|
PureBytes Links
Trading Reference Links
|
This was discussed in an earlier post....the premise here is that for a
given set of logic and rules that any one trade's return can be highlighted
as better than another. Restated: If the system's EXPECTED win rate is 60%
you expect 6 out of 10 FUTURE trades will be winners......then which of the
10 trades will have the best return ?
The response from some knowledgeable systems traders was: If you knew in
advance which trades are better, why trade the rest ?
MY TAKE:
If you extensively profile your backtesting results and come-up with
additional external factors PRESENT AT THE TIME OF THE ORDER that correlate
with actual win amount, then it sounds as though this would be a basis for
increasing the "bet" for an impending trade that meets or exceeds those
conditions. Possibly some sort of "scoring" approach may work here (1 thru
10 where 10 indicates a "100%" winner).
If the conditions are present DURING THE TRADE POSITION, this would be a
basis for increasing the number of contracts in the position at a certain
time-in-trade as a percentage of the average trade length.
Ryan Jones mentions this briefly in his new book "The Trading Game".
I've also recently been introduced to a concept called Maximum Favorable
Excursion which is the point of profit at a certain time-in-trade whereby
statistically the chances of a loss are low. This sounds like a basis for
increasing the number of contracts DURING THE TRADE POSITION. This is
strictly a statistical approach and should be applied only to backtested
results of hundreds of trades IMHO.
Warning: BETTER KNOW YOUR STATS ! Increasing bet size based on statistics
can result in serious, serious drawdowns in equity just as applying Max
Adverse Excursion can have serious, serious detrimental effects to positive
performance of a system.
Bob - your comments are welcome on this topic.
> -----Original Message-----
> From: HDD95@xxxxxxx [mailto:HDD95@xxxxxxx]
> Sent: Friday, July 14, 2000 8:31 AM
> To: omega-list@xxxxxxxxxx
> Cc: bfulks@xxxxxxxxxxxx
> Subject: Re: Vendor System, Real Profits for 3 Years
>
>
> > As for bet-sizing, there are two parts of this problem:
> >
> > > Deciding how many contracts to trade on each trading signal
> >
> > > Deciding how much of your capital to allocate to each commodity
> >
> > You find that some trades will have a higher reward-to-risk-ratio
> > than others so you need to trade more contracts for those cases. This
> > obviously will improve the overall profitability.
>
> I am interested in thoughts about how to have some idea in advance which
> trades will have a higher reward-to-risk-ratio. What you say
> certainly makes
> sense, but I lack ideas for how to determine reward-to-risk ratio
> in advance
> of a trade. I'm not sure it is possible. Any ideas?
>
>
|