Good analysis John. I've often thought that Hurst / Millard, etc. would fit well with this methodology. MIKR had a takeover offer, so there's not going to be much tradeable volatility there.
Mischief is absolutely right. You falsely and deliberately attributed a quote to Spyder which was made by poster dbphoenix. Here is what Spyder actually stated in his post on the page that you referenced: "Nwbprop (the originator of the futures method thread on ET) recently closed the thread, however, he has enjoyed success using the Equities Trading Methods. Discussion regarding Jack Hershey's Equities Method (referred to in the original post) continues in another thread located here." Jack's past differences with ET moderators have absolutely nothing to do with Spyder's current implementation of Hershey trading techniques. No, the easy and correct thing to do is to ignore deceitful individuals such as you. In the spirit of Usenet: *plonk*
Thanks mischief! Your info on MIKR gave me goosebumps. I even looked at the chart again to confim that I had the turnaround/cycle low on 26 Jan 2007. OK, so here's the deal: which came first the established cycle or the takeover. Argh! I have to stop now, my head is spinning. Can't ... see ... the ... submit ... button ...
Should anyone find a particular post off topic, or believes a specific contributor fails to provide information pertinent to the discussion, I encourage liberal (and frequent) use of the ET "Ignore" and "Complain" features in an effort to respond to inappropriate commentary. Adherence to the above guidelines should allow for an open and friendly discourse and foster an environment conducive to learning. - Spydertrader
mischief, Thanks for posting the info and the Amibroker code. I've been thinking of writing an autotrading program in WealthScript, and that will help me work it out. Have you actually been running the autotrading program, and if so are your results anywhere near the backtest results? I often find a huge difference between backtesting and reality... good work!
Palinuro, I'm running a slightly different script at the moment, which was a 'kludge' that overcame the limitations of IB's 30 day backfill. In essence, I was finding dry up stocks on an EOD database and then putting these into a watch list on my intraday database. Because I didn't have enough data to truly calculate dry up volume, I was using the previous days volume of stocks in dry up as a proxy. This has worked pretty well. However, I recently switched to IQFeed for data and on Friday Ambroker recently a new plug-in for IQFeed that allows up to 8 months worth of 1 minute backfill. This has allowed me to remove the kludges and calculate dry up and intraday signals in the one script. I've traded live since 1 December, with some hiccups along the way. My results are a Win percentage of around 48% with a profit factor of approx. 2.2. I've created a backtestable version of my 'kludge' script and over the next couple of days am going to compare the backtest results to reality to see how much slippage I am getting. I'll post the results if you like. However, a backtest of the posted script from 1 Dec 2006 gives a win percentage of 55% with a profit factor of 2.73. Considering I've missed some good trades due to stuff ups, I'd say that my reality and the backtest are pretty close. Of course I've missed some bad trades as well, but a missing a bad trade has avoided a 2% loss, whereas missing a good trade has cost me up to a 10% profit target..
Very cool. It looks like the cash is not being fully utilized most of the time though (position size and opportunity constraints). What happens if you increase the position size to 5% or 10% equity risk per trade? Just for fun: how far can you push it before it blows up or gives a DD > 40%, say?
Hi Pointone, The problem is (and I know this from my production trades), is that you quickly run out of margin. By way of example (remembering all trades are entered with a 2% stop loss). 25k equity 20.00 entry price. Risk/share = 20 * .02 = 0.40 1% equity risk = 250 2% equity risk = 500 5% equity risk = 1250 10% equity risk = 2500 Shares to buy (& position size) 1% = 250/0.4 = 625 = 625 * 20 = 12500 2% = 500/0.4 = 1250 = 125 * 20 = 25000 5% = 1250/0.4 = 3125 = 3125*20 = 62500 10% = 2500/0.4 = 6250 * 20 = 125000 Reg T (overnight) margins are max 50%. This means that I couldn't hold a 5% or 10% equity risk overnight. I couldn't even do a 10% equity risk on an intraday basis. At the moment, in my real trading, I am risking 1.5% per trade. This allows me to enter multiple trades (up to 3 full positions and 1 with reduced size) and still be able to hold overnight. The holding in cash is simply a reflection of not getting any entry signals, perhaps due to the wider market tanking. I have coded and backtested a script that shorts stocks as well, which provides superior net profits and uses your equity more efficiently, but it has much greater drawdowns. Admittedly, this is only backtested over the last 5 months (a bull market) so may be different in a bear market. You may also not get fills as easily as in a backtest because of the uptick rule. If anyone has 1 minute data for 2001 - 2002 I'd be happy to backtest
I understand the stop loss : position size maths - it does seem to be inefficient though. I am curious with your backtest what the limits are if you utilized all the cash available. (Not equity, but free cash.) In the spirit of how Jack started out - all in - the risk is already taken care of in your stock selection and cyclical analysis. Let's pretend the 10K or 25K is only 2% of our net worth so we are still conservatively within conventional money management guidelines... Thanks for your response. EDIT: to be clearer (hopefully!), the average cash balance on your backtest appears to be around 12K. If this was fully invested into the stock selections the equity curve would be somewhat steeper. Forget conventional position sizing for this exercise.
Ok, ran a couple of tests. I've zipped the backtest results and attached them. They were: 'Standard' 1.5% risk, no reduced position size allowed, 50% margin. 3% risk, reduced position size allowed, 50% margin 5% risk, 25% margin requirement (to simulate a larger account with a quarantined portion of equity allocated to this method). As you would expect profit factor and percentage win rate didn't move that much, since we are in essence taking the same trades on each backtest. Clearly drawdown and end net profit did change though with the greater risk. The third backtest actually highlights something I recall reading in one of acrary's posts. To boost returns without increasing drawdowns, increase the risk/trade on a portion of your equity. He stated that you need to run some tests to work out the ratios, but this (single) test seems to bear it out. One last test I did was to double the account size, then pull back to 2.5% risk with 50% margin to simulate the same dollar per trade risk. The 5%, 10k, 25% margin account returned a net profit of 76671 with 56% winners and a profit factor of 3.63 and a system drawdown of 4965 or 27.48% The 2.5% 20k 50% margin account returned a net profit of 61736 with 56% winners and a profit factor of 3.63 and a system drawdown of 4249 or 15.71% It's a fascinating exercise, since you'd almost expect the numbers to be the same, or at least similar. For an extra 4965 - 4249 = 716 in drawdown, you boost returns by 76671 - 61736 = 14935 In percentage terms: 716/4249 = 16.8% 14935/61736 = 24.2% Of course, you'd need to test this exhaustively, but it is still food for thought. Thanks for sending me down this path....