Back on the topic of risk management, here goes a long post :wink:

I've been doing various analysis over the past few days and some results and ideas have come to mind after running numerous scenarios.

These are not gospel, just my ideas after observation for minimising risk.

1. Calculate the risk of your "system" by comparing the maximum drawdown with POT.

Stebbo demonstrated this earlier and he got me thinking - thanks stebbo!

Ideally your risk must be less than your POT.
Otherwise your risk is greater than your profit and sooner or later you'll bust.

The greater the difference or ratio between POT and risk e.g. 2:1 the better.

This is no guarantee of future profits, but based on the data available, this would be perhaps the best guide.

Maximum Drawdown should always be less than the profit, as the entry point of a series of bets can make or break or even give false optimism.

Example: Start with $1,000 have a good run and win $800. Have a bad run next and lose $1,000 but you're only down $200 on your initial bank.
Had you entered at the second point you'd have lost your entire bank.

2. An increase massive increase in profit due to staking can also give false optimism.

Elsewhere I've heard:
"But it's working"
"I haven't gone bust yet"
"I don't give a ### about POT, profit is all that matters"

At the end of the day profit is not all that matters, to me, risk is all that matters.

What's the use in making 500% profit if the risk of losing it all is 600% or even 1000%?

Sooner or later all that profit will be gone unless some form of insurance or safeguard is applied.

3. As a guideline to system viability, I use the profit divided by the average win dividend.
This is a guide only as a couple of longshots will distort the result.

Example:
Profit is $79.50
Average Dividend is $2.10

SV calculation is $79.50 / $2.10 = 37.86

So the profit comes from 37 plus winners which is very good.

While this is not exactly Risk Management by the general definition, it is salient to the argument, because the continuing viability of a "system" (for want of a better word) must be considered the major risk of the investment regardless of the staking or investment method.

Hope this generates some discussion.

Paging becareful to the foyer please :smile:

__________________
Out in front - you make your own luck!

I felt sure my latest addition to this thread would have sparked some interest.

Ho hum.

:sad:

I have found the discussion in this thread interesting but I feel some of the analysis has a serious flaw and that is basing future bets on past results. For example using past POT as a indication for future POT and the basis of future staking strategies. One thing my years at punting has shown is that racing does not follow simple patterns and this is why most systems fail.

My idea of risk management is minimum outlay for maximum return. I have discussed my level staking ideas at length (and possibly ad nauseum) in another thread but no one has yet shown me the flaw with my analysis - other than the need for patience.

John

Chrome,
I'll have a go at a reply.

What you were saying sounds correct on paper but you cant work out things like profit = 500% but risk was 600% [?]. The method is known as "Special pleading"[create a false "fact" and then base your argument on it]. The only fact is the 500% profit, not the 600% [or whatever] risk.

Risk [odds if you like] is a subjective thing in horse racing as in everything and can't be measured with maths [the goal of a system creating a golden egg], If so there would be a maths formula to work out precise odds of a horse winning which as you know, can't be done. Your 600% [or whatever] risk might have in fact been .01% or 1000% depending on the subjective measure [point of view or a collection of subjective measures].

Such things as tossing a penny can be worked out as a 50% chance of head or tails being equaly thrown in eventual totality [pure maths] but not where you are going to be score wise even after a million future throws. You could be rich or you could have lost a fortune and your shirt at any future point.

Gunny makes a good point about backfitted stats. They can never be a calculation of a certain future position profit wise. It's the old penny tossing scenario. Have a thousand flips and note your score, then have another thousand and you will see what Gunny means. You might get a similar result but it could be wildly different too just as easily. I have seen the Maths. on this penny thing by a mathamatician and it soundly dumps the notion that future probability is equal to past probability except to an infinate point of a 50% x 50% result [and we will never live that long].
Your chances are always 50% of throwing your choice per throw but say after your 1000th. throw you could have 300 wins to 700 losses just as easily as 700 wins and 300 losses. Your chances of either are exactly 50% [or if you like, your "risk" of either result is 50%].

A punter could back a horse for $10000 to win at 2/1 and think his chances of winning are even. You could be [while choking on your pie] thinking, "what a mug, that nags chances with that weight at this distance are about 1000/1. The horse wins. What were his chances or true odds of winning? Backfitted they were 100%.

Chance or risk in Horse racing I believe is nothing more than a subjective guide. The reason some punters win and some loose is that some guess [judge future probability and therefore are less wrong] better than others most of the time.

Risk management can only ever be subjectively measured and as humans we do that well enough to survive as a species but not well enough that we all can get rich on the punt.There can never be a formula [but we can all have a great time trying to find one].

Cheers.









[ This Message was edited by: crash on 2003-11-01 04:36 ]

Hi gunny,

You are of course correct and level staking is the least risk.
But it doesn't really cover the maximising the return part.
I'm looking at the best combination of both.
There's absolutley nothing wrong with level staking though and is the safest.

As for basing staking systems on past results....

I agree, but when you have nothing else to work with....
What I normally do is get a very large sample of results,(this is always preferable) and run a staking plan - then use a filter option on an excel spreadsheet to sort the results numerous times so that different patterns emerge.
Then you can work out "IF" the results occurred at different times or in a different order etc.

Chrome Prince....i do agree with your thoughts on level staking.......the only difference i use is bank percentage bet....i guess this is not level staking in the normal sense but it depends on when you recalculate your bank....i do it when my bank has risen 20% i have done many tests over time and i find that thie type of bet performs as good if not better than most of the staking plans in terms maximising profits....if you could make 10% on your bank every month and used a percentage bet you can double your bank every 8 months.....i am not saying there are better staking plans out there....but for someone just starting out punting and is looking for the safest staking plan to start with....there is nothing better than bank percentage

Hi crash,

I'm interested as to why you think that profit can be measured but risk cannot.
As you outlined one cannot be 100% accurate as to a horse's actual chances, BUT if you have for example 1000 bets and 50% of them win and your average dividend is $2.20, it really doesn't matter what odds an individual horse should be.
You still end up that you have a 50% chance overall so you are making 20% profit.

Of course there will be horses you consider have a 50% chance but have less, and others which have an 80% chance but you rated only a 50% chance.

Looking at it as a group of bets rather than the individual.

I calculate risk as the maximum drawdown as a percentage of the starting bank.

EG
Bank = $1,000
Maximum Drawdown $48.50
Risk = 4.85%

This is the risk of going bust on this particular series, you can shuffle the results various ways to see worse case scenarios etc.

This is also the reason that I take a series of results and shuffle lots of times and record the various patterns. I think I mentioned earlier, that you could be lucky and win straight away and never touch your starting bank too much, but at a different point you'd half lost half.

This is why I measure maximum drawdown from the highest point of the bank to the next lowest and apply that figure to the starting bank.

Others calculate it differently, but kid themseves that they had 0% risk, which is inconceivable.

You make some good points crash, thanks for your reply.

[ This Message was edited by: Chrome Prince on 2003-11-01 11:46 ]

Chrome,

Your "Bank" is a present fact.
Your "maximum drawdown" is a past result.
Your "risk" is also a past result.

Presenting them all as future fact is "special pleading" [1 present fact + 2 past results = 3 future facts]. Your measurable risk then is based on assumption which is not usefuly measurable but could be used as a guide for future prediction [only].

If your maximum drawdown is exceeded at some future time, you must create a new maximum drawdown and risk % as new "facts". Meaning your earlier MD was not a maximum at all.

If however I am incorrect and your MD wont in the future be exceeded [is a known and measurable fact] then you are correct and risk can be measured. If that is so then could you please tell me the maths formula you used to mesure your MD [that was then used to measure risk %]. No maths formula, no facts.

It's all a bit like plotting the path of hurricanes based on their previous paths I'm afraid. Cannot be accurately done. A general guide sure, but reliable never [not even after examining millions of previous ones].

If you can measure MD or risk with maths [I am waiting on the formula not the argument], we are both wrong about not being able to measure exact odds using maths, but somehow I don't think so. Do you ?

Cheers.

Cheers.



[ This Message was edited by: crash on 2003-11-01 23:57 ]

Hi crash,

I agree with what you say.

What I am getting at though is the "best" way of measuring or presenting things.

Nobody can predict accurately the future, but we can make a pretty good guess. How accurate that guess is depends on the research done I believe, otherwise may as well stick to level stakes or such.

In answer to your question...

Maximum Drawdown = (Bank Highest Point) - (Bank Subsequent Lowest Point)

Risk = (Maximum Drawdown / Starting Bank) x 100/1

I hope this was what you were after.

What I do though is randomise the results many times and run simulations as mentioned earlier and calculate various outcomes for each scenario - it's a better guide.

All of the above is a guide, as you say there are no guarantees, I'm just looking at the best way of making a better prediction.