|
PureBytes Links
Trading Reference Links
|
Ken,
I have responded several times to Herman's posts. But, you said it all very well in your post. The only thing more to say about Tharp's little exercise on random entries is that, when he was confronted by someone about this in his public forum awhile ago, he quickly retorted that this was not an exhaustive test of random entries, and he was by no means espousing the use of random entries when trading. He was only making the point that MM is crucial to the profitability of any trading system, and that MM made a random entry system profitable. But even he avers that MM cannot make a losing system profitable. So, you DO have to start off with a profitable system first (positive expectancy), and then when you apply MM techniques, you will stay in the game long enough to reap the benefits. Of that I am convinced. Al Venosaavcinci@xxxx
>From: "Ken Close"
>Reply-To: amibroker@xxxxxxxxxxxxxxx
>To:
>Subject: RE: [amibroker] Re: 2 cent worth on MoneyManagement
>Date: Mon, 21 Oct 2002 08:22:12 -0400
>
>Herman:
>
>I have held up commented on your repeated request because I thought others
>much more skilled and practiced in Van Tharp's book and his principles would
>have replied. Al V has not so perhaps I am misinterpreting Van Tharp.
>
>He repeatedly says that the system is less important that the money
>management techniques. He does not say it is unimportant. He does say:
>
>"...you could make money consistently with a random entry as long as you
>have good exits and size your position intelligently" (quoting a seminar
>participant, pg 200). He agrees with the participant and goes on to do
>various studies which confirm the comment.
>
>"That's it. That's all there was to the system--a random entry, plus a
>trailing stop that was three times the volatility, plus a one percent rick
>algorithm to size positions."...."This system made money on 80% of the runs
>when it only traded one contract per futures market. It made money 100% of
>the time when a simple 1 percent risk money management system was added.
>That's pretty impressive. The system had a relieability level of 38%, which
>is about average for a trend following system." (pg 201).
>
>I believe your premise is that you could make MORE money, according to the
>system test results you have generated. Can you actually trade them is the
>question.
>
>It seems to me that there is a vast difference between running an
>optimization over and over and getting percent returns in the triple digits
>and actually trading the same trades with real money in real time.
>Psychology is the difference and in spite of all you say and all the testing
>you do, you will not -- and can not -- trade as consistently as the
>backtest. If one can do that, unconcerned with drawdowns (or at least
>really able to not second guess, not swing with emotion), then perhaps the
>largest return system result will wind up in the bank account. I know I can
>not do this.
>
>Meanwhile, while I have not done Tharp's extensive testing, I believe him
>that the money management and position sizing results in the best balance in
>the bank account when all else is said and done.
>
>Ken
>
>
>
>-----Original Message-----
>From: Herman van den Bergen [mailto:psytek@xxxx]
>Sent: Monday, October 21, 2002 6:04 PM
>To: amibroker@xxxxxxxxxxxxxxx
>Subject: RE: [amibroker] Re: 2 cent worth on MoneyManagement
>
>
>Hi William and all,
>
>Thanks for your reply. You say "See attached excel file on how to calc risk
>of ruin"... sorry but there was no attachment. Would you mind emailing it
>again, it might have been stripped off by Yahoo, my email is
>psytek@xxxx Many thanks!
>
>The sample result attached to my earlier email was an Overall Performance
>Report for 100 stocks (N100). This brings up the point of whether the
>Expectancy test is valid on groups of stocks. Any comments on that? Perhaps
>this is why my results were way off?
>
>I do not understand when you say "You are also right the example you have
>with payoff ratio of less than 1, the problem might just be at the stop
>loss". I don't use stops in system design... I believe stops and
>position-sizing distorts the true nature of a system and handicaps equity
>analysis and development. Can you explain your statement?
>
>A point that worries me is that to apply strict money management techniques
>one must have big sums to trade. I would guess that if the minimum
>requirement is $200K than many subscribers on this list don't need to worry
>about MM :-) Any comments?
>
>So, this brings me back to the question of which comes first: developing a
>good trading system or a MM system. I think the trading system comes first,
>MM will only be as good as the system we trade. Right?
>
>However, I am intrigued with the screening possibility offered by the
>Expectancy factor. This can be implemented right in the trading system.
>Also, 2D/3D Expectancy charts for optimizations are very interesting. But I
>would like to find a better formula, the one I am using looks too much like
>my Equity curves (if normalized); not sensitive enough. There must be many
>variations of it out there... I'll keep looking.
>
>Best regards,
>Herman.
>
> -----Original Message-----
> From: William Wong [mailto:williamwongab@xxxx]
> Sent: 21 October, 2002 3:42 AM
> To: amibroker@xxxxxxxxxxxxxxx
> Subject: RE: [amibroker] Re: 2 cent worth on MoneyManagement
>
>
> Hi Herman,
>
> Essentially you are right. It is a mathetical expecation. You are also
>right the example you have with payoff ratio of less than 1, the problem
>might just be at the stop loss. Such low payoff ratio means the risk of
>ruin could be quite high. See attached excel file on how to calc risk of
>ruin. I would focus on getting a system with > 1 pay off ratio and > 50%
>times profitable.
>
> Btw, the Risk Adj Ann Return% in AB's report is not how one would
>interpret it. This return is "adjusted" by the % of exposure you have in
>the market. I.e., if your system seldom takes position in the market, this
>return will be inflated, it assumes when your money is not working for you
>in the market, it is working some where else which generates the same
>return. This may not be the case.
>
> The unadjusted return is more realistic as usually our money is idle in a
>trading account, a more conservative number. Anyway, your exposure is close
>to 100%, hence this return comes close to the unadjusted one.
>
> One more thing, focus on the drawdown, your trade and system drawdown are
>very high, close to 90%. This means you need to be very strong in
>conviction to carry through using this system as you will be beaten almost
>flat before it makes money for you. Most professional restricts a drawdown
>of a single trade down to 2% as a rule of thumb, this got to do with the
>money mgmt. For example, if you have $10,000, you have identify a trade to
>buy SUNW at $2.50 with a stop at $2.00. The risk (include commission $0.02
>per share per round trip) is $0.52. Using 2%, I would only risk $200, i.e.,
>$200 / $0.52 = 384 shares in this trade.
>
> Something I learnt through paying tuition fees to the market :-)
>
> William
>
> Herman van den Bergen wrote:
>
> Thank you William,
>
> Can you help me link this to a typical AB BackTest report (attached)?
>
> avg win$ = Average winning trade: 15.71
> avg loss$ = Average losing trade: -17.04
> %Win = Percent profitable: 61.8
>
> Plugging these numbers in your formula I get:
>
> Expectation = ( 1 + 15.71 / 17.04 ) * .618 - 1 = 0.172
>
> I assume the minus sign for Average losing trade is dropped? This was a
>backtest on the N100 so I am not quite sure how to interpret this if we
>apply the formula to groups... ideas? Taking the top ten stocks from this
>test i get
>
> Expectation = ( 1 + 35.19 / 42.94 ) * .706 - 1 = 0.284 still not very
>good...
>
> It looks like my decision to work first on trading systems before
>tackling MM was correct: i am a long way off from 0.7 !
>
> Still, I am not quite sure what to think of this... the first test gave
>me 161%/ann average for 100 stocks, the second test gave me 660%/ann average
>for 10 stocks (of course assuming I could screen the best stocks) over about
>5000 trades. To get 0.7 I'll have to be making thousands of percent per
>year... where did i go wrong? I guess the problem is that the losers are too
>big, adding stops might improve the Expectation rating but would reduce
>profits... hmmm sounds familiar :-)
>
> So, we need to Optimize for Expectation... now, wouldn't that be fun?! I
>might just work on that this weekend :-) AmiBroker can do anything, right? I
>love a good afl challenge.
>
> Intuition tells me the formula ought to have a factor for
>frequency/distribution of trading/profits and the testing period under
>consideration. This formula would give good results if I had only two very
>profitable trades and two small losers - not right...
>WinTradeDuration/LosingTradeDuration probably ought to be in there too...
>How can we factor those in?
>
> Many thanks triggering those sleepy neurons,
> Herman
>
> PS. Sorry for not letting this thread end...
>
>
> -----Original Message-----
> From: William Wong [mailto:williamwongab@xxxx]
> Sent: 18 October, 2002 3:31 AM
> To: amibroker@xxxxxxxxxxxxxxx
> Subject: RE: [amibroker] Re: 2 cent worth on MoneyManagement
>
>
> My 2 cents worth on money management.
>
> A system performance is governed by a combination of %winning trades
>and pay off ratio (avg win$ / avg loss$). If an expectation of a system is
>negative, no matter how good the money management, it only slows down an
>eventual death.
>
> Expectation = (1 + pay off ratio) * %win - 1
>
> e.g. If a system gives pay off ratio of 2:1 but only profitable 30%
>of the time,
>
> Exp = (1 + 2) * 0.3 - 1 = -0.1, not worth using
>
> But if the accuracy is improved to 40%,
>
> Exp = (1 + 2) * 0.4 - 1 = 0.2, worth using
>
> The higher the expectation, the better the system. Rule of thumb is
>to improve to at least 0.7.
>
> Now when a system has a positive expectation, then money management
>comes into play. MM is about reducing the Risk of Ruin, the chance of
>lossing x% of your portfolio before you decide to quit. It about staying in
>the game. It is the different between punting and making a living on
>trading. A mediocre system (positive expectation) with a sound money mgmt
>is good enough to trade for a living.
>
>
>
>
>
>--------------------------------------------------------------------------
> Do you Yahoo!?
> Faith Hill - Exclusive Performances, Videos, & more
> faith.yahoo.com
>
> Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
> (Web page: http://groups.yahoo.com/group/amiquote/messages/)
>
> Check group FAQ at:
>http://groups.yahoo.com/group/amibroker/files/groupfaq.html
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
>
>
>
> Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
> (Web page: http://groups.yahoo.com/group/amiquote/messages/)
>
> Check group FAQ at:
>http://groups.yahoo.com/group/amibroker/files/groupfaq.html
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
> Overall performance summary
>
> Total net profit: 16397.46 Total commissions paid: 0.00
> Return on account: 167.32 % Open position gain/loss -525.51
> Buy&Hold profit: -3024.58 Bars (avg. days) in test: 25480 (375)
> Buy&Hold % return: -30.86% System to Buy&Hold index: 642.14%
>
> Annual system % return: 160.40% Annual B&H % return: -30.18%
>
> System drawdown: -82.66 B&H drawdown: -92.85
> Max. system drawdown: -1191.56 B&H max. drawdown: -260.31
> Max. system % drawdown: -93.74% B&H max. % drawdown: -95.74%
> Max. trade drawdown: -581.29
> Max. trade % drawdown: -88.23%
> Trade drawdown: -444.52
>
> Total number of trades: 5295 Percent profitable: 61.8%
> Number winning trades: 3272 Number losing trades: 2023
> Profit of winners: 51391.78 Loss of losers: -34468.81
> Total # of bars in winners: 16905 Total # of bars in losers:
>13187
> Commissions paid in winners: 0.00 Commissions paid in losers:
>0.00
>
> Largest winning trade: 258.56 Largest losing trade: -357.47
> # of bars in largest winner: 4 # bars in largest loser: 7
> Commission paid in largest winner: 0.00 Commission paid in
>largest loser: 0.00
>
> Average winning trade: 15.71 Average losing trade: -17.04
> Avg. # of bars in winners: 5.2 Avg. # bars in losers: 6.5
> Avg. commission paid in winner: 0.00 Avg. commission paid in
>loser: 0.00
> Max consec. winners: 15 Max consec. losers: 8
>
> Bars out of the market: 194 Interest earned: 0.00
>
> Exposure: 99.2% Risk adjusted ann. return: 161.63%
> Ratio avg win/avg loss: 0.92 Avg. trade (win & loss): 3.20
> Profit factor: 1.49
>
>
>
>
>
>
>
>----------------------------------------------------------------------------
>--
> Do you Yahoo!?
> Y! Web Hosting - Let the expert host your web site
>
> Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
> (Web page: http://groups.yahoo.com/group/amiquote/messages/)
>
> Check group FAQ at:
>http://groups.yahoo.com/group/amibroker/files/groupfaq.html
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
>
>
> Yahoo! Groups Sponsor
> ADVERTISEMENT
>
>
>
>
>Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
>(Web page: http://groups.yahoo.com/group/amiquote/messages/)
>
>Check group FAQ at:
>http://groups.yahoo.com/group/amibroker/files/groupfaq.html
>
>Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
>
Unlimited Internet access -- and 2 months free! Try MSN. Click Here
|