I know a number of us here would actively participate in a thread dedicated to the mangement of risk within horse racing. There have been lively debates on the question of staking plans already so I would like to see discussion go beyond that level.

Hi Doc,

I'll take the bait, and do the starting at this stage.... :grin:

For me, risk management is about bank preservation. And bank preservation is being able to predict the maximum drawdown that may occur.

I define maximum drawdown (MaxDD) as being the difference between a bank high, and the lowest point that the bank reaches at any subsequent time from that high. For example, if you start with a 500 unit bank, get a few winners and the bank goes to 650 units, then you hit a bad patch and the bank drops to 450 units, the drawdown at this point is 200 units. If that's the worst retrace that the bank ever does, then you only need a 200 unit bank to ensure that you don't go broke. Not that I'd set aside a 200 unit bank for this particular system, as 1) I'd probably not bet a system with a 200 unit max drawdown, and 2), I'll always allow another 20 to 50% of this as a buffer, as the next MaxDD is just one more bad losing streak away.

Before I understood about MaxDD, I used to consider the longest run of outs as a measure of bank risk. I was naive in thinking that if the LRO was 9, then I'd add a few to be safe, so I'd only need a 13 or 14 unit bank..... I used to bet 5% of my highest bank nonreducing, and proceeded to lose my first two $500 banks...

Since I've started basing my bet size on the MaxDD of any particular system, I have yet to lose an entire betting bank. This is not to say I'm winning on all these banks, just that I haven't exhausted them yet :smile:

Another thing that I do as part of my risk management is to setup different banks for each of my systems. It makes for an interesting day if I have 5 bets from 5 different systems, as I may have 5 different bet sizes to worry about. However, I feel that this allows me to separte the risk for each system, and ensures that I don't rob the proceeds from one system to help another ailing system.

Cheers,
Chris.

JFC - please enlighten us all on your superior risk management methods!

I have looked at Kelly quite extensively but have dismissed it as too risky and too impractical for horse racing - obviously you disagree so would love to hear your opinion.

I use percentage of bank as my preferred system with the percentage depending on the system I am using and the level of risk for that bank (like Chris I allocate seperate banks to different betting systems). For my two current horse-racing "systems" I use 2% and 5% of bank (reducing) and for tennis I use 5% of bank non-reducing.

Have you fellows ever used Malcolm Knowles BET SMARTER AND WIN.
In my opinion thats the best risk management stratedgy of all time.
Cheers.
Darky.

If you are referring to "The Power of 10" staking plan in that book then yes I have done thorough testing on it (a lot of it I published on this forum). That testing left me less than impressed with it underperforming percentage of bank staking in most of the test cases I ran through it (these test cases were real betting results from my own systems and from several other people who sent me their betting records for testing).

__________________
"Computers can do that????" - Homer Simpson

Hi Becareful,

I see that you're using a reducing percentage of bank. This is an excellent way of protecting the betting bank, but I've always felt that it performs worse than flat stakes, since you'll normally be betting less on the winners and more on the losers. Have you done any modelling that supports or refutes this?

Cheers,
Chris.

Chris,

I probably should clarify - I don't generally recalculate the bet amount after every bet and I round to the nearest $10 anyway. This means the bet size will only reduce after a few losing bets (or quite a few on the 2% system). For example on the 5% racing system if my bank was $1000 my bet size would be $50. I would only reduce it to $40 if my bank got down to $800 (so 4 losing bets in a row for example).

I have found this "stepped" reducing bank method seems to give the best overall results for me. It takes advantage of the winning streaks as your bet size steps up after a winner and if you get a few nice wins in a row it will take advantage of this. On the other hand in a losing patch your bet size will decrease to preserve the bank until you start hit the winners again.

In practice the average bet size on winners and losers is identical as essentially what I am doing is betting flat stakes at a given level for a while and then readjusting the staking level either up or down depending on the bank (hopefully always adjusting upwards!!!)

In theory with the % of bank approach you will never go bust as you will always only be betting a small portion of your remaining bank (although in practice of course there are minimum bet sizes so once you get down to that level you are in trouble!).

_________________
"Computers can do that????" - Homer Simpson

[ This Message was edited by: becareful on 2003-10-20 11:30 ]

BECAREFUL.Its kept me ahead of the mob and quite frankly i wouldn,t use anything else.
BUT everyone to their own choice.
Cheers.
Darky.

Hi Becareful,

sounds like a good plan, and is similar to what I plan to do with my staking. I have one particular longshot system which backtested shows a maximum drawdown of 65 units. I add another 50% to that, so call it 100units. I'm betting 1% of my highest bank non reducing. My plan is if (ok, ok, when) it gets to a 65unit drawdown, I shall halve my bet until it gets back above the -65 units, which effectively gives me another 65 units drawdown.. If it gets to that level, then the system is a croc and deserves to be chucked!

Cheers,
Chris.

The theory sounds good Stebbo, but what bothers me is if you put your system into a graph form of peaks and troughs, the problem becomes when to enter the graph with bet 1 [assuming level stakes bets here].
Allowing an MRO of say 13 [and that would require a strike rate nearing 50%] and using your example of 200 of max drawback, if I enter the graph with my bet 1 when you were on 650 and a 200 drawback hit, you would be on 450 and I on 300. You would be on a 150 buffer purely due to entry point luck. If a series of MD's hit, or even close to MD [it does happen] with small gains between, I wouild be out and you in [maybe], purely due to luck.

Becareful,

Wouldn't stepped bets often produce the bigger bets hitting loosers and smaller bets hitting winners producing no advantage other than allowing more bets for your $ [perhaps] but not necessarily a greater chance of winning due to ever smaller collects ?

Cheers.

[ This Message was edited by: crash on 2003-10-20 17:01 ]

[ This Message was edited by: crash on 2003-10-20 17:06 ]

[ This Message was edited by: crash on 2003-10-20 17:13 ]

On 2003-10-20 16:55, crash wrote:
The theory sounds good Stebbo, but what bothers me is if you put your system into a graph form of peaks and troughs, the problem becomes when to enter the graph with bet 1

Hi Crash,

the point you make is the exact thing that my MaxDD measures. My MaxDD is the largest value between ANY peak and a subsequent trough. I do this for exactly the reason you state... where I enter a series of bets will determine my required bank, and that pr1(< Murphy has predetermined that I always enter a series the very bet AFTER a bank high, and the bank is about to enter a long period of drawdown.

Cheers,
Chris.