.

Bundle a continuing series of bets into Blocks of say 20 bets (30, 40 or whatever). The Starting Block Bank remains static throughout each block of bets irrespective of current win/loss - and is only recalculated at the commencement of each new Block of bets.

Individual bet size is calculated as a % of the prevailing Block Bank. This allows for different sized bets (1%, 2%, 3% etc etc) and all bet sizing whims including Kelly.

At the end of each Block a profit or loss is declared. A proportion of Profit maybe siphoned off for preservation - a new starting Block Bank is calculated from the residual profits plus the proceeding Block Bank. Hence, individual bet sizes in the forthcoming Block of bets will be proportionately larger.

In the event of a Block of bets producing a loss the new Block Bank remains EXACTLY the same as the proceeding Block Bank. Accordingly, the following series of individual bets remain the same (by proportion).

Sizing of individual bets as a percentage of Bank must pay homage to Maximum historical DDown and then some (Stebbo has either been reading my mail or has read the same book).

As for Staking Plans such as Power of Ten the good news is that I own the Software Edition. What am I bid?

With due respect to all posters, I think we have all started well on the subject of risk management but ended up on staking plans. Have a quick squiz and you will see what I mean. Taking a leaf from Stock Market capitalese, the correct term I think is "Asset allocation management".

Cheers.

[ This Message was edited by: crash on 2003-10-21 18:13 ]

In that case the thread may as well stop here and either bet level stakes or percentage of bank reducing.

My offer stands to anyone interested in testing my staking plan which maximises profit and reduces risk. (see above).


[ This Message was edited by: Chrome Prince on 2003-10-21 19:20 ]

crash

Risk Mangement for me includes two competing requirements, GROWTH and CAPITAL PRESERVATION. Aside from backing only (more)winners (or less losers) GROWTH can only be achieved by increasing bet size (increased RISK). The Kelly criterion is proven to optimise GROWTH but, as others have pointed out, only at unacceptable levels of exposure to CAPITAL. So, how to balance GROWTH vs CAPITAL PRESERVATION???

I think you mentioned representing the ups and downs of equity within a graph. The plan outline above is specifically designed to smooth out those bumps and troughs, and furthermore, is in sympathy with the punters psychologic health (i.e. sets short term goals, take rewards).

I thought it un-necessary to spell this out in my original post.

You take the risk in order to make a profit.

* Criminals take the risk in order to make a profit.

* You buy an investment house to take a risk in order to make a profit.

Let's talk profit and then risk will come into the discussion.

I haven't heard risk yet.

* Short odds less risk versus long odds high risk.

* High strike rate less risk versus low strike rate high risk.

* Calculated run of outs before loss of bank versus Stop strategies to loss of bank.

* Streaks and THE RISK of streaks.

* How to evaluate your own Edge and how that equates with streaks

* How you back a race on identified factors versus how you back a race on actual measurements of ability.

* How much complacency and inertia stall one's ability and accuracy in finding accuracy in measuring risk.

* How to rate a method. Which is how risky a method really is.

* What are the 'what ifs' that are never considered that make up risk.

* What new things make risk analysis difficult to measure in a given field

etc etc etc.

Risk is relative.
Would you risk 100% of your bank to make 1% profit?
The discussion has been criticized for tilting towards staking, yet surely the staking is key to both the profit and risk element.
In my opinion there are various criteria to be considered before manging risk and staking to arrive at a determined or undetermined profit.

a)Does your method show a level stakes profit and if so over how many bets?

b)Does all the profit come from one or two outsiders or is it consistant?

c)What is your average dividend?

d)What is your strike rate?

e)What is your current longest run of outs and what is the mathematical expectation for same?

From here you can calculate the level stakes risk.

Perhaps this is the time to mention measuring whether or not your method ,should hold up in the future.

Obviously nobody can predict the future, but risk can be minimized by having the greatest amount of data available to arrive at a reasonable conclusion.



SEE ABOVE



One can calculate the maximum run of outs and calculate a stop loss procedure similar to Stockmarket strategies.
What to do then once you've reached your stop loss?
Start from square one and cop the loss or modify the method or staking again.
I do not hold too much with stop loss strategies in general because this game only enables a very small relative POT, and to regain losses either now or in the future involves delving into unchartered territory.



What if my internet connection dies in the middle of placing a bet, something happens to distract you and you miss the one bet that would have made a profit for the month?
What if the power goes out, the ISP packs it in, you have to reboot (MacroHard), the family needs you attention for the day etc etc etc.

Over to someone else...

To All: Apologies if my earlier comments about staking plans implied I didn't think staking plans should be discussed... What I meant to say was that I didn't want this discussion to become one about the merits of one staking plan vs another. Discussions about how staking plans affect risk are clearly part of the discussion.

To FryingPan: You make some excellent points about elements of risk that have not yet come under discussion directly. A few posts have touched on some of the points you raise. Becareful's selection of a 5%-of-bank for his 60% strike rate system for example.

Your points about odds and strike rates are valid, but I think these are decisions that are taken well before we start to think about risk. Personally, I don't particularly care about odds and strike rate, as long as SR times by Odds is greater than 1. I must admit to leaning towards the longer priced selections, as it's an incredible high to pick a $20 longshot, whereas picking Lonhro to win at $1.80 is boring... (In my opinion).

* How to rate a method. Which is how risky a method really is.

I think this is a very interesting point, and one that I hope people here have some ideas about. I personally use MaxDD as a measure of a system's RISK and I also very much like Chrome's SV calculation as a system measure. I've tried to make the point (perhaps badly) that I think Longest Run of Outs is a very poor measure of a system's risk.

Personally, I'm on my third "system risk measure". When I first started this journey, I used strike rate as a measure of a system's risk. "Gee, a 50% strike rate, I can be half my bank and be rich in no time"... I kid you not, that was an initial thought I had many months ago when I was very, very green......

Another extremely vital aspect of risk that I think it was crasher who mentioned in passing is DISCIPLINE (BTW Crasher - I hope you got some of the $6 or $7 being offered about Delzao by the bookies????? - I took $6 on IAS early on Sat as I had to go out :smile: )

Discipline is perhaps THE most important element in risk management. I for one find it very difficult to stop myself from looking for something to bet on when my systems throw up none or just 1 selection for the day... I don't have any impulses to bet on more nags if I've just scored a big win, only when my systems have only a few bets on a given day.....

Another issue that hasn't yet been touched upon is the natural inclination to fiddle with a given system. On another thread here I've stated how I like simple systems, and I think that once you find a system that seems to work, you should run with it for a period of time... I know a lot of people will keep "fiddling", cutting a few losing bets here, adding another filter there, all the while trying to get that "super system".... I try to avoid this urge as best I can.

Cheers,
Chris.