Avg Win: 2.7% (24 winning trades) Avg Loss 2.0% (7 losing trades) Wash: 0% 1 trade NOTE: My first post did not count trades that resulted in a G/L of under 0.5%. This post includes all trades, so there is a difference. Doug
I don't know if you practice any position sizing, but based on your track record (if these results were typical for your trading), you could calculate: Kelly ratio = %win - (1-%win) * avgLoss/avgWin = .77 - .23 * 2/2.7 = 60% To bring the drawdown probability to acceptable levels you could risk i.e. 6% of your account at any given time (this is 1/10 of Kelly ratio). As you grow confident that these results are really typical for your trading you could raise it to i.e. 15% (1/4 of Kelly ratio). This will greatly increase your returns, accepting a higher drawdown probability because you've already built your confidence that drawdowns are an inherent part of your profitable system.
Not to start a discussion on Kelly but I'm pretty well convinced it does not apply to trading since it is based on a Bernouli distribution which is not what we see in trading. Also, when you say risk 6% of your account, that means you are actually taking a position roughly equal about half your entire account. Since the amount you risk is P*Lmax, where P=invested amount and Lmax is the anticipated worst loss. So, for these trades, if I use 10% for Lmax, about twice my worst loss, I would place trades equal to about 60% of my account; too much for me. P = 0.06xAccountBalance/0.1 Personally, I use a Monte Carlo method to figure out the amount to risk and I don't have enough trades to do that work up yet. DS
Is there any post or document regarding position sizing related to Jack Hershey's equities or commodities trading methods? Could you or somebody else point me to it? Thanks. Dougcs, how do you currently size your positions traded with this method? Your 7 week results are great, and confirm Spydertrader's published results.
third post down, Grob109 talk about Kelly and money management. http://www.elitetrader.com/vb/showthread.php?s=&threadid=57364&perpage=6&pagenumber=5
Would like to resolve a discrepancy. I have been running a score scan and tabulating the results, in order to try and visualize the cyclic flow of the stocks on my list. Recently, I did one of these manually, and found that it did not match the scanned result. For example CUTR has a scan score of 2 for this evening. Doing it manually, I come up with a score of 7 (1,1,1). Spydertrader, please comment. Thanks in advance for your help.
Thanks. I haven't read yet anywhere what's Jack's position on risk taking. Spydertrader posted a link to a document about setting stops, but it didn't say anything about how big positions should be taken. Jack writes a lot about money making velocity, but obviously how much you put in your trades affects dramatically both your returns and your drawdowns.
I believe the discrepancy results from an improper calculation of the binary score used during your manual calculations. For price to receive a binary score of one (1), price must show an increase of 1% over the previous day close. For volume to receive a binary score of one (1), volume must show an increase of 10% over the previous day volume total. For Accumulation / Distribution to receive a binary score of one (1), Balance of Power (BoP) must show a level greater than 0.30 on the BoP Indicator (indicating Accumulation). Turning our attention to CUTR on the evening of 2006-01-19, we see a price increase of .09 over the previous day close ($26.47 on 2006-01-18 and $26.47 on 2006-01-19). Since a .09 per share increase remains less than a 1% increase (.26) in price, CUTR receives a binary price score of zero (0). According to QCharts datafeed, Volume levels for CUTR show 130,561 shares traded on 2006-01-18 and 155,041 shares traded on 2006-01-19 (a difference of 24,480 shares). Since volume for CUTR increased more than 10% over the previous day, CUTR receives a binary volume score of one (1). Lastly, using Qcharts datafeed again, the Balance of Power Indicator (BoP) shows a current (2006-01-19) level of 0.0118 for today. Since 0.0118 remains less than 0.30, CUTR receives a binary Accumulation / Distribution (A/D) Score of zero (o) - indicating Distribution. To determine the Overall Score, we use Jack's PVAD Formula: (P)² + (V)¹ + (A/D) As a result, CUTR's three binary scores (0, 1, 0) for Price, Volume and Accumulation / Distribution (PVAD) calculate out a Total PVAD Score of Two (2). I hope the above clarifies the calculations for you. - Spydertrader
A discussion regarding appropriate risk taking and money management occurred in the Original Journal. Within the discussion, one can find links to Jack's documents, several opinions on how to best calculate risk, as well as, an attached 'risk calculator' available for download. - Spydertrader