philosoraptor and I have been discussing some analysis he had
done regarding risk for the Multi system. If you look at the
Statistics page for the Multi Entry System you'll see that Matt
points out that he adjusted the reporting on the Win percentages
because "I treat a 'scaled in' trade as one trade, whereas TS
treats them individulally." Matt had also pointed out TradeStation
reports the maximum drawdown looking at each execution as an
individual trade. The maximum drawdown of a single trade was 10.11%
in terms of the capital allocated to the trade and amounted to
4.05% of the total capital. Tradestation reports the 4.05% number
because it's looking at the total capital.
However if we were in a trading office with a risk manager we
suspect that the number they would be considering would be daily
(or moment to moment) TOTAL % risk as a percentage of the trader's
capital. Therefore one may wish to consider the aggregated
risk of the 4 trade entries that are all open for part of the same
period when there are 4 signals. This number can be arrived at by
weighting the individual drawdowns based on percent of capital
allocated per entry OR by looking at the drawdown in the average
price per share for the total trade.
This linkhas a spreadsheet which lays out these
calculations for the same trade series that included the
TradeStation max drawdown trade. Both calculation methods yield a
drawdown number of 7.74% which is obviously well above the
TradeStation's 4.05% number. A drawdown of 7.74% is still
excellent in the context of the amazing system Matt has developed.
However the difference might lead some to re-think the total
capital allocated or whether to use leverage. One rule of thumb is
to assume that the "real" drawdown going forward will be twice what
was experienced during backtesting. So based on that the trader
might think in terms of looking at the account down 15% from it's
high (assuming NO leverage) and think if he or she could stay with
the system through that. We should note that we are assuming that
7.74% was the maximum aggregated risk but without running the
calculations for all the trade series we don't know for sure. It's
unfortunate that TradeStation cannot be programmed to analyse
"scale-in" trades as single trades because then this would all be
automatic. All constructive thoughts welcome -- hey we would
rather be able to think in terms of 4%
Please see this earlier post as to why the max drawdown for the
multi system should be reported as 7.74% rather than 4%. We were
surprised this did not generate more of a response. If someone
thinks this analysis is not correct I'd be interested in hearing
why.
Thanks for reposting this. I forgot to look at your worksheet
the last time when you first posted it. Yes, I pretty much agree on
the numbers. The 4% could be very misleading for the reasons that
you explained.
Thanks bkout. I think this is very important. I'm going to look
at it in more detail over the weekend.
I love Matt's system. But one thing I want to mention is that it
may not be realistic to expect continued 90%+ winning trades. When
a system goes live, it's very difficult to get the same results as
what was backtested. I'm only mentioning this because I'm afraid
some might overleverage too much. I believe this system will
continue doing extremely well because I really like the way it
enters trades. It will still have a high percentage of winning
trades and a very high profit factor imo. I hope I'm wrong and we
continue the amazing results, but 90%+ is very hard to duplicate
when the system goes live imho.
Your work is really helpful as I, and I suspect other members as
well, continue to weigh how much capital to allocate to the system.
As we all know, Matt's system dodged the Flash Crash, but I think
we all have to allow for the possibility that we might not be that
lucky next time there is a Flash Crash or some kind of
out-of-the-envelope Black Swan. Also, we all know that the Flash
Crash rallied back VERY quickly, so had the system been long going
in to the crash, the ultimate exit from the trade would not have
been all that bad. But had the system been long going in to a
1987-like event, then the loss could easily be over 20%. And using
a 15% Crash Stop as Matt discussed as a possibility, might or might
not help. We also know that many Crash stops in the Flash Crash
actually executed WAY below the stop points. So given all this --
your 15% rule of thumb number seems like a very reasonable point of
reference.
In trying to think through risk parameters for the system
itself, we also have to remember that the system does not exist in
a vacuum, and if we are long in other positions at the same time,
then what we really need to be looking at is how much total capital
we have in the market at any moment in time, and what is our risk
tolerance for our entire portfolio position.
Thanks again for all your excellent work on this!
Newsletter
Subscribe to our email list for regular free market updates
as well as a chance to get coupons!
Title: Alternative risk analysis for
Posted by bkout3 on 20th of May 2011 at 11:30 am
philosoraptor and I have been discussing some analysis he had done regarding risk for the Multi system. If you look at the Statistics page for the Multi Entry System you'll see that Matt points out that he adjusted the reporting on the Win percentages because "I treat a 'scaled in' trade as one trade, whereas TS treats them individulally." Matt had also pointed out TradeStation reports the maximum drawdown looking at each execution as an individual trade. The maximum drawdown of a single trade was 10.11% in terms of the capital allocated to the trade and amounted to 4.05% of the total capital. Tradestation reports the 4.05% number because it's looking at the total capital.
However if we were in a trading office with a risk manager we suspect that the number they would be considering would be daily (or moment to moment) TOTAL % risk as a percentage of the trader's capital. Therefore one may wish to consider the aggregated risk of the 4 trade entries that are all open for part of the same period when there are 4 signals. This number can be arrived at by weighting the individual drawdowns based on percent of capital allocated per entry OR by looking at the drawdown in the average price per share for the total trade.
This linkhas a spreadsheet which lays out these calculations for the same trade series that included the TradeStation max drawdown trade. Both calculation methods yield a drawdown number of 7.74% which is obviously well above the TradeStation's 4.05% number. A drawdown of 7.74% is still excellent in the context of the amazing system Matt has developed. However the difference might lead some to re-think the total capital allocated or whether to use leverage. One rule of thumb is to assume that the "real" drawdown going forward will be twice what was experienced during backtesting. So based on that the trader might think in terms of looking at the account down 15% from it's high (assuming NO leverage) and think if he or she could stay with the system through that. We should note that we are assuming that 7.74% was the maximum aggregated risk but without running the calculations for all the trade series we don't know for sure. It's unfortunate that TradeStation cannot be programmed to analyse "scale-in" trades as single trades because then this would all be automatic. All constructive thoughts welcome -- hey we would rather be able to think in terms of 4%
alseq -- See this post
Posted by bkout3 on 8th of Aug 2011 at 07:11 pm
alseq -- See this post from May 20 looking at max drawdown analysis for Single vs Multi Entry system
Please see this earlier post
Posted by bkout3 on 10th of Jun 2011 at 02:51 pm
Please see this earlier post as to why the max drawdown for the multi system should be reported as 7.74% rather than 4%. We were surprised this did not generate more of a response. If someone thinks this analysis is not correct I'd be interested in hearing why.
bkout, Thanks for reposting this. I
Posted by cw12 on 10th of Jun 2011 at 03:02 pm
bkout,
Thanks for reposting this. I forgot to look at your worksheet the last time when you first posted it. Yes, I pretty much agree on the numbers. The 4% could be very misleading for the reasons that you explained.
Thanks bkout. I think this
Posted by cw12 on 20th of May 2011 at 11:58 am
Thanks bkout. I think this is very important. I'm going to look at it in more detail over the weekend.
I love Matt's system. But one thing I want to mention is that it may not be realistic to expect continued 90%+ winning trades. When a system goes live, it's very difficult to get the same results as what was backtested. I'm only mentioning this because I'm afraid some might overleverage too much. I believe this system will continue doing extremely well because I really like the way it enters trades. It will still have a high percentage of winning trades and a very high profit factor imo. I hope I'm wrong and we continue the amazing results, but 90%+ is very hard to duplicate when the system goes live imho.
Great work Bkout! I think we also need to keep in mind Flash Crash/Black Swan Risk
Posted by puma on 20th of May 2011 at 01:30 pm
Your work is really helpful as I, and I suspect other members as well, continue to weigh how much capital to allocate to the system. As we all know, Matt's system dodged the Flash Crash, but I think we all have to allow for the possibility that we might not be that lucky next time there is a Flash Crash or some kind of out-of-the-envelope Black Swan. Also, we all know that the Flash Crash rallied back VERY quickly, so had the system been long going in to the crash, the ultimate exit from the trade would not have been all that bad. But had the system been long going in to a 1987-like event, then the loss could easily be over 20%. And using a 15% Crash Stop as Matt discussed as a possibility, might or might not help. We also know that many Crash stops in the Flash Crash actually executed WAY below the stop points. So given all this -- your 15% rule of thumb number seems like a very reasonable point of reference.
In trying to think through risk parameters for the system itself, we also have to remember that the system does not exist in a vacuum, and if we are long in other positions at the same time, then what we really need to be looking at is how much total capital we have in the market at any moment in time, and what is our risk tolerance for our entire portfolio position.
Thanks again for all your excellent work on this!